technical analyst

Preview – IFTA Conference 2004

sept/o

ct 2004

.technicalanalyst.co.uk wwwThe publication for trading and investment professionals

DeMark Indicators US Presidential ElectionsOutlook for the NasdaqIPO data informs technicals looking for signals

in the FX market

how do markets react?

TA in SpainTA in Spain

London Stock Exchange Advertisement

WELCOME

We are very proud to have been chosen as an official collaborator byIFTA for its annual conference in Madrid this year. Our pre-conferencecoverage provides you with everything you need to know about the TAevent of the year. We look forward to reporting on the conference'shighlights in our next issue.

In this issue we also present an introduction to DeMark, whose indi-cators are gaining a growing reputation amongst trading professionalsfor their clarity and reliability. The Technical Analyst visits a Londondealing floor to discover how using DeMark has enhanced tradingreturns for one City dealer. Finally, all eyes will be on the US presiden-tial election in November, so we take a look at how the election cyclehas affected markets in the past.

Matthew ClementsEditor

Editor: Matthew Clements (MSTA)Managing Editor: Jim Biss

Editorial Board:Mikael Bask, Umea University, SwedenTai-Leung Terence Chong, The Chinese University of Hong KongWing-Keung Wong, National University of Singapore

Marketing: Vanessa GreenSales: Christopher LeighDesign: Paul Simpson

The Technical Analyst is published byClements Biss Economic Publications Ltd,Unit 201, Panther House,38 Mount Pleasant, London WC1X 0AN Tel: +44 (0)20 7833 1441Web: www.technicalanalyst.co.ukEmail: [email protected]

SUBSCRIPTIONS

Subscription rates (6 issues) UK: £140 per annumRest of world: £165 per annumFor information, please contact: [email protected]

ADVERTISING

For information, please contact:[email protected]

PRODUCTION

Art, design and typesetting by all-Perception Ltd.Printed by The Friary Press

ISSN(1742-8718)

© 2004 Clements Biss EconomicPublications Limited. All rights reserved.Neither this publication nor any part of itmay be reproduced, stored in a retrievalsystem, or transmitted in any form or byany means, electronic, mechanical, photo-copying, recording or otherwise, withoutthe prior permission of Clements BissEconomic Publications Limited. While thepublisher believes that all information con-tained in this publication was correct at thetime of going to press, they cannot acceptliability for any errors or omissions that mayappear or loss suffered directly or indirect-ly by any reader as a result of any adver-tisement, editorial, photographs or othermaterial published in The TechnicalAnalyst. No statement in this publication isto be considered as a recommendation orsolicitation to buy or sell securities or toprovide investment, tax or legal advice.Readers should be aware that this publica-tion is not intended to replace the need toobtain professional advice in relation to anytopic discussed.

September/October 2004 THE TECHNICAL ANALYST 1

CONTENTSSEPTEMBER/OCTOBER 2004


Industry News

The Technical Analyst Talks To…Nina Cooper, president of the AAPTA

Book ReviewTechnical Trading Tactics by John Person

Software ReviewMTPredictor RT 4.0

Commitments of Traders Report

Long-Term Technicals

Training & Events Diary

05

32

39

40

44

46

48

→→

Special preview of the IFTA CONFERENCE 2004page 13

Contents continue overleaf

4 THE TECHNICAL ANALYST September/October 2004

Market ViewsNasdaq - weak IPO confidence dents technical outlookEUR/USD - a new challenge to a two-year trendUS Treasuries - bears may soon be back in control

IFTA Conference 2004 - PreviewBehind the scenesConference programmeA short history of IFTA Conferences & essential facts

TechniquesImpressive signals from DeMarkThe US Presidential Election Cycle - fact or fiction?Introduction to Kagi Charts

Subject MattersBacktesting predictors of the S&P 500When does technical analysis work …and when doesn't it?

070911

141618

202528

3436

The US PresidentialElection CycleWhat to look for over the next 4 years andwhy a win for Kerry could be the best out-come for stock markets.

DeMark IndicatorsKurt Magnus from Westpac discusses theDeMark Sequential Indicator TM and explainswhy this lesser known study is central to hisFX trading.

20

25

Industry News


The Interactive Data Corporation

(IDC), a leading global provider of

securities pricing, financial information

and analytic tools, is to acquire

FutureSource for $18 million.

FutureSource is a privately owned

provider of real-time data to the

futures, commodities and foreign

exchange markets. "The acquisition of

FutureSource will further enhance,

expand and complement our eSignal

division," says Stuart Clark, president

and chief executive officer of

Interactive Data. Interactive Data

plans to combine FutureSource's prod-

ucts and services into its eSignal divi-

sion and so will add 5,000 global cus-

tomers with 6,500 terminals to

eSignal's customer base.

For more information, contact Jeremy

Berghorst:[email protected]

eSIGNAL ANNOUNCES NEWQUOTREK 1.1

INTERACTIVE DATA CORPORATION BUYSFUTURESOURCE

Thomson Financial has announced thecreation of a new team dedicated to theUK hedge fund community. The team willbe led by Mark Jackson, sales managerreporting to Richard Garnier, managingdirector, European sales and account man-agement. With over 100 hedge fundsclients in the UK, Thomson's hedge fundoffering in Europe will provide a combina-tion of their products and content sets.These include I/B/E/S, First Call, AutExand Worldscope drawn from across theThomson ONE application suite andThomson Datastream, which itself isbeing integrated into Thomson ONE.This can be integrated with portfolio man-agement tools from Thomson PORTIA.

For more information contact: [email protected]

THOMSON FINANCIALTARGETS HEDGEFUNDS

eSignal, a division of InteractiveData Corporation has announcedthe immediate availability ofQuoTrek® 1.1, with user interfaceimprovements including free worldindices and added compatibility withsmart phones and the BlackBerryWeb Client service from ResearchIn Motion (RIM). QuoTrek providesstreaming quotes, charting and ana-lytics to wireless devices such asBlackBerry Wireless Handhelds.QuoTrek is currently available on a30 day free trial basis.

For more information on QuoTrek,visit their website:www.quotrek.com.

Industry News


CQG has released its newIntegrated Client order routingproduct. The new software allowschart analysis and trading to takeplace in one application withfunctionality enabling connectionto E-CBOT, CME Globex,LIFFE Connect, Eurex andEurex US, trading both futuresand options and futures. Tradingdirectly from charts, the user hasa graphical representation ofworking orders and filled tradeswith online access to 30-days oftrade history. CQG say their

Integrated Client product allowsusers to switch between analysisand trading without changingcomputer. Forthcoming featureswill include a spread trading func-tionality, automatedlogging and auto-matic trading trig-gered by pre-deter-mined studies.

Meanwhile, CQGhas opened newoffices in Yerevan,Aremia, Samara

and Obninsk, Russia and Kiev inthe Ukraine.

For more information visitwww.cqg.com.

TRADESTATIONANNOUNCES SELF-CLEARING FOR EQUITIES

CQG OFFERS INTEGRATED ANALYSIS & TRADINGPLATFORM

Updata, the point and figure chart-ing software specialist, hasenhanced its Technical Analyst sys-tem to run in conjunction withBloomberg terminals. The compa-ny reports that anincreasing number ofBloomberg users fromwithin banks and hedgefunds are adopting thesystem which combinescharting software withBloomberg data.

Updata Technical Analyst costs£97 per month to run onBloomberg and can be downloadedfrom:www.updata.co.uk.

UPDATA INTRODUCESADVANCED TECHNICALANALYSIS FOR BLOOMBERG

TradeStation Securities, a sub-sidiary of TradeStation Grouphas commenced equities self-clearing operations for its activetrader client base. TradeStationcompleted the conversion to self-clearing of its client account basein September. All stock tradesmade by TradeStation's activetrader client base on or afterSeptember 8th, and all optionstrades made on or afterSeptember 10th are now clearedby TradeStation.

www.tradestation.com

Market Views


The time line remains a tough nut to crack when tech-nically forecasting movements in any market.Establishing direction and range is relatively easy by

comparison. When using technical tools to provide answersto questions which are wholly impacted by fundamentalevents, the fact that your analysis is based on a study of his-torical pricing allows you to relax in the knowledge that theanticipation of these events is built into the current priceaction. All of this fundamental information is thereforebuilt into your technical indicators. Accordingly, you mayfeel comfortable that you have a grip on the direction andhow far in that direction you are heading. However, this stillleaves open the matter of how long it will take to reachyour objective. The fol-lowing suggestion maygo against the grain incertain technical analysiscamps but a calendarmarking the key upcom-ing fundamental eventsis going to be our bestfriend in this area.

The NASDAQ, unlikethe other Americanstock markets, may haveanticipated benefitingover the past 3 monthsfrom a seasonal spate ofIPO's. Despite the nerv-ous trading pattern pre-vailing post the dot combubble and 9/11, the NASDAQ has continued to attractnew business to the market with mixed results. Market con-ditions have been unfavourable since the beginning of theyear with a reported average drop of 4% since issue price.This alone is likely to apply a sharp handbrake to theapproach of any encouraging and potentially profitable newissues. The technical resistance in this market, which linesup well with those gloomy fundamental results, is lookingfirm in the 1450 area as marked on Figure 1.

The extremely discouraging signs highlighted on this chartsuggest the technicals are matching the caution currentlybeing exercised by those banks who are in a position to leadfurther IPO's. NASDAQ in particular is a market which can

have its temperature taken by the performance of its recentIPO's. The weak supply of new business therefore suggestsa lack of confidence, at least from institutional investors,that this market has turned a corner and is forming a risingsupport base. The evidence here in Figure 1 indicates thatwe have a potential downside of 1119 at present and thatthis target level is declining at an average rate of 4.00 pointsper week. By 31st December, if we continue at the currentrate of decline, this target will have lowered to 1058.

Sell the ralliesSelling rallies back into the initial declining resistance at1420 with a stop above 1460 and a target at 1119 produces

an attractive risk rewardtrade of 7.53:1. Theshort term target for ini-tial profit taking is 1294into the Ranger -1 level(a proprietary term foran initial exhaustive sup-port level), as marked.The declining stop is setat the light blue indica-tor and a break abovethis level (currently1451) would suggest thatmomentum is turningpositive and a move upinto 1578 will beachieved. The 1578 levelis a trend reversal point

which opens a recovery path into 1653 and then 1829.These Fibonacci wave cycle levels take account of theweekly range on each price bar and will rise/fall accordingly.Traders will therefore be kept in a profitable position forlonger and the trailing stop level will provide a potentiallyprofitable exit should it decline beneath the original entrylevel.

Should the market fail to find support at Target 2 on thischart, the outlook for the first quarter of 2005 will be bleakto the tune of 835 (a declining level). Until the light bluetrailing stop has been broken, a new low point may beinevitable.

Uncertainty about the future, coupled with relief at

NASDAQ WEAK IPO CONFIDENCE DENTS TECHNICAL OUTLOOK by Karen Griffiths

→→

Karen Griffiths

Market Views


being in a position to lock in some much sought after profitis perhaps causing the market to remain firmly rooted inthis declining channel. The upmoves are worth being longfor - a low point of 1303 pushing back up to 1420 is worthconsideration. The chart indicates that the next breakout tothe downside will be beneath 1294 looking for the pull backto 1119 and below.

The US election results and the anticipated interest rate risesover the coming three months should provide the fundamental

clarity which the market is waiting for. Technically, a furtherbreak to the downside is favoured. As the stop loss leveldeclines each week, the turning point for a shift in momentumto positive is a level which will aim to reduce losses while spec-ulating from the short side.

Karen Griffiths is chief technical strategist with PronetAnalytics in London.

Figure 1.


Market Views

After a 2-year rally, EUR/USD peaked early in 2004finding strong resistance against a key Fibonacciretracement level at 1.2925/50. EUR/USD is one

of the few sectors in the market that tends to show a lot ofrespect for the 75% retracement and this time around is noexception. After reversing against this key retracement levelat 1.2925/50, the market lost 12 big figures to find newsupport against the 2-year trendline at 1.1750. UnlikeSeptember 2003 when EUR/USD re-tested and held thistechnical support, momentum measures and sentiment datadid not support a strong bounce in 2004. On the contrary,momentum measures indicated a very weak market, whilesentiment highlighted a continuously overbought environ-ment. This technical phenomenon resulted in a choppyconsolidation that lasted for most of the summer monthsand, as we entered the presidential election campaign, builta possible head-and-shoulders top in this market.

Elliott Wave technicians would also note that the 2-yearrally in EUR/USD (February 2002-February 2004) hascompleted a 5-wave pattern witha peak at 1.2925/50 and hasbegun the 1-2-3 correction intothe latter part of 2004. Puttingthis all together, there seems tobe much technical risk in themarket as we move into the finalstages of 2004 and, if con-firmed, should pave the way fora weaker euro and stronger dol-lar into the end of the year.

There have been many recentdiscussions about the geopoliti-cal risks that might keep thedollar subdued ahead of the USelections in November 2004.However, we have witnessedsimilar discussions associatedwith both the Democrat andRepublican conventions, as wellas the Olympic Games held ear-lier in August. As we enter

autumn 2004, the risks associated with the above eventshave dissipated, paving the way for the next directionalmove in EUR/USD as well as other major currency crossesaround the globe. In the US, the dollar index (Figure 2)challenges resistance against the 2-year trendline around89/90. Similar to EUR/USD, this sector has been in achoppy consolidation since May 2004 and a confident moveabove 90 will be needed to open the door for a strongerperformance from the USD.

A move through this important technical level of 90would likely correspond with a break below the necklinesupport in EUR/USD, in turn offering a first serious chal-lenge to the broader dollar bearish trend that has been inplace since 2002. As food for thought, the measured objec-tive of the current head-and-shoulders formation wouldoffer a technical target against 1.150. This is well below theApril 2004 low at 1.1760 and would correspond with a 62%Fibonacci retracement from the September 2003-February2004 rally in EUR/USD at 1.1590/80.

EUR/USDA NEW CHALLENGE TO A TWO-YEAR TRENDby Michael Trefel

Figure 1.

→→

Market Views


Figure 2.

However, while many technical signals are pointing to alower resolution for EUR/USD into the end of 2004, somecaution is warranted. The two-year trendline has in the pastbeen a formidable support and may extend the summer2004 consolidation further. This may keep EUR/USD in awide range of 1.1950-1.2500 for the remainder of Q3 '04.

All in all, the currency markets around the world are nowat a key technical junction. We will cautiously await therange resolutions for better guidance toward a more direc-tional move from both the euro and dollar into the latterpart of 2004 and early 2005.

Michael Trefel is chief technical strategist for globalfixed income and foreign exchange at LehmanBrothers, New York.

Important Note: This report is the sole possession of Lehman Brothers and is beingprovided to The Technical Analyst for educational purposes only. It is by no meansintended as an investment solicitation. Nonetheless, there are important investmentdisclosures on the last page. These disclosures should be read carefully. In addition,this report is copyrighted and may not be reproduced under any circumstances. Anydistribution of this publication outside of its intended use, i.e., the The TechnicalAnalyst must be approved by the Fixed-Income Publications Department : (212-526-6268 or [email protected]).

“THERE SEEMS TO BE MUCH TECHNICAL RISK IN THE MARKET AS WE MOVE INTO THE FINALSTAGES OF 2004.”


Market Views

US TREASURIESBEARS MAY SOON BE BACK IN CONTROL by Thomas Anthonj

10-year Treasury

10-year Treasury notes have been in a long-term uptrendthat started in 1981 (Figure 1). But after reversing sharplylower in a key-reversal week on 20th of June 03 and break-ing below key-support at 112.15 (last bottom) 12 monthsago, the market finally confirmed a long-term trend rever-sal. Having formed a wave 1 down to 108.02, the marketperformed a 2nd wave rebound up to 116.18 in March, alevel right in its target zone between the 61.8 % and the76.4 % retracement (115.22-117.16).

Accelerating downwards thereafter in the manner of a 3rdwave impulse (target 96.19), we see another wave 1 sub-count completed at 106.1. Having already retraced 76.4 %(114.10), the risk of resuming the downtrend to test thenext Fibonacci support level at 103.28 is now very high.Only a weekly close above 114.10 would start to alter thebigger picture.

30-year Treasury

Stalling at a projected price target for a 5th wave top at122.30 in a head-and-shoulders pattern in mid-2003 (Figure2), the 30-year Treasury market retreated and broke below akey support at 117.26 (last bottom) - see Figure 2. Havingalso broken the row of higher lows at 106.13, we receivedthe final certainty that we have seen a long-term turnaroundat 123.03. Bouncing from strong support at 104.00, themarket performed a typical 2nd wave rebound up to 116.12and resumed its bigger bear trend thereafter, which meansthat we are most likely performing a 3rd wave impulsedownwards with a calculated target at 85.08 (wave 1 x1.618).

The latest rebound from the weekly tendline support at103.06 is most likely nothing else but another internal 2ndwave rebound that was expected to stall at 113.06 (76.4 %)at the latest. Having tested this level two weeks ago we'd

Figure 1.

→→

Market Views


only need to break below an old top at 109.15 to receivestrong evidence that the bears are back in control. Thiswould mean a test of the weekly trendline support, whereasa break below would open substantial downwards potentialtowards 97.06 (old low), and strong support between 92.07and 89.00 (38.2 % / old bottom).

Thomas Anthonj is chief technical analyst at ABNAmro in Amsterdam.

Figure 2.

“WE ONLY NEED TO BREAKBELOW AN OLD TOP AT109.15 TO RECEIVE STRONGEVIDENCE THAT THE BEARSARE BACK IN CONTROL.”


IFTA CONFERENCE 2004 - PREVIEW

¡Bienvenidos aMadrid!I would like to thank The TechnicalAnalyst magazine for the great job theyhave done in this issue to bring theconference to all its readers. Thisyear's conference in Madrid is going tobe a worthy meeting and the technicalanalysis community should be madeaware of this unique event. We havemade every effort to get first classspeakers such as John Murphy, JohnBollinger, Perry Kaufman, Martin Pring,Trevor Neil, David Krell and BernardLietaer among others. Special care hasbeen taken in the selection of confer-ence facilities to make people comfort-able and relaxed. Finally, tourist excur-sions and fine cuisine will be just someof the highlights of the conference. Forall these reasons, I encourage you tojoin us and enjoy this magical threeday event. ¡Bienvenidos a Madrid!

Marc MichielsConference Chairman, 2004


BB EE HH II NN DD TT HH EE SS CC EE NN EE SS

With the celebration of the 17thIFTA Conference in Spain onNovember 4, 5 and 6, Madridbecomes the capital of technicalanalysis for three days. TrevorNeil, John Murphy, John Bollinger,David Krell and many others willbe there, as will The TechnicalAnalyst covering the event for itsreaders. We talked to MarcMichiels, daily manager of theSpanish Association of TechnicalAnalysts (AEAT) and 2004 IFTAConference Chairman, to getsome idea of what goes intopreparing an event for almost 900attendees.

TTA: The Spanish TA association has been around since 1992 but neverbefore applied to host the conference. Why now?

MM: The time felt right. The current boom in technical analysis, both glob-ally and in Spain particularly, and the fact that it was the turn of a Europeansociety to host the conference, convinced us to put forward our candidature.

TTA: How did the AEAT succeed in persuading IFTA to hold theConference in Spain?

MM: We have always been interested in IFTA matters and our relationshipwith most of the national associations and IFTA has always been excellent.The most important prerequisite for us to bring the conference to Madridwas to be sure the big names of technical analysis would be present. Thanksto our relationship with our partners, it was not difficult to get some of thebest technical analysts to speak at Madrid.

With such a program, we only had to manage a few other things to buildup a strong dossier. A location for the conference headquarters (one of thebest hotels in Madrid), and a marvellous 3-days tourist program to entertainattendees' partners. To keep to our budget, we have relied on sponsor andco-sponsor contributions which have allowed us to maintain reasonable feesfor attendees and offer them a top quality conference. As you can guess theSpanish candidature was unbeatable!

TTA: How did you go about gathering speakers?

MM: The key factor in attracting such first rate speakers to a conference isto already have a pair of prestigious names on the conference program. Forthe Madrid conference, we first made contact with David Krell,International Securities Exchange founder, a very good friend of JorgeBolívar - President of Spanish Society of Technical Analysts - and asked himif he was willing to participate. David wanted a successful conference andsaid yes directly. Other keynote speakers that have been on the programsince the very beginning are John Bollinger and Martin Pring. With thosethree keynote speakers, our job of getting good speakers became much eas-ier. In November 2003, we went to Washington D.C. and participated in theXVI IFTA Conference. We came back with acceptance of participationfrom two other giants of technical analysis: Jonh Murphy and PerryKaufman. Five keynotes speakers were already scheduled on the program inwhat was a great bonus for our organization. After that, there was no moreneed to look out for speakers. Since then, speakers have contacted us to bepart of the program. In Spain, local technical analysts have been queuing upto offer their services!

More recently, Bernard Lietaer accepted our invitation to make a presenta-tion about his best seller "The future of money" on the gala dinner of theconference. The last star to sign at the conference was Trevor Neil,Bloomberg chief European technical analyst, in July 2004. In fact,Bloomberg are one of the conference sponsors and will be installing two flatscreens in the lounge with channels in both English and Spanish.

So far, the conference comprises seven keynote speakers, a record for anIFTA Conference. Another five local keynote speakers and twelve othersfrom different parts of the world constitute the full speaker team. We arevery proud of our speakers and their topics, and we hope that attendees willappreciate them.

TTA: What does the AEAT hope to achieve with this Conference?

MM: Our main objective is to bring together local technicians and offerthem the possibility of exchanging ideas with foreign technicians. We expectthe conference will be the first step in creating a strong local TA communi-ty. But, above all, the conference should be viewed from a international per-spective, as an international forum for TA professionals to discuss newtrends in the discipline.

Marc Michiels


TTA: So what's the thinking behind choosing "Technical Analysis in ActivePortfolio Management and Risk Control" as the conference theme?

MM: The selection of this topic is motivated by the absolute willingness ofthe conference promoters to organize a meeting that will really help finan-cial market professionals in their day-to-day work.

The behaviour of the financial markets in the last few years has shownclearly the importance of adopting active management strategies which helpprofessionals avoid or even take advantage of bearish phases of the markets.Timing is essential if we want to preserve our capital in adverse situationsand make the most of bullish phases. Absolute return is equally, if not moreimportant, than relative return.

Managing a portfolio essentially involves managing its risk. As managers,we cannot decide on the direction and volatility of the prices. But we canadjust our investments (exposures or weights) to ensure they do not exceedthe maximum risk of acceptable loss. Quantitative technical analysis allowsus to manage our portfolio dynamically, track prices and volatilities, and toidentify buy and sell opportunities. In addition, technical analysis can also beused in a 'relative' way (intermarket analysis) to select between differentbuy/sell alternatives.

Technical analysis is already widely applied in portfolio management as anadditional tool for taking investment decisions. We want to go far beyond thecurrent use of technical analysis, bringing all these professionals to Madridto talk about their methodologies and preferences in relation to managingportfolios with technical analysis.

TTA: How will this theme be investigated throughout the conference?

MM: We have divided the talks into three subsets: The first day of the con-ference will focus on technical analysis and portfolio management, with talksfrom John Murphy and Perry Kaufman. On the second day, John Bollinger,among others, will talk about technical analysis and risk control and on thethird day, David Krell, Trevor Neil and Martin Pring will talk mainly abouttechnical analysis and the markets. Obviously, all the talks at the conferenceare related to the main topic of the event, Technical Analysis in ActivePortfolio Management and Risk Control.

TTA: And what about extra-curricular activities. Is there anything planned?

MM: Lots of them! We have organized three special events for attendeeswith a global pass. On Thursday evening, they are invited to a welcome din-ner with David Krell, president of the International Securities Exchange,one of the largest equity options exchanges. On Friday midday, attendeeswill visit Segovia and its Roman legacy, and will taste a typical medieval lunchwith products from the region Castilla y León. On Saturday evening, therewill be a gala dinner, held at one of the most emblematic buildings inMadrid. Bernard Lietaer will give a presentation on the international curren-cy markets. He is one of the co-founders of the single European currencyand was elected the best currency trader in the nineties by Business Week.Evening visits and many other activities will keep attendees busy all the time!

TTA: What is the best way to register for the conference and what wouldyou say to a potential attendee?

MM: The best way is through the web site www.aeatonline.com. We havealready sold three-quarters of passes so if someone is really interested inattending the conference, he or she should book early to avoid disappoint-ment.

To all professionals involved in Technical Analysis I strongly recommendnot missing the chance to listen to and meet specialists from all over theworld and of keeping up-to-date with new developments in this discipline.Welcome to Madrid!

The Conference Team

Fernando Bolívar, AEAT vice-chairman &

CEO Expert Timing Systems

Marc Michiels, AEAT General Secretary &

Conference Chairman

Jorge Bolívar, AEAT chairman &

TechRules.com CEO

María Pardo, AEAT Marketing Manager

Jorge Pérez, AEAT IT Manager

Mercedes Martínez, AEAT Administration Manager


II FF TT AA CC OO NN FF EE RR EE NN CC EE

All passes

09.00 Talk by Gerald Butrimovitz (US),President of Gerald Butrimovitz andAssociates Advisory Service, “Whatindicators worked post 2003”

10.00 Talk by John Bollinger (US),President of Bollinger CapitalManagement, Inc, "Bollinger Bandsaround the world"

11.00 Coffee break

11.30 Talk by Matthieu Gilbert (CH),Head of Currency Overlay Department,La Compagnie de Trésorerie Benjaminde Rothschild "The integration of quan-titative models in a currency overlayapproach"

Global Pass

12.30 Visit to Segovia province, 6 hourexcursion

Daily Pass

12.30 Talk by Roberto Knop (SP), Head

All passes

09.00 Opening

Global pass

09.30 Walkabout

Daily pass

09.30 Talk by Fernando Bolívar (SP),CEO Expert Timing Systems Int,"Quantitative asset allocation for fund offunds"

10.30 Talk by Jorge Bentué (SP),Finance Professor, TechRules School ofFinance, "TA Strategies for options"

All passes

11.30 Coffee break

12.00 Talk by John Murphy (US),Author of Technical Analysis of theFinancial Markets, "Combining intermar-ket analysis and Exchange Traded Fundsin asset allocation and portfolio manage-ment"

13.00 Talk by Jorge Bolívar (SP),Founder and CEO TechRules.com &President AEAT, "Nearest neighbours.Pattern recognition"

14.00 Spanish lunch

15.30 Talk by Perry Kaufman (US),Author of Trading Systems and Methods,"Portfolio allocation for active traders"

16.30 Talk by Josep Codina (SP),Founder and Director of TASCInvestor, "Current TA trends and tech-niques applied in Spanish portfolio man-agement. A real case study"

17.30 Coffee break

18.00 Round table: Emerging Markets

21.30 Welcome dinner with David Krell(US) (Global Pass)

THURSDAY NOVEMBER 4TA & PORTFOLIO MANAGEMENT

FRIDAY NOVEMBER 5TA & TA & RISK CONTROL

Paella!


of Balance Sheet & Treasury Risk.Barclays Bank, "VaR & EaR when meas-uring and managing risk"

13.30 Lunch

15.00 Talk by Roberto Moro (SP),Independent financial analyst and advi-sor, "Understanding new investmenthabits from Technical Analysis side"

16.00 Talk by Miguel Angel Cicuéndez(SP), Capital Manager and TechnicalAnalyst in Afina-Pentor AV, a memberof Commerzbank group, "Speculativestrategies with volatility, a combinationof technical analysis and market feel-ings"

17.00 Coffee break

17.30 Talk by José Luís Cava (SP),Author of Speculation Systems in StockMarkets, "Elliott, which guidelines dowork?"

Other activities

16.00 DITA Level 1 2 hours exam

All passes

09.00 Japan Hour: Hiroshi Okamoto,Chairman of the Board of Directors ofNippon Technical Analysts Association,"Enhanced trend Analysis using a spe-cial triangle ruler"

Yoshito TED Tetsuda, Deputy GeneralManager, Investment Strategist, FixedIncome Research Department, DaiwaSecurities SMBC Co. Ltd. (JP),"Technical tools of active portfoliomanagement"

10.00 Talk by David Krell (US), DavidKrell is Founder & CEO of theInternational Securities Exchange, "TheInternational Securities Exchange - cata-lyst for change"

11.00 Coffee break

11.30 Talk by Rick Bensignor (US),Chief Technical Strategist, MorganStanley, "Seeing clearly through theclouds: an introduction toIchimokukinkhouyou"

12.30 Talk by Martin Pring (US), Authorof Technical Analysis Explained,"Using technical analysis to minimizerisk and maximize profits"

13.30 International lunch

15.00 Bloomberg talk by Trevor Neil(UK), European Head of TechnicalAnalysis, Bloomberg, "Mixing funda-mentals and technicals - the ultimateweapon?"

16.00 Coffee break

16.30 Talk by Manfred Huebner, FundManager with Deka Investment GmbH& Rolf Wetzer, Senior currency managerwith MEAG Asset Management GmbH(D), "Measuring market sentiment usingthe Sentix Market Index"

17.30 Round table: TA Education

21.30 Gala dinner with Bernard Lietaer(B), Visiting Professor at NaropaUniversity & co-founder of ACCESSFoundation (Global Pass), "The futureof money, beyond greed and scarcity

SATURDAY NOVEMBER 6TA & MARKETS

PP RR OO GG RR AA MM MM EE 22 00 00 44

Segovia Province

Cibeles Square by night


A SHORT HISTORY OF IFTA CONFERENCESThe International Federation of Technical Analysts,Inc. (IFTA) was incorporated in 1986 and is a globalorganization of market analysis societies and associ-ations in 26 countries. This not-for-profit federationhas four main goals, one of which is to "providemeetings and encourage the interchange of material,ideas and information". It was with this goal inmind, that IFTA held the first conference back in1988 and it has continued to do so every year since.

So how have the conferences varied over the 16years? Michael Smyrk, IFTA Business Manager1993-2003, tells us more about what is now themost prestigious TA event of the calendar year.

How many attend these conferences and who arethey?

Numbers have varied enormously, with the result usuallydepending on the health (or otherwise) of the financial mar-kets in the run-up to the booking deadline. Japan, for example,had unusually low numbers because of poor markets before-hand, and the effect of 9/11, which even led to late cancella-tions. Mostly, the attendance by local members of the hostsociety makes up around 1/3rd to 1/2 of the total - so a largehost society has an immediate advantage. Nevertheless, IFTAhas usually tried to place the conference wherever it is mosthelpful to the largest number of TAs, while trying to keep tothe "pendulum" - Year a in North America, Year b in Europe,Year c in Asia/Australasia, Year d in Europe, Year e in NorthAmerica again. No doubt this will change as local member-ships increase/decrease, but traditionally Europe has had thelargest number of TAs overall. Attendees are usually about2/3rds institutional (including data vendors etc), and 1/3rdprivate investors (mainly local).

Do the formats of the conferences vary?

Each Conference in my experience has been different, takingits character from the host society. The host societies havebeen generally allowed to go their own way, apart from stick-ing to a fairly traditional format.

The format itself has not changed much over the years. Themain difference has been in the social events around the mainevent, including "partner programmes" - sometimes very

good, sometimes non-existent - and the final night party/cele-bration, which has sometimes been highly memorable.Unusual parties have included a "Pirate Ship" in Amsterdam, acruise around Manhattan Island & the Statue of Liberty, anddinner at the Sydney Opera House, for example. And specialside visits for delegates and partners have also been wellreceived - such as a reception in the NYSE DirectorsBoardroom.

What have been the best things about the IFTAconferences?

Traditional parts of the Conference that I hope will always bethere include the "Japan Hour", which has helped many peopleto accept and study those special methods, and Ian Notley's"Walkabout", which is both a terrific ice-breaker and a mostuseful way of finding out other people's favoured TA tools -on which subject, I would suggest that methods discussedhave not reflected linear progress, but more a circular re-visit-ing of recurring methodologies - P&F has come and gone andcome and gone again over the years, Chaos Theory was brieflyin favour, ditto Artificial Intelligence. Last year I think we wereback to very traditional tools like RSI, MACD and Stochastics.

A beneficial side-effect of the Conferences, not often men-tioned, is the availability of high-level TA authors and experts,who are willing to talk to anyone - there is enormous goodwill.And a further spin-off is the impact of these "important"people on local data providers, who are made to realise theimportance of what they produce.


DATESThursday 4 November to Saturday 6 Novemberinclusive.

LOCATIONHotel Castellana InterContinental, Madrid

GETTING THEREIf you fly with Iberia, the official conference carrier,you will receive a 30% discount on the full ticketprice. Quote code OSI IB BT4IB21MPE0313 whenmaking your reservation at any Iberia office orthrough your local Iberia phone number.

COSTAttendees can choose to go for one day (€450) orfor the complete program (€1,100 for IFTA col-leagues and €1,250 for non-IFTA colleagues).Special events (welcome dinner with David Krell,Trip to Segovia and Gala dinner with BernardLietaer) are only for those with complete programpasses.

BOOKING AND FURTHER INFORMATIONAll information concerning the conference as wellas registration online can be found at www.aeaton-line.com. If you need more information concerningthe program, you can e-mail the organisers [email protected] POLICY:Registration Fees: No refunds will be given unless there are exceptional circum-stances.

ACCOMODATIONDiscounted accommodation is available at the con-ference headquarters, Hotel CastellanaInterContinental. The Castellana InterContinental isa distinguished hotel in the heart of the city, withcomfortable lounges, rooms and other facilities aswell as fine cusinie. The Castellana InterContinentalsupports the conference through special room ratesfor attendees. To receive these special rates, reserva-tion must be made at www.aeatonline.com

CANCELLATION POLICY:Hotel cancellation policy: A refund of 85% of the total payment will be returned tothose who notify, in writing, by October 1, 2004. A refund of 50% of the total pay-ment will be returned to those who notify, in writing, by October 30, 2004 .Afterthis date no refunds will be given unless there are exceptional circumstances

ESSENTIAL FACTS

Techniques


The technical analysis indicatorsdeveloped by Tom DeMarkenjoy a reputation for reliability

amongst its small group of marketusers that exceeds that of the averageprice or volume indicator. DeMarkremains one of the lesser known mar-ket indicators and is seldom covered intechnical analysis syllabi or textbooks.This is largely because DeMark avail-ability has been confined to the profes-sional market and its reputation hasspread mainly by word-of-mouth.

Kurt Magnus is a DeMark devoteeand applies it to all his FX trading andstrategy decisions. "Probably onlyaround 3% of London traders useDeMark", he says. "This is because theytake time and effort to master and haveto be uploaded onto your screen. Thisis an inconvenience. Consequently,DeMark enjoys a certain degree ofexclusivity and so is not yet part ofmainstream technical analysis theory".

Magnus and his team deal only in for-eign exchange although there is also asmall fixed income desk at Westpac inLondon specialising in the Australianand New Zealand bond markets. Forobvious reasons, the Australian dollarfeatures highly in Magnus’ daily tradingbut he considers that DeMark remainsreliable, no matter how obscure thecurrency cross he may be dealing. "Irecall that a recent backtest of theSequential signals showed them to be

IMPRESSIVE SIGNALS FROM DEMARK

Kurt Magnus, head of foreign exchange sales at Westpac bankin London, discusses Tom DeMark's (TD) SequentialIndicatorTM, his preferred technique for timing position taking in the FX markets.

Techniques


around 70% accurate. DeMark is essen-tially a risk-reward strategy and its stop-loss positioning means that even whenthe indicators occasionally underper-form, losses are cut to a minimum. Inmy experience, the TD SequentialIndicator is more than 70% reliable; itis closer to 90%.”

The DeMark SequentialIndicator Magnus uses Bloomberg charts whosesoftware automatically recognizes anddisplays TD Sequential Indicators asprices change from day to day. TheSequential is perhaps the most com-monly used DeMark indicator and hasan impressive record of identifying andanticipating turning points across theFX, bond, equity and commodity mar-kets. Furthermore, the indicators pro-vide signals not only on a daily, weeklyand monthly basis but also intraday.

The Sequential Indicator identifieswhen a trend is becoming, or hasbecome, exhausted. On daily charts, forexample, DeMark identifies preciselywhich day to enter into a new positionor liquidate an existing one. This totalabsence of ambiguity with regard tomarket timing makes the Sequentialstand out. Using the indicator doesrequire a leap of faith however, as sig-nals often appear prematurely. As such,a buy signal may appear before a down-trend has completed so the trader

Kurt Magnus→→

may have a nervous ride before themarket finally turns. Magnus warns thatit is crucial the indicator is properlyunderstood. "Unless you understandexactly the maths behind the signals,you can make costly errors. There areonly two guys in the London FX mar-ket who can explain these signals withauthority. Jason Perl at UBS in Londonand I often talk to make sure we get itspot on.”

SetupsThe TD Sequential Indicator consistsof two patterns, a TD Setup and a TDCountdown. Setups are the shortest induration, lasting for exactly nine pricebars when completed. For example, abuy Setup exists when there have beennine consecutive price bars in whicheach bar's close is lower than the closefour price bars earlier. When a price barcloses below that of four price barspreviously a '1' appears below the bar.If the next price bar also closes belowthat of four bars earlier a '2' appearsand so on. These appear in Figure 2 ingreen, a chart of EUR/USD fromFebruary to May '04. If before price bar9 is reached a price bar fails to closebelow that of four bars previously thenthe Setup is abandoned and the num-bers are automatically deleted. Oncenine consecutive price bars have beencompleted the trader will be looking fora "perfected" Setup; one that is nowvalid for trading. A buy Setup is per-fected when the low of either price bar8 or 9 is less than the lows of bothprice bars 6 and 7. Perfected sell Setupslook for a high of either price bar 8 or9 that is greater than the highs of both

Techniques


Figure 2.

Figure 1. is a daily chart of the Dow from mid-2003 showing how the DeMark SequentialIndicator appears with price bars. The numbers in green and red represent TD Setups and TDCountdowns respectively. The purple dotted lines are the DeMark stop loss levels automaticallygenerated by the software.

Techniques


price bars 6 and 7. Figure 2 clearlyshows perfected sell Setups in Februaryand April marked with a red arrow.

CountdownsA TD Countdown occurs after a com-pleted Setup. A buy Countdown con-sists of thirteen price bars whose closeis lower than or equal to the low twobars earlier. The corresponding num-bers appear below the price bar. Unlikethe Setup, a Countdown doesn't have to

consist of consecutive days. TheCountdown is a bigger pattern than theSetup in that it can take months for aCountdown to form and often signifiesa larger market move once the trendchanges. Like the Setup, theCountdown also has "perfection" crite-ria. For a buy Countdown this requiresthat the low of price bar 13 be less thanor equal to the close of price bar 8.Similarly, sell perfections require thatthe high of price bar 13 be greater than

or equal to the close of price bar 8.Figure 3 illustrates how signals havebeen generated in EUR/USD sinceJune 2003 using Countdowns. The buysignal generated in September '03 andsell signal in February '04 anticipatedlarge market moves which includedcompleted, yet unperfected Setups.

Stop losses The placing of stop loss levels is a cru-cial component of the SequentialIndicator and they are generated auto-matically only after the completion of aCountdown. For a buy signal, their levelis calculated by identifying the lowestprice bar of the entire Countdown(whether numbered or not) and thensubtracting the low of that price barfrom its high, or the prior price bar'sclose, whichever is the greater. Thisvalue is in turn subtracted from the lowof that same price bar and the criticalstop loss level is established. The stoploss is only executed when there is aclose above the stop loss level followedby a close below it. The next price barmust also open below the stop loss

TD Setup TD CountdownDuration 9 price bars Unlimited

Buy signal

9 consecutive price bar closes

that are less than the close 4

price bars earlier

13 price bars where each close

is less than or equal to the low 2

price bars earlier

Perfection - buy The low of either price bar 8 or

9 must be less than the lows of

both price bars 6 and 7

The low of price bar 13 must be

less than or equal to the close of

price bar 8

Sell signal9 consecutive price bar closes

that are greater than the close 4

price bars earlier

13 price bars where each close

is greater than or equal to the

low 2 price bars earlier

Perfection - sellThe high of either price bar 8 or

9 must be greater than the highs

of both price bars 6 and 7

The high of price bar 13 must be

greater than or equal to the close

of price bar 8

“CONVENTIONAL INDICATORS ARE TYPICALLY TREND FOLLOWERS WHEREAS DEMARK IS

DESIGNED SPECIFICALLY TO ANTICIPATE TREND REVERSALS.” TOM DEMARK

Table 1.

→→

but must also have a low that is belowits open. Magnus concludes, "There issome debate as to what close should beused in determining the stop loss levelas this can have some impact on overallprofits and losses. In my view, theLondon rather than the New Yorkclose is more valid because of thegreater liquidity in the London market,at least as far as foreign exchange isconcerned.”

Including the Sequential Indicator,there are 17 TD indicators in total. TomDemark told The Technical Analyst,"The DeMark indicators are propri-etary market timing tools that are reallyonly available to professional investors.These indicators are not to be confusedwith conventional technical analysisthat relies more upon subjective inter-pretation of price charts. Rather, theyare quantitatively derived and groundedin market psychology and are totallyobjective. Many traders, even if they arefundamentalists, rely upon the indica-tors to time their trading decisions.Conventional indicators are typicallytrend followers whereas DeMark isdesigned specifically to anticipate trendreversals.”

Tom DeMark is president of MarketStudies and has been involved in theinvestment industry for over 30years. He has served as a consultantto the Soros Group, JP Morgan,Citicorp and Goldman Sachs amongothers. In the 1980s he was executivevice president of hedge fund Tudorand for the past eight years has beena special consultant and partner toSAC Capital. www.tomdemark.com

Techniques


Figure 3.

“IN MY EXPERIENCE, THE TDSEQUENTIAL INDICATOR ISMORE THAN 70% RELIABLE; IT IS CLOSER TO 90%.” KURT MAGNUS, WESTPAC

Techniques


THE US PRESIDENTIAL ELECTION CYCLE FACT OR FICTION?

The theory behind The Theory isthat policies announced by thepresident after an election vic-

tory, such as increased taxes and regula-tion, are generally negative for the cor-porate sector and so have a dampeningeffect on stock markets, whereas halfway through the four-year term, thestock market picks up as the presiden-

tial manifesto becomes more corporatefriendly in preparation for the nextelection.

But despite the intuitive explanationand impressive record cited by Hirsch,the Presidential Election Cycle Theoryis now largely discredited for the simplereason that it has been wrong as manytimes as it has been right in the years

since 1967.Recent research by Wing-Keung

Wong and Cehn Dujuan at theUniversity of Singapore looked at thebehaviour of the S&P500 in the yearsleading up to and following an election.Wong and Dujuan's analysis of the pre-vious ten elections dating back toLyndon Johnson in 1966 showed

The impact of the US presidential election on the financial markets is a subject that hastraditionally been the territory of economists. Nevertheless, that hasn't stopped technicalanalysts from attempting to find repeatable patterns. The most noteworthy and oft citedexample is that of Yale Hirsch's Presidential Election Cycle Theory. In 1967, he showedthat stock markets performed better in the second half of the four-year term than the firsthalf in around 70% of cases going as far back as the mid-1800's.

→→

Techniques


that in only half of the cases did theS&P500 enter a bear market in the twoyears following the election.Furthermore, even when the S&P didmove downwards, the cycle period wasoften imprecise and failed to accuratelycoincide with the four-year electioncycle.

Yet the influence of the presidential

election cycle may still survive in aslightly different guise. A more conclu-sive pattern emerges when looking atthe year prior to the election year itself.Table 1. shows that in every year priorto the 15 election years since 1944, theDow has finished higher. In fact, theonly severe loss in a pre-election yeargoing back 84 years occurred in 1931

during the Depression.So why should the Dow rally in the

pre-election year, rather than the elec-tion year, with such regularity? The rea-son may lie in the timing of crucialpresidential policy changes. Year three(the pre-election year) of the presi-dent's four-year term can be crucial inestablishing a favourable economic cli-

President

Election year

Prior year change

Next year change

Roosevelt 1944 +14 +27

Truman 1948 +2 +13

Eisenhower 1952 +14 -4

Eisenhower 1956 +21 -13

Kennedy 1960 +16 +19

Johnson 1964 +17 +11

Nixon 1968 +15 -15

Nixon 1972 +6 -17

Carter 1976 +38 -17

Reagan 1980 +4 -9

Reagan 1984 +20 +28

Bush 1988 +2 +27

Clinton 1992 +20 +14

Clinton 1996 +34 +23

Bush 2000 +25 -7

Bush/Kerry 2004 +25 ?

% change in Dow before and after election years

Table 1.

“THE ONLY SEVERE LOSS IN A PRE-ELECTION YEAR GOING BACK 84 YEARS OCCURRED IN 1931 DURING THE DEPRESSION.”

Techniques


10.0%

8.0%

6.0%

4.0%

2.0%

0.0%

-2.0%

12.0% Republican vs. Democratic AdministrationsDow Jones Industrials Average % Change Since 1901

7.5%

6.0%

9.1%10.0%

7.8%

9.3%

6.8%

8.4%

11.5%

-1.2%All Years Rep Pres Dem Pres Rep Cong Dem Cong Rep P/

Rep C Rep P/Dem C

Rep P/Split C

Dem P/Dem C

Dem P/Rep C

All years

RepPres

DemPres

RepCon

DemCon

Rep PresRep Con

Rep PresDem Con

Rep PresSplit Con

Dem PresDem Con

Dem PresRep Con

7.5 6.0 9.1 10.0 7.8 9.3 6.8 -1.2 8.4 11.5

Dow (% change) since 1901

DDEEMMOOCCRRAATTSS VVEERRSSUUSS RREEPPUUBBLLIICCAANNSS GRIDLOCK IS BEST by Jeffery Hirsch

Who should Wall Street be rooting for? There are six possible scenarios on Capitol Hilldepending on who has the presidency (Republican or Democrat) and who has control ofCongress (Republication, Democrat or split).

Looking at the historical performance of the Dow under Democrat and Republicanpresidents, we see a pattern that is contrary to popular belief. Under Democrats, the Dowhas performed much better than under Republicans. The Dow has historically returned9.1% a year under the Democrats compared to only a 6.0% return under a Republicanpresident. The results are the opposite with a Republican Congress, yielding an average10.0% gain in the Dow compared to a 7.8% return when the Democrats have control ofthe Hill.

With total Republican control of Washington, the Dow has been up on average 9.3%.Democrats in power over the two branches have fared a bit worse with 8.4% gains. Whenpower is split, with a Republican president and a Democratic Congress, the Dow has notdone very well averaging only a 6.8% gain. The best scenario for investors is a Democratin the White House and Republican control of Congress, with average gains of 11.5%.The direst of circumstances occurs with a Republican president and a split Congress,averaging a net loss of 1.2%. There has never been a Democratic president and a splitCongress.Jeffrey A. Hirsch is editor of the Stock Trader's Almanac and Almanac InvestorNewsletter and president of The Hirsch Organization.

mate before the president embarks onhis election campaign, something thatwill consume most of his time in yearfour. Policy measures to boost businessand the economy must be put into placeearly to take effect; waiting until the elec-tion year itself is too late. Any new poli-cies announced in year three will have animmediate affect on the markets and, assuch, this year rather than the electionyear will reap the benefits. A good exam-ple of this is President Bush's State ofthe Union speech in January 2003 inwhich he announced plans for incometax cuts and the elimination of taxes ondividends.

Other major policy changes may alsobe announced mid-term, such as shiftsin the administration. December 2002saw the resignation of TreasurySecretary Paul O'Neill as a result of hisinconsistent US dollar policy statementsand his opposition to the large fiscalstimulus to the US economy that Bushwas then planning.

As for the year immediately followingthe election year, the results are mixed.With the exception of 2001, there havebeen four consecutive post-election yearrallies since 1989. Prior to that, however,the Presidential Election Cycle Theoryhas little to offer in helping to predictstock market direction in the year afteran election. So, with the importantexception of the regularity of the pre-election year rally, in most respects thecycle theory appears redundant. Withthat in mind, ignore it - at least until2007.


Techniques

The kagi chart is a unique kind of linechart designed to filter out short-termmarket noise. It offers an interestingalternative for identify trends, sup-port/resistance levels, and reversals.Although believed to date back to the1870s when the Japanese stock marketbegan trading, kagi was only recentlyintroduced into western TA by SteveNison in his book ‘BeyondCandlesticks’. Going back to its origins,The Technical Analyst asked IkutaroGappo - a kagi chart expert at theJapanese association of technical ana-lysts - to explain how they work andhow they can be used.

The kagi chart is a non-timeseries chart. Non-time seriescharts, which also include such

varieties as the neri, shin-ne (new price)and P&F (point-and-figure), are dimen-sional charts which show stock priceson the vertical axis but indicate no timefactor on the horizontal axis. Thesecharts differ in the degree to whichemphasis is placed on the recording ofprices, but share the aim of identifyingtrends and changes in trends as accu-rately as possible.

How to draw kagi chartsThe kagi chart, which is also called thenehaba (price range) chart, shows mar-ket fluctuations with turns in a line. Itcan be used to forecast stock pricetrends from price changes that exceedeither a certain range or a certain rate.The former is called a fixed price move-ment kagi chart and the latter is called afixed rate kagi chart.

The price range or rate is determinedin advance, e.g. JPY10, JPY20, JPY100,or 5%, 10% or 20%. When the stock

price moves beyond the given rate orrange, a line is drawn to that price.Then as long as the price moves in thesame direction even by a yen or two, thechart is extended accordingly. If theprice moves in the opposite directionbeyond the predetermined range orrate, only then is a new line formedshowing a change in direction. If themove in the opposite direction to thetrend is less than the range or rate, itwill not be regarded as significant andtherefore not recorded onto the chart.The trend prior to a turn is regarded asstill continuing. The closing price isused for kagi charts.

Take, for example, the JPY10 pricerange chart. When the price rises morethan JPY10 from the starting point, ayang (red, thick) line is drawn up to thenew level (no matter how many days ittakes to reach that level). If the pricecontinues to rise, each rise is added tothe red line. If a fall of more thanJPY10 yen occurs (no matter howmany days it takes), a new line (in red)connecting this previous line should bedrawn down to the lower price. Whenthis fall goes below the previous bot-tom, however, the line is changed to ayin (black, thin) line although it remainson the same vertical.

The black line is extended downwardas long as the price continues to declinewithout a rally of more than JPY10. Ifthere is a recovery of more than JPY10,the line changes direction and movesupward to the new price as a black line.If and when this rise exceeds the previ-ous high, the line changes to a yang(red, thick) line from that point.Popular price ranges are JPY5, JPY10,

JPY20, JPY50, JPY100 and JPY200.Popular rate ranges are 1%, 2%, 5%

and 10%. The choice of which to usediffers according to which stock isbeing traded. Greater price and rateranges are used for stocks with higherprices because their upward and down-ward movements are larger. For lowerpriced stocks, smaller price and rateranges are used. The point is to use aprice or rate range that suits one'sneeds.

How to read kagi chart patterns

The line and center point in thekagi chartThe yang (red, thick) line of the kagichart indicates the buy force and a yin(black, thin) line the sell force; when theyang and yin lines are equal, it meansthat the buy and sell forces are bal-anced; when the yang line is longer thanthe yin line, it means that the buy forceis stronger; when it is shorter, the buyforce is weaker; the center point (indi-

INTRODUCTION TO KAGI CHARTSby Ikutaro Gappo

1

Ce

nte

r

Bu

y f

orc

e

2

Se

ll f

orc

e

3

Ba

lan

ce

d f

orc

e

4

Do

min

an

t B

uy

fo

rce

5D

om

ina

nt

Se

ll f

orc

e

Figure 1. Kagi chart and the centre point

Techniques


cated by "X") is the turning pointbetween the buy and the sell forces andthus is very important for judging thestrength of the trend (See Figure 1).

Types of kagi charts

(1) One-stage breakWhen the stock price exceeds theimmediately preceding shoulder, buy-ing is indicated. When the price fallsbelow the immediately preceding waist,selling is indicated. If the price riseswithout crossing the preceding centerpoint (line), it means very strong buyingmomentum. This is a highly reliablebuy signal. Similarly, if the pricedeclines without crossing the centerpoint (line), it is stronger than a mereone-stage break sell signal. See Figure 2.

(2) Double windowWhen the right and left lines do notoverlap, the space between these twolines is called a "window." When thewindow opens on both sides on thesame level, it is called a "double win-dow." When a price rises through adouble window, it confirms a marketbottom. This is an important

ShoulderBuy

(positive turn)

WaistSell

(negative turn)

Center

BuyCenter

Sell

Figure 2. One-stage break

Buy

Double Window

Double Window

Sell

Figure 3. Double windows

Figure 5. Reverse Sanson

Buy

BuyBuy

Buy

Inverse Inverse Tengu

Inverse Okame Shoulder down, waist down

Regular Tengu

Sell

SellSell

Sell

a

a

a

a

b

b

b

bc

cc

c

d

dd

dOkame

Shoulder down, waist down

Figure 4. Sanson

→→

"buy" signal. When the price declinesthrough the double window, it confirmsa ceiling. This is an important "sell" sig-nal. As long as a window is open, themarket is still moving in the same direc-tion. Only when both windows areopen does the move become meaning-ful. See Figure 3.

(3) Sanson (head-and-shoulders)When two shoulders (a, b) and twowaists (c, d) are equal, respectively, thisis called the "regular sanson". It is thesame as the Head and Shoulders top.When the center rises higher, it is calledthe "tengu sanson" and when it islower, it is called the "okame sanson."Sanson with its shoulder (b) and itswaist (d) lower than (a) and (c), respec-tively, is called the "shoulder-down,waist-down sanson." All these varietiesconfirm the top, but the tengu sansonis said to be the most effective.

A reverse sanson is a signal that con-firms the bottoming out of a marketand also has many varieties. See Figures4 and 5.

(4) GokenAs Figure 6 shows, in the buy gokenthe outer waist goes up (1, 2, 3), and sodoes the inner waist (4, 5), surpassingthese five points. The closing price fol-lowing the high (5) (shoulder) repre-sents a buy signal. The selling gokengoes in the reverse order. This form iscompleted when the closing pricedeclines below the low (5) (waist). Inthe buying goken, the pattern (3) (high-er than the center line) is stronger thanthe pattern (3') (dip buying force isstrong). In the selling goken, the pat-tern (3) (lower than the center line) is

Techniques


1

2

2

4 4

5

5

3

3

3

1

Buy

goke

n

Center Center

Sell

Gok

en

3

Figure 6. Goken (Five step points)

1

1

2

4

5

3

2

4

5

3

Figure 8. Successive buy and sell step points

1

2

4

5

3

1

2

4

5

3

Soar

ing

buy

goke

n

Prec

ipita

nt se

ll gok

en

Figure 7. Precipitant Goken

Techniques


stronger than the pattern (3').

(5) Soaring or precipitous gokenIn the soaring goken, the waist declinescontinuously (3, 2, 1) and so does theshoulder (4, 5), as shown in Figure 7.But the price turns upward from thelow (1) and when it exceeds the shoul-der (4), it rises further and surpasses theshoulder (5) in one big movement. Inthe precipitous goken, the shouldergoes up (3, 2, 1) and so does the waist(5,4). From the shoulder (1), the priceplummets and goes below the waist(4)and (5).

(6) Successive buy and sell steppointsThis pattern is formed when the sell(buy) goken is immediately followed by

the buy (sell) goken. The pattern withthe greater price range is said to be bet-ter. See Figure 8.

(7) Numbers in the kagi chartThe number "nine" has special weightin Sakata's chart and all other Japanesecharts. A pattern called the three-stagerise (fall) in the traditional Japanesechart is based on the notion that thestock price tends to rise (fall) in threestages. Each stage is composed of threeminor stages. A movement from floorto ceiling, therefore, includes nine (3 x3) stages (see Figure 9). This emphasison "9" is often applied to the kagi chart.Experience shows that price tends topeak (bottom) in three stages, as inFigure 10.

ConclusionThe world of technical analysis isstrewn with different charting tech-niques. They often provide the sameinformation, but in various formats.Kagi is no different in this respect. Butthe way in which kagi charts illustrateprice action means they are effective infiltering out distracting market noise,while remaining sensitive enough toprovide clear trading signals. The veryfact they have survived since the nine-teenth century is testimony to their ele-gant power as a trading tool.

Ikutaro Gappo is an adviser for theNippon Technical AnalystsAssociation

Bott

om

Bott

om1/2

dro

p

1/3

dro

p

1/2

reco

very

1/2

reco

very

1st

- sta

ge

rise 2n

d -

stag

e ri

se

3rd

- st

age

rise

Top

1st

- sta

ge

fall

2nd

- st

age

fall

3rd

- st

age

fall

1

1

2

2

33

4

4

5

5

6

6

7

7

9

9

8

8

Figure 9. Three-Stage rise and fall

1

1

2

2

3

3

4

4

5

5

6

6

7

7

9

9

8

8

1st

- sta

ge

rise

2nd

- st

age

rise

3rd

- st

age

ris

e

1st

- sta

ge

fall

2nd

- st

age

fall

3rd

- st

age

fal

lFigure 10. The Kagi Ashi of Three-Stage rise and fall

TTA: You first entered the market back in the 1980swhen you joined the institutional bond desk at the FirstNational Bank of Chicago. How has the practice of TAand its perception changed since then?

NC: Technical analysis has become more accepted andaccessible. That doesn't mean that it is highly regardedhowever. The mainstream investment community stillregards TA as some sort of mumbo jumbo of dubiousvalue. Yet even the most vocal opponents of the tech-nique use it in some form or another. The remarkablething is that as a forecasting approach TA is more oftencorrect - across any time frame you select - than straightfundamental analysis. But like Rodney Dangerfield alwayswhined, technical analysts don't get much respect

TTA: Why is that?

NC: In part the problem is largely semantic. Mainstreaminvestment theory and academia tend to dismiss anythingthat is not currently in the established body of theory.They forget that every new addition starts as an idea oran observation - some revolutionary concept. Those witha vested interest are always dismissive. But with time andresearch new ideas are woven into the accepted theory.Then it - TA or any other new idea - is no longer contro-versial. TA is becoming accepted as behavioral financeor quantitative analysis. Many people do not realize thatTA may be graphically represented but much of theanalysis is based on maths. For example, trendlines plotthe slope of a line. Bollinger bands track standard devia-

tions from a mean. Technical analysts don't talk in math-ematical terms but if you investigate the underlyingprocesses, you'll find that TA is seriously mathematical.Terms that sound quantitative or academic seem to makethe concepts more palatable to the fundamentally orient-ed.

Another factor in the acceptance of TA is its readyavailability because of the broad use of computers andtechnology. TA might have seemed more mysteriouswhen analysts drew charts by hand. Now with a comput-er on every desk, charts and technical studies are readilyavailable to everyone. Of course, with those tools athand, younger investment professionals whip throughthe graphics programs and trading platforms without asecond thought. Older financial pros face the dilemma ofnot knowing what their juniors take for granted. It's agenerational problem too.

TTA: You're known for teaching Elliott Wave analysis.Why did you choose this as your preferred tool?

NC: I have found that Elliott Wave is a remarkable tool.Actually it is more of a concept or system than a tool perse. Elliott's research explains that all market price actionis systematic - cyclical - in a predictable way. The mostpersuasive fact for me is that Elliott Wave explains all ofthe features in mainstream TA, especially patterns.

TTA: In what ways have you developed your TA overthe years?

Interview


THE TECHNICAL ANALYST TALKS TO…

Nina Cooper

Nina Cooper is president of the American Association ofProfessional Technical Analysts (AAPTA). She has beenadvising clients, trading and managing money in the capitalmarkets since 1980, in both the United States and London.In 1995, Nina set up Pendragon Research Inc., an independ-ent research firm. Her other undertakings include teachingElliott Wave Theory, phi analysis and advanced stochasticsat the Chicago Mercantile Exchange and writing for ElliottWave International.

Interview

NC: When I first started working with charts in the early80s, I looked at many of the studies that seem intuitive -moving averages, RSI, volume. What I found is that eachhas value but only at certain times. Elliott Wave was thefirst technical approach I found that seemed to be uni-versal - explaining all of the action, even at times whenthe wave counting was not clear.

In recent years I've developed beyond Elliott Wavebecause it does suffer from some shortcomings. Namely,it is ambiguous at times and it gives little reliable insightfor timing purposes. My work has expanded into phianalysis which builds on the fractal structure of marketsbased on phi (1.618). That number is often attributed toFibonacci who in fact identified a summation series withphi properties. Phi has been around for thousands ofyears. It's a pretty amazing number. You can't imaginehow predictive it is. In addition to fractals and phi, Ihave done a great deal of work with stochastics as a tim-ing tool. Most traders do not adequately understand thepower of this study. When combined, fractals, phi andstochastics can forecast what direction the market isgoing, where it will stop and reverse and when. You can'tget much better forecasting information.

TTA: Since we're clearly in conference season at themoment, I'd just like to talk about your involvement withTA societies and your decision to found a new associa-tion earlier this year. What's the background behind this?

NC: I first learned of technical analysis professionalbodies while I was working in the UK. Someone intro-duced me to the STA and I was bowled over by being ina room full of people doing what I (of necessity) hadbeen doing secretly in my job. When I returned to theUSA in 1993 I joined the Market TechniciansAssociation (MTA) and was an active member in theChicago chapter. I held board level positions for a num-ber of years and was the Annual Seminar Chair in 2001.

In 2003 I resigned from the MTA. Shortly afterwards agroup of other technical professionals sharing the sameview decided to form the AAPTA. Why the change?Well, if I can step back and be objective, organizationsare like people, with different personalities and goals. Fora period, an organization may be broad enough to pur-sue many agendas which means that it is inclusive andpermits many points of view. In recent years the MTABoard has become more focused on specific goals.Unfortunately many of us felt that the MTA's pursuit of

growth came at the expense of supporting professionalsin the field. Our needs were not being met. We believedthat a professional organization should be focused on itsprofessionals and practitioners. And so AAPTA wasborn to be a meeting place for professionals who activelyuse technical analysis in their careers.

TTA: So what are the requirements for prospectiveAAPTA members?

NC: To become a member, a person must have at leastseven years of professional experience using technicalanalysis as a major component of their job. This showsthey have a high level of competency and have learnthow to survive professionally. Also of importance, theymust hold to and practice high ethical standards.

TTA: As president, what are your ambitions for theAAPTA?

NC: AAPTA's mandate is pretty simple: our organiza-tion exists as a forum or network to enable professionaltechnical analysts to connect and to interact. We want toshare research, draw on other analysts' expertise and justenjoy the fellowship of those with a common interest.We do not have any evangelical goals like educating thepublic or credentialing individual analysts. We hope to beable to support our members with easy means of com-munication with one another, to help our members withnetwork, help with tools to help strengthen their profes-sional lives. There may be other specific desires that willsurface as AAPTA grows.

With regard to membership, we have more than 50members at present and expect that number to growsteadily. We want to encourage others who meet ourrequirements and value that collegial relationship to joinus and build our organization in the direction the mem-bers' desire. The potential for growth is substantial butgrowth for growth's sake is not on our agenda. There isno target number projected.

For more information on the AAPTA, seewww.aapta.us.

For more information on Pendragon Research, seewww.phicharts.com.


Subject Matters


Most stock market participantswould agree that the characteris-tics of an attractive company

depend to some extent on the sector thecompany is in. For example, in the utilitiessector high dividends may be a soughtafter sign of stability, while in the technol-ogy sector they may be shunned as a signthe company does not have high growthpotential. Even technical indicators mayvary in their usefulness across sectors -companies in some sectors may exhibitshort term mean reversion, while in othersectors long term trends may be moreimportant. In this study, we set out to sys-tematically evaluate what were the bestpredictors of excess returns within each often sectors of the S&P 500 index (usingthe S&P GICS sector definitions). Wefound some expected results and somesurprising ones, as well as substantial dif-ferences between sectors. This articledescribes our methodology and gives anoverview of the results.

MethodologyOur first task was to define a set of candi-date factors. The approach used at NedDavis Research, which we believe to benearly unique in the industry, is to useboth technical indicators and fundamentalaccounting measures to gain insight intothe dynamics of a company's share price.Following this philosophy, we selected 17technical indicators (including short termand long term momentum, stochasticoscillators, and candlestick codes) and 22fundamental measures as a base set of fac-tors. We added three factors which repre-sent known return anomalies (low price,small market cap, and positive coskew-ness) and three which represent measuresof risk. Finally we included three sensitiv-ities to investment "style" and five sensi-tivities to macroeconomic conditions, fora total of 53 factors. Examples of thesefactors are given in Box 1.

We chose to define the fundamental, risk,and anomaly factors so that higher ranksrepresent "better" companies; i.e. highervalues are historically associated with pos-itive excess returns. To accomplish this, weinverted the signs of several familiar ratiossuch as debt/equity and workingcapital/sales. Since the technical factorscan display either trending or mean-revert-ing behavior depending on the sector, thedirection in which these factors are rankedis arbitrary. The same is true of the styleand macroeconomic factors.

Two aspects of this study were crucial toavoid some errors which commonly arisewhen backtesting quantitative models:

With the factors suitably defined, and theabove rules in place, we calculated thereturn to each factor as follows:

Note that performance of the sector as awhole has no effect on the factor return,since the portfolio is always long and shortequal dollar amounts. Table 1 shows theannualized return, standard deviation, andSharpe ratio for each of the factors ineach sector. The remainder of this articlediscusses some themes we found in theresults.

FindingsMean Reversion: The dominant themeamong our purely price-based technicalfactors appears to be reversion to themean. Stocks that were the best perform-ers in their sector in a given month tendedto be the worst performers the followingmonth. See for example Figure 1, whichshows the return to one month momen-tum in the Consumer Staples Sector (theresult of holding the 20% of the sectorwith the highest return the prior monthand shorting the 20% with the lowestreturn the prior month). The steadydownward trend evident on this chartshows that it would have been a winningstrategy to buy the stocks with the lowestrecent returns and short the stocks withthe highest recent returns. Similar resultswere found in almost every sector, butwere most pronounced in Financials,Materials, and Industrials. This is consis-tent with the findings of other research;see for example Jegadeesh (1990).

Another technical factor exhibitingmean-reverting behavior is the candlestickcoding technique introduced inLikhovidov (1999). To implement thistechnique, we retrieve the high, low, clos-ing, and opening prices for the stock ineach month, and then assign to it one of128 possible codes representing the color,body size, upper shadow, and lower shad-ow of the candlestick. We found in eightof the ten sectors that it would have beena consistently profitable strategy to buystocks with the lowest candlestick readings

BACKTESTING PREDICTORS OF THE S&P 500by David Whitaker and Chun Wang

Sort the stocks in each sector by thevalue of the factor at the end of eachmonth - e.g., from highest earningsyield to lowest Form a portfolio that is long the topquintile (20%) and short the bottomquintile of stocks ranked by the factorMeasure the return to this portfolioduring the following monthRe-balance the portfolio at the end ofthe month using updated ranks andsector members

We eliminated survivor bias by usingthe full historical list of sector con-stituents at each point in time (i.e., weincluded companies which later dis-solved, were acquired, or migrated to adifferent sector)We minimized look-ahead bias by onlyusing accounting information on orafter the reporting date. For example,balance sheet items that are measuredas of December 31 but not reporteduntil March 31, are included in thestudy from March 31 forward.

Subject Matters


and short those with the highest readings.See for example figure 2, which shows thereturn to the candlestick factor in thefinancial sector.

Valuation Measures: Perhaps not sur-prisingly, the traditional measures of valu-ation - earnings yield, cash flow yield, andbook to market - performed best in thevalue oriented sectors. Materials,Industrials, Energy, and Consumer Staplesall show positive returns to these factors.Conversely, these factors had little or novalue in the growth oriented sectors, suchas Health Care and Technology. Figure 3compares the cumulative return to earn-ings yield in the Industrials sector to thatin the Technology sector. Clearly, earningsyield has been a good predictor of excessreturns for Industrials but a poor one forTechs.

Efficiency Measures: Similarly, measuresof operating efficiency (such as assetturnover and working capital to sales) per-formed well in the growth oriented sec-tors, but not in the value oriented sectors.Figure 4 shows the return to assetturnover in the Health Care sector vs. thesame factor in the Materials Sector.

Factor Rotatio: We also found evidenceof factors rotating into and out of favor.Figure 5 shows the return to earnings yieldin the Consumer Discretionary sector.The return to this factor was generallypositive, except during a 12 month periodfrom October 1992 - October 1993 andduring the 'bubble' years of 1997 - 2000.Both of these periods showed strong neg-ative returns to this factor. Cash flow yieldand EBITDA/Enterprise value exhibitsimilar patterns in this sector. An interest-ing avenue for future research might be toexamine the predictability of "cyclical"factor returns such as these, looking for

trends, reversals, or underlying economicconditions that would help to predictwhen these factors will be in or out offavor.

Other Findings: One of the more inter-esting results concerns the distribution ofcash to shareholders. Repurchase yield(the sum of stock repurchases by thecompany over the prior 12 months divid-ed by total market capitalization) was aconsistent predictor of excess returns forseveral sectors: Materials, Industrials,Health Care, and Financials. In contrast,dividend yield was not a significant factorfor any sector. In the US, capital gains aretaxed more favorably than dividends. Ourresults may reflect a preference by US tax-able investors to receive cash via repur-chases rather than dividends. It may alsoindicate that company managers correctlyrecognize when their stock is undervaluedand initiate repurchase programs at thesetimes.

We found that the change in inventory(as a percent of total assets) was a goodpredictor for the Industrial and ConsumerDiscretionary sectors, with decreasinginventory being more favorable. Rapidincreases in inventory may signal a reduc-tion in demand for a firm's products, lead-ing to reduced sales and profits in thefuture. Decreasing inventory may signalrising demand. We also found that theaccruals ratio, which represents the non-cash component of reported net income,may be a good indicator of earnings"quality". Stocks with lower accruals ratiosproduced superior returns in theIndustrials, Health Care, IT, and Utilitiessectors.

ConclusionWe found strong support for the idea thatstock selection should be done on a sec-tor-by-sector basis, with the importantpredictors of excess return varying con-siderably by sector. We also found evi-dence of mean reversion in short termreturns, consistent with prior academicresearch. We hope this study provides use-ful information to practitioners andinspires future research into the pre-dictability of stock market returns.

David Whitaker and Chun Wang areanalysts at Ned Davis Research, Inc.

ReferencesJegadeesh, Narasimhan (1990). Evidence ofPredictable Behavior of Security Returns, Journal ofFinance 45, 881-898.

Likhovodov, Viktor (1999). Coding Candlesticks,Technical Analysis of Stocks & CommoditiesNovember 1999, 38-46.

22 Fundamental Factors, e.g.:Earnings Yield, Dividend Yield andDebt/Equity

17 Technical Factors, e.g.:Momentum 1, 2, 3, 6 & 12M; RSI 26 & 52Wand Stochastic 26 & 52W

3 Risk Factors, e.g.:Beta and EPS Stability

3 Anomaly Factors, e.g.:Price and Market Cap

3 Style Factors, e.g.:Value v. Growth and Cyclical v Consumer

5 Macroeconomic Factors, e.g.:Crude Oil; 10-Year T-Note and Credit Spread

Examples

Box 1.

Subject Matters


From the literature survey, it wasfound that those earlier academicstudies tested whether stock prices

follow a random walk by using statisticaltests that are in fact designed to uncoverlinear patterns in stocks prices. However,the lack of linear dependencies does notnecessarily imply the series are random asthere might be other more complexforms of dependencies that cannot bedetected by these standard linear method-ologies. Even Fama (1965) admitted thatlinear modeling techniques have limita-tions as they are not sophisticated enoughto capture complicated patterns that thechartist sees in stock prices.

One of the possible hidden patternsthat went undetected in earlier studies isthat of non-linear dependency. After thefirst evidence of non-linearity reportedby Hinich and Patterson (1985), moreand more evidence has emerged to sug-gest non-linearity is a universal phenome-non. This new feature of the data sup-ports the idea of stock market pre-dictability. In this regard, Lim and Liew(2004) argued that non-linearity favoursnon-linear technical analysis techniques,and their view is further supported by theempirical work of Andrada-Félix et al.(2003) who demonstrated the profitabili-ty of non-linear trading rules. Given the

mounting empirical evidence of pre-dictability, the pendulum has swung infavour of professional analysts, andCochrane (1999) has even labeled stockmarket predictability as a 'new fact infinance'.

Though the issue of stock market pre-dictability is still hotly debated, there is apossible win-win solution for bothgroups. The repeated demonstrations bySchachter et al. (1985) and Hood et al.(1985) via sub-period analysis stronglyhighlights the fact that there are timeswhen market movement is random, whileat other times, the market moves in a sig-nificantly non-random and dependentpattern. Another recent work byAmmermann and Patterson (2003) alsofound that the stock and index returns ofthe Taiwan Stock Exchange follow a ran-dom walk for long periods of time, onlyto be interspersed with brief periods ofstrong linear and/or non-linear depend-ency structures. These findings, on theone hand, suggest that stock market pre-dictability is mainly a short-horizon phe-nomenon, while at the practical level,highlight the relevance of market-timingstrategies.

The main objective of this study is toutilize recent statistical advances, the win-dowed testing procedure, to provide fur-

ther empirical support to the conjectureof Schachter et al. (1985) and Hood et al.(1985) that that there are times whenmarket movement is random and timeswhen it is not. To conserve space, thisarticle only provides a brief discussion ofthe methodology. Interested readers canrefer to Hinich and Patterson (1995) andHinich (1996) for a full theoretical deriva-tion of the test statistics involved. Thepresent methodology is robust for at leastthree reasons: First, the portmanteau cor-relation (denoted as C) and bicorrelation(denoted as H) test statistics employed inthis windowed testing procedure aredesigned to detect linear and non-lineardependency structures in the data respec-tively; Second, it permits a closer exami-nation of the precise time periods whenmarkets moves randomly and those peri-ods when it does not. Third, both the Cand H test statistics have good sampleproperties over short horizons of data.

This study looked at daily closing pricesfor three South Asian stock marketindices: Colombo SE All Share (SriLanka), India BSE National (India) andKarachi SE 100 (Pakistan). These indiceswere collected from Datastream and aredenominated in their respective local cur-rency units for the sample period1/1/1990 to 31/12/2003. From this

WHEN DOES TECHNICAL ANALYSIS WORK …AND WHEN DOESN'T IT? by Kian-Ping Lim

In the financial academic literature, one of the most enduring questions

concerns the predictability of stock prices. Much research has been devot-

ed to forecasting stock prices in order to "beat the market". The general

consensus drawn from earlier empirical work is that stock prices move in a

random fashion, suggesting that analysis of past prices to forecast future

price movement is meaningless because patterns observed in the past

occurred purely by chance. This finding poses a direct challenge to techni-

cal analysts, to the extent of implying their work is of no real value to stock

market investors. However, this hardly makes sense given the wide usage of

technical analysis in the investment world.


Subject Matters

data, the percentage daily returns arecomputed based on the price move fromthe close of one trading day to the next.In the windowed testing procedure, thedata is split into sets of non-overlappingwindows of 35 observations in length,approximately seven trading weeks.

Evidence of random and non-ran-dom walk movementThe results of the window testing arereported in Table 1. The fourth rowshows the number of windows where theproposition of pure noise is rejected bythe C statistic (indicating the presence oflinear dependency structures), with thecorresponding percentage in parenthesis.The statistics for significant H windows(indicating the presence of non-lineardependency structures) are also displayedin the same table. Since both significant Cand H statistics indicate departure from arandom walk, the final row of Table 1provides the total number of windows orsub-periods in which the returns seriesare non-random. A common finding isthat all three South Asian stock returnseries do not follow a random walk all thetime. For instance, in the case of BSEN-AT, 6 out of a total 104 sub-periods(equivalent to 5.77%) move in a signifi-cantly non-random and dependent pat-tern, while for the remaining majority ofsub-periods the market moves along at aclose approximation to a random walk.This corroborates the findings ofAmmermann and Patterson (2003), andprovides additional empirical evidence tosupport the conjecture of Schachter et al.(1985) and Hood et al. (1985). In particu-lar, these three South Asian stock marketsjoin the list of exchanges that at timesmove randomly and at other times donot.

Figure 1: Significant C and H Windows for South Asian Stock Returns Series

→→


Subject Matters

Graphical IllustrationA graphical depiction of the results couldprovide a closer examination of the pre-cise time periods during which the seriesdeviate from a random walk. The his-tograms in Figure 1 show those windows(sub-periods) in which the series are non-random, either due to a significant C or Hstatistic, or both. Since the windowedtesting procedure breaks the full sampleinto equal-length and non-overlappedwindows, it is possible to identify theexact dates when the series under studydeparts from a random walk movement.For instance, in the case of BSENAT,there are 6 windows or sub-periods thatthe series move in a significantly non-ran-dom and dependent pattern. In particu-lar, this occurs in window-8 (11/12/90-28/1/91), window-21 (8/9/92-26/10/92), window-45 (28/11/95-15/1/96), window-59 (14/10/97-1/12/97), window-79 (20/6/00-7/8/00)and window-97 (19/11/02-6/1/03). As awhole, Figure 1 clearly demonstrates that

all three South Asian stock return seriesfollow a random walk for long periods oftime, only to be interspersed with briefperiods of strong linear and/or non-lin-ear dependency structures.

Implications for technical analy-sisThe present study throws some interest-ing light on the ongoing debate of stockmarket predictability. Though the returnsfor the South Asian stock market indicesfollow a random walk for long periods oftime, there were times when it does not,suggesting the potential of profitabilityfor technical trading rules. In particular,during those periods when the marketsmove in a significantly non-random anddependent pattern, it is possible forinvestors to devise a trading rule toexploit those detected linear and non-lin-ear dependencies to earn abnormal ratesof returns. Furthermore, the results high-light the relevance of market-timingstrategies, as the dependency structures

appear only sporadically, and hence sug-gest that predictability is mainly a short-horizon phenomenon.

Kian-Ping Lim is a lecturer at theLabuan School of InternationalBusiness and Finance, UniversitiMalaysia Sabah, Malaysia.

References

Ammermann, P.A. and Patterson, D.M. (2003). Thecross-sectional and cross-temporal universality of non-linear serial dependencies: evidence from world stockindices and the Taiwan Stock Exchange. Pacific-BasinFinance Journal, 11, 175-195.

Andrada-Félix, J., Fernadez-Rodriguez, F., Garcia-Artiles, M.D. and Sosvilla-Rivero, S. (2003). An empiri-cal evaluation of non-linear trading rules. Studies inNonlinear Dynamics and Econometrics, 7(3), Article 4.

Cochrane, J.H. (1999). New facts in finance. EconomicPerspectives, 23, 36-58.

Fama, E.F. (1965). The behavior of stock market prices.Journal of Business, 38, 34-105.

Hinich, M.J. (1996). Testing for dependence in theinput to a linear time series model. Journal ofNonparametric Statistics, 6, 205-221.

Hinich, M.J. and Patterson, D.M. (1985). Evidence ofnonlinearity in daily stock returns. Journal of Businessand Economic Statistics, 3, 69-77.

Hinich, M.J. and Patterson, D.M. (1995). Detectingepochs of transient dependence in white noise. Mimeo.University of Texas at Austin.

Hood, D.C., Andreassen, P. and Schachter, S. (1985).Random and non-random walks on the New YorkStock Exchange. Journal of Economic Behavior andOrganization, 6, 331-338.

Lim, K.P. and Liew, V.K.S. (2004). Nonlinearity favoursnonlinear TA techniques. The Technical Analyst, Mayissue, 38-40.

Schachter, S., Gerin, W., Hood, D.C. and Andreassen, P.(1985). Was the South Sea bubble a random walk?Journal of Economic Behavior and Organization, 6,323-329.

BSENAT

CSEALL

KSE100

Total number of windows

104

104

104

Window length

35

35

35

Number of lags

4

4

4

Significant C windows

2

(1.92%)

20

(19.23%)

6

(5.77%)

Significant H windows

4

(3.85%)

6

(5.77%)

7

(6.73%)

Significant C and H windows

6

(5.77%)

25

(24.04%)

12

(11.54%)

Table 1. Windowed-Test Results for South Asian Stock Returns Series Note: BSENAT- India BSE National; CSEALL- Colombo SE All Share; KSE100- Karachi SE 100.

Book Review


A COMPLETE GUIDE TOTECHNICAL TRADING TACTICS

A Complete Guide to Technical TradingTactics: How to Profit Using Pivot Points,Candlesticks & Other Indicators By John L. Person Published by Wiley Trading 266 pages, £39.99ISBN 0-471-58455-X

John Person's book is available from theTechnical Analyst bookshop at the reducedprice of £33.99 plus P+P. To order please call01730 233870 and quote the "TechnicalAnalyst magazine". Books are usually postedwithin one working day of your order.

John Person's new book falls into two parts; a basic introduc-tion to technical analysis techniques and methods for thenovice, and more interesting sections looking at trading strate-gies that use well know and some lesser well known TA tech-niques. Unlike many of the numerous books published ontechnical analysis, Person's book is well written with clearcharts and easy to follow examples. It is also written purelyfrom a trader's perspective and contains much that will be ofinterest to the professional analyst and trader. Perhaps themost interesting section regarding trading techniques isPerson's presentation of pivot point analysis which, he says, iswidely used amongst day traders, brokerage firms and marketprofessionals in the US. Pivot point analysis is essentially amethod for calculating major support and resistance levels.

A pivot point number (P) is the sum of the high, low andclosing price of a period divided by three. From this numbersupport and resistance levels are derived. For example, the pri-mary resistance level in the next period of trading (ie, day,week, month etc.) is calculated by multiplying P by 2 and sub-tracting the low price for the period. The major support levelis calculated by multiplying P by 2 and subtracting the highprice for the period. The reasoning behind these calculations isthat the pivot point represents an equilibrium around whichtrading occurs in any given period and so the support andresistance levels contain the range of prices when trading veerseither side of this equilibrium in subsequent periods.

As an example, Person uses a monthly sugar futures chartfrom September 2002 to forecast the low of the next month,October. In September, the high was 7.80, the low was 6.40and it closed at 6.63. This produces a pivot point number of6.943 with a corresponding support number of 6.09. The lowfor October was in fact 6.11, just two ticks away. The supportlevel of 6.09 therefore produces the optimal entry point forOctober. However, Person suggests that these levels are bestused in conjunction with conventional chart analysis and goeson to present his own version of pivot point analysis calledP3T (Person's pivot Point Trade signal) that incorporates can-dlesticks and stochastics.

His sections on pyramiding, scale trading and options strate-gies prevents much of the book from going over familiarground. This together with the clarity of writing and presenta-tion means that Person's book is, without doubt, a notchabove the average TA publication. Furthermore, techniquessuch as pivot point analysis will be relatively unfamiliar tomany non-US traders and so the book will offer an insightinto a new and exciting technique to these professionals.

Software Review


MTPredictor has only been around since 2001. Yet it alreadyhas a loyal following of mostly private traders, many in the US,and the company has been able to charge an amount for itsEnd-of-Day (EOD) software that deters many casual or lowcapital private traders. As such, it is aimed squarely at mid tohigh-end private traders and small institutions.

Now, with the release of Real-Time 4.0 (RT), MTPredictorhave simply transferred their "isolation approach" to the realmof intraday trading.

The isolation approachLike EOD, the RT software is built around the company's"isolation approach", which the developer, Steve Griffiths, for-mulated over the 17 years he spent trading on his own account.The isolation approach claims to offer a solution to the manyproblems associated with Elliott Wave trading. Namely:

To overcome these problems, the software looks for just onepart of an Elliott pattern, the ABC correction (or zig-zag), toidentify when a price reversal is imminent. Being able to identi-fy such likely areas of price reversals provides good opportuni-ties for low risk / high reward trades.

Contrary to normal Elliott Wave trading, the software consid-ers it unnecessary to work out how the correction fits into theoverall Elliott count. The thinking being that if you have suc-cessfully identified an ABC correction, any of the next waves,whether it is Wave 1, 3, 5, or Wave C of a larger correction,could all be potentially profitable.

Once the ABC correction is identified, the software can thentell you what the risk/reward ratio is for each of the possiblewave outcomes, giving the trader essential information indeciding whether to take the trade.

There are three key trade set-ups that RT 4.0 looks for:

TS1 - is where the ABC correction is part of a Wave 2 or Bcorrection. The trade set-up aims to take advantage of anensuing Wave 3 (usually the most profitable wave). (Figure 1).

TS2 - is where the ABC correction is part of a Wave 4 correc-tion. The trade set-up aims to take advantage of an ensuingWave 5. (Figure 2).

TS3 - identifies an ABC correction of unknown context. Thetrade set-up aims to take advantage of an ensuing wave ofunknown count, but probably a Wave C or Wave 3 of largerdegree. (Figure 3).

The software is able to scan at three Elliott Wave timescales -minor, intermediate and major. It also gives the option tosearch for ABC corrections where Wave C is terminating orhas terminated in a predicted Wave Price Target (WPT), calcu-lated using Fibonacci retracements/projections.

To add further comfort, the software can scan the marketsfor the above ABC patterns, but with the added requirementthat the last bar on the chart is a red (sell) or blue (buy) bar.These reversal bars are derived from several standard reversalpatterns plus an MTPredictor proprietary oscillator.

Compatibility options and availability should growAt present, RT 4.0 is a stand-alone system that relies on aneSignal datafeed. The company is also in the process of mak-ing the software compatible with Townsend Analytics'RealTick data.

MTPREDICTOR REAL-TIME 4.0

As MTPredictor looks to break into the institutional market with the release of its new real-time software (RT 4.0), we assess the software's viability for professional traders.

forcing wave counts on charts where no wave count is obvious

the difficulty of trading alternative counts

changing wave counts as new data arises, leaving a trader strand-ed with a "wrong" position

having to use more and more complex Elliott Wave analysis tomake it work, e.g. the X wave

Figure 1.

Software Review


But exposure to the professional market is ultimately depend-ent on having its software distributed through the likes ofBloomberg and TraderMade. In this regard, MTPredictor isalready making some headway - it is planning to link up as a"partial plug-in" with TradeStation. This means thatMTPredictor will offer two of the key modules ('Show ElliottWaves' & 'Trade set-ups') as an optional extra for TradeStationusers.

In terms of the data available, there is no problem.Everything you would expect is available through eSignal,including stocks, indices, mutual funds, futures, options, forex.The data is real-time and supplied on a tick-by-tick basis,although the MTPredictor software aggregates the tick datainto time bars down to a 1-minute minimum. There have beena few niggles with the interface between eSignal andMTPredictor but these, we have been told, have been resolved.

Easy to use packageThe software comes on a CD Rom with a hefty training manu-al, which is very clear, if slightly repetitive. It could easily havebeen half the weight it is (though it is now also available indownloadable colour PDF format). But those new to ElliottWave Theory will appreciate the assumption of no priorknowledge.

Once up-and-running, the software can be used at its mostbasic level with ease. This means running scans for the threetrade set-ups and analysing those set-ups with the risk/rewardmodule. But the advanced functions are also fairly intuitive,allowing the trader to do their own manual analysis of graphs,go back in time, put their own Elliott Wave counts onto charts,find likely areas for waves to end and so on.

In addition to these Elliott Wave related functions, RT 4.0also includes other indicators and studies. These includeBollinger Bands, moving averages, RSI, stochastics and volumedata. But even Tony Beckwith, MTPredictor director of salesand marketing, admits they are there to keep people happy.They can be used to give further comfort for the trades thatthe software suggests, but they are, he said, 'peripheral nice tohaves'.

Thorough software and supportThe software is clearly well thought out. Both Beckwith andGriffiths use it to trade on their own account and because ofthis, they have addressed almost every detail or problem that atrader is likely to encounter - either in the software itself, in themanual or on the website.

The trading manual emphasises their practical and down-to-earth approach further. It doesn't promise miracle returns, butsimply says that disciplined risk trading is a means of earningan income, albeit a volatile one. MTPredictor say their tradestypically win 40-50% of the time, but that the winning ones

are at least two to three times the size of the losing ones. Plusthere are the occasional big winners that come along to make itall worth while. Used in conjunction with the sensible risk andmoney management techniques outlined in the manual, RT4.0seems to offer a sound and complete system for trading.

The developer as trading kingIn essence, the MTPredictor software tries to mimic the suc-cessful trading style of its developer, Steve Griffiths. Thisexplains the eclectic mix of extras that can be scanned for -DOJIs, inside days, 80/20 days and minor pullbacks,

Figure 2.

Figure 3.

→→


Software Review

which can all be used to provide further confirmation of thetrade set-ups.

There is no doubt RT4.0 is very good software, offeringclear trading signals with precise entry and exit points. Thelong periods when no trade set-up is found are to be wel-comed since they prevent over-trading. But even the best soft-ware can throw up spurious results and MTPredictor is no dif-ferent. The company recognises this and urges all its customersto check the Elliott Waves to make sure they look correctaccording to Elliott Wave theory. It also encourages its cus-tomers to make sure the trade does not run contrary to theoverall market context, and if it is, that it can be justified.

This is clearly the side of trading that calls on human judge-ment and as previous research has shown, this may be themost important factor in separating the successful trader fromthe unsuccessful one. In recognition of this, MTPredictor pro-vides daily reports and "hotComm Web seminars" on themembers section of the website. The daily report providesgood insight into the way Steve Griffiths is thinking and helpsthe trader develop the necessary skills that allow him to choosewhich set-ups to trade.

TradingAll this counts for nothing if it doesn't make money.MTPredictor has carried out its own tests, the results of whichare presented in brief in Table 1. (Further details can beobtained from MTPredictor, upon request).

The results are certainly impressive. To provide a basis for

replicable assessment, very prescriptive rules were used aboutwhich trades to take and how to manage them once they wereopen. These rules can be found on their website (called"Trading Guidelines"). Readers should be aware, however, thatthe sample is not statistically large nor necessarily representa-tive of using the software outside of the sampling period. TheTechnical Analyst can not verify or endorse these results.

SummaryThere is no doubt that MTPredictor will appeal to novice orrelatively inexperienced traders. And rather like your local sci-entology outfit, it should also thrive on unsuccessful traderswho have lost their way and are looking for confidence andfirm direction.

But can it break into the wider professional market? Thereare a few availability issues that need to be addressed first, butultimately if the software has a positive impact on a trader'sP&L, then there is no reason why not. Early indications aregood and the success of its EOD version lends further weightto MTPredictor's argument.

Price $1,995 for year one, available on a 30-day money back trial(minus $95 administration fee). Year two onwards at $495 perannum. Price does not include datafeed.

www.mtpredictor.com

RT 4.0 Test: Trade Record for US index and ETF trading, 26 July to 25 August 2004

Number of Trades

Number of Winning

Trades

Number of Losing Trades

Profit on wining trades

(units of risk)

Loss on losing

trades (one unit of risk per trade)

Total P/L

TS1 Set-up

TS2 Set-up

TS3 Set-up

Total

381132

207

17

46.752246

-20-7

-17

26.751529

184

15

81 37 44 114.75 -44 70.75

Table 1. Note: 1) Securities – ES (E-mini S&P 500â); NQ (E-mini Nasdaq-100â); YM (mini-Dowâ futures); SPY (Exchange Traded Fund tracking S&P 500);QQQ (Exchange Traded Fund tracking Nasdaq-100); DIA (Exchange Traded Fund tracking Dow Jones Industrials). 2) Timeframes – 3min. and 5 min. 3) Noaccount has been taken of slippage and commissions. 4) P/L risk units rounded to nearest 0.25. 5) No trades left open overnight (closed at session end if neces-sary) 6) No trades were actually taken

MTPredictor TM

The software solution for complete trading excellence

Designed exclusively to find, assess and manage only the very best trades in stocks, currencies and commodities

This is the type of trade MTPredictor can automatically uncover for you….

A Profit of approximately 7x the initial risk required to take the trade, ignoring

slippage and commissions, in the UK stock GKN (October 2003)

TAKE CONTROL OF YOUR TRADING WITH THE NEW MTPREDICTOR 4.0 SERIES!

MTPredictor Ltd www.mtpredictor.com [email protected] Tel +44 (0) 208 9776191

End-of-Day and Real-time programs with automatic routines for:

· Ideal trades: Find exceptional set-ups with outstanding Risk/Reward prospects· Ideal trade size: Control your position size· Ideal trade management: Display the exit stop strategy on-screen· Ideal trading psychology: Consistent, logical trading, time after time· Systematic Elliott Wave software: Avoid the pitfalls of standard Elliott analysis· Advanced strategies: Expert trade opportunities and management plans



COMMITMENTS OF TRADERS REPORT25 May - 7 September 2004Futures only (open interest)Non-commercial net long positions and spot rates

10-year US Treasury Source: CBOT

Dow Jones Industrial Average Source: CBOT

5-year US Treasury Source: CBOT

Swiss franc Source: CME

Pound sterling Source: CME Yen Source: CME

-250000

-200000

-150000

-100000

-50000

0

50000

100000

150000

May-25 Jun-08 Jun-22 Jul-06 Jul-20 Aug-03 Aug-17 Aug-31

3.00

3.20

3.40

3.60

3.80

4.00

4.20

4.40

4.60

4.80

5.00

10-yr Treasury Spot

0

50000

100000

150000

200000

250000

300000


2.40

2.60

2.80

3.00

3.20

3.40

3.60

3.80

4.00

4.20

5-yr Treasury Spot

-6000

-5000

-4000

-3000

-2000

-1000

0

1000

2000


9600

9800

10000

10200

10400

10600

10800

11000

DJIA Spot

-15000

-10000

-5000

0

5000

10000

15000

20000

25000

30000


1.21

1.22

1.23

1.24

1.25

1.26

1.27

1.28

1.29

Swiss franc Spot

0

5000

10000

15000

20000

25000

30000

35000


1.70

1.72

1.74

1.76

1.78

1.80

1.82

1.84

1.86

1.88

Pound sterling Spot

-15000

-10000

-5000

0

5000

10000


100

105

110

115

120

125

Japanese yen Spot



Euro Source: CME

Nasdaq Source: CME

3-month eurodollar Source: CME

Nikkei Source: CME

Gold Source: CEI US dollar index Source: NYCE

0

5000

10000

15000

20000

25000

30000

35000

40000


1.18

1.19

1.20

1.21

1.22

1.23

1.24

1.25

Euro Spot

-800000

-600000

-400000

-200000

0

200000

400000

600000


0.00

0.20

0.40

0.60

0.80

1.00

1.20

1.40

1.60

1.80

2.00

3-month eurodollar Spot

-14000

-12000

-10000

-8000

-6000

-4000

-2000

0

2000

4000


1650

1700

1750

1800

1850

1900

1950

2000

2050

2100

Nasdaq Spot

-1000

-500

0

500

1000

1500

2000

2500

3000

3500

4000

4500


10000

10200

10400

10600

10800

11000

11200

11400

11600

11800

12000

Nikkei Spot

0

10000

20000

30000

40000

50000

60000

70000

80000

90000


365

370

375

380

385

390

395

400

405

410

415

Gold Spot

-6000

-4000

-2000

0

2000

4000

6000


113

113.5

114

114.5

115

115.5

116

116.5

117

US dollar index Spot



LONG-TERM TECHNICALS

Provided by Thomas Anthonj, ABN Amro, Amsterdam

As long as the market remains below 1.8067 we have toexpect lower prices towards 1.7375, 1.6958 and maybe even1.6676 as a completing C-leg down of a bigger A-B-C correc-tion pattern.

GBP-USD

Breaking above the last top at 112.34 the overall negativepicture has almost been neutralized. But unless necklineresistance has also been cleared we are still in danger oftesting the last bottom at 101.25 or the H+S target at 95.75.

USD-JPY

Two key-reversal weeks down plus the break below 1.2334delivered strong evidence that we are at least displaying abigger 4th wave setback with a minimum target of 1.1179.The latest rebound has to be classified as a B-wave rallyonly as long as 1.2652 has not been broken decisively.

EUR-USD

Showing a strong bounce up from a Fibonacci-support clusterand closing above neckline resistance one year ago the mar-ket confirmed a medium-term bottom in place. But hitting key-resistance at 1161/77 and closing below trend line support itis very likely that the market performs an internal 4th wavecorrection down towards 1020 before the up-trend is expect-ed to resume.

S&P 500



Breaking above the old 12081 top the market deliveredstrong evidence for a long-term turnaround. But missing a2nd wave setback we have to expect more corrective actionover the coming weeks and months particularly as long asthe market remains below 11969.

Nikkei

The market seems to perform an internal 3rd wave impulseup with 45.95-47.26 as the next immediate target zone. Abreak above would re-open the upside towards 56.52-58.64before any bigger setback could be expected again.

Brent Crude Oil

Stalling right at key-resistance the market retreated, whichsignaled that we are due for a setback that will most likelyform a 4th wave down to 9479 over the next few weeks.Only a break and close above 10525/71 would now signal astraight resumption of the old bull-trend.

Dow Jones

Failing to stabilize above 2099 the market retreated in whatlooks like an internal 4th wave setback as long as trendlinesupport holds. A decisive break and a close thereunderthough would call for a much deeper setback.

Nasdaq


8-10 OCTOBER

Course:AAPTA 1st annual conference,

Phoenix Arizona

Organiser: AAPTA

Contact: [email protected]

13OCTOBER

Event:STA meeting

Organiser: Society of Technical Analysts


25/26OCTOBER

Course:Technical analysis and charting

Organiser: Chartwatch


28/29 OCTOBER

Course:Advanced technical analysis

Organiser: Chartwatch


4-6NOVEMBER

Event:IFTA Conference, Madrid

Organiser: AEAT


10NOVEMBER

Event:STA meeting

Organiser: Society of Technical Analysts


10 NOVEMBER

Course:Introduction to technical analysis

Organiser: 7city


15NOVEMBER

Course:Introduction to technical analysis

Organiser: Quorum Training


25/26NOVEMBER

Course:An introduction to charting &

technical analysis

Organiser: International Petroleum Exchange


For submissions please email us at: [email protected] *All venues are in London, unless otherwise stated.

TRAINING AND EVENTS DIARY

Training and Events Diary

technical analyst

Documents