machine learning for quantitative finance€¦ · machine learning for quantitative finance: fast...

25
Fast derivative pricing Machine Learning for Quantitative Finance Sofie Reyners Joint work with Jan De Spiegeleer, Dilip Madan and Wim Schoutens

Upload: others

Post on 02-Oct-2020

24 views

Category:

Documents


1 download

TRANSCRIPT

Page 1: Machine Learning for Quantitative Finance€¦ · Machine Learning for Quantitative Finance: Fast Derivative Pricing, Hedging and Fitting, Quantitative Finance, forthcoming. 22 Machine

Fast derivative pricing

Machine Learningfor Quantitative Finance

Sofie Reyners

Joint work with Jan De Spiegeleer, Dilip Madan andWim Schoutens

Page 2: Machine Learning for Quantitative Finance€¦ · Machine Learning for Quantitative Finance: Fast Derivative Pricing, Hedging and Fitting, Quantitative Finance, forthcoming. 22 Machine

Derivative pricing is time-consuming...

I Vanilla option pricing• European-type→ Fast Fourier transform

• American-type→ Tree methods

I Exotic option pricing→ Monte Carlo simulations

1 Machine Learning for Quantitative Finance

Page 3: Machine Learning for Quantitative Finance€¦ · Machine Learning for Quantitative Finance: Fast Derivative Pricing, Hedging and Fitting, Quantitative Finance, forthcoming. 22 Machine

... but time is money!

time-consuming algorithms

continuously moving markets

→ prices are outdated when available, overnight calculationscannot be performed in one night, ...

2 Machine Learning for Quantitative Finance

Page 4: Machine Learning for Quantitative Finance€¦ · Machine Learning for Quantitative Finance: Fast Derivative Pricing, Hedging and Fitting, Quantitative Finance, forthcoming. 22 Machine

Let a machine learn the pricing function

product, market andmodel parameters model price

time-consuming method

machine learning

Expensive pricing function is summarized with machine learning.

3 Machine Learning for Quantitative Finance

Page 5: Machine Learning for Quantitative Finance€¦ · Machine Learning for Quantitative Finance: Fast Derivative Pricing, Hedging and Fitting, Quantitative Finance, forthcoming. 22 Machine

Let a machine learn the pricing function

product, market andmodel parameters model price

time-consuming method

machine learning

When training is completed, prediction is extremely fast!

3 Machine Learning for Quantitative Finance

Page 6: Machine Learning for Quantitative Finance€¦ · Machine Learning for Quantitative Finance: Fast Derivative Pricing, Hedging and Fitting, Quantitative Finance, forthcoming. 22 Machine
Page 7: Machine Learning for Quantitative Finance€¦ · Machine Learning for Quantitative Finance: Fast Derivative Pricing, Hedging and Fitting, Quantitative Finance, forthcoming. 22 Machine

Gaussian process regression (GPR)

Consider a training set

(X,y) = {(xi, yi) | i = 1, . . . , n}.

Find a relation between inputs and outputs:

yi = f(xi) + εi

where f(x) is a Gaussian process and εi ∼ N (0, σ2n) are i.i.d. random

variables representing the noise in the data.

4 Machine Learning for Quantitative Finance

Page 8: Machine Learning for Quantitative Finance€¦ · Machine Learning for Quantitative Finance: Fast Derivative Pricing, Hedging and Fitting, Quantitative Finance, forthcoming. 22 Machine

Gaussian process

A Gaussian process f(x) is a, possibly infinite, collection of randomvariables, any finite subset of it having a joint Gaussian distribution.

I Mean function: m(x) = E[f(x)

]I Kernel function: k(x,x′) = Cov(f(x), f(x′))

=⇒ f(x) ∼ GP (m(x), k(x,x′))

5 Machine Learning for Quantitative Finance

Page 9: Machine Learning for Quantitative Finance€¦ · Machine Learning for Quantitative Finance: Fast Derivative Pricing, Hedging and Fitting, Quantitative Finance, forthcoming. 22 Machine

Gaussian process

If f(x) ∼ GP (0, k(x,x′)), then

f ∼ N (0,K(X,X))

where (X,f) = {(xi, fi) | i = 1, . . . , n} is a sample from f(x) and

K(X,X) =

k(x1,x1) . . . k(x1,xn)... . . . ...

k(xn,x1) . . . k(xn,xn)

6 Machine Learning for Quantitative Finance

Page 10: Machine Learning for Quantitative Finance€¦ · Machine Learning for Quantitative Finance: Fast Derivative Pricing, Hedging and Fitting, Quantitative Finance, forthcoming. 22 Machine

GPR: a Bayesian method

I Don’t model the relation as one function, but as a distributionover functions.

I Procedure:

1 Start from a prior GP→ prior knowledge: smooth function, periodic function, ...→ prior distribution over functions

2 Include observed data points

3 Compute a posterior GP

7 Machine Learning for Quantitative Finance

Page 11: Machine Learning for Quantitative Finance€¦ · Machine Learning for Quantitative Finance: Fast Derivative Pricing, Hedging and Fitting, Quantitative Finance, forthcoming. 22 Machine

Posterior distribution

Only consider functions that agree with the data.

I Take new inputs X∗, with corresponding (unknown) functionvalues f∗

I Joint distribution of training outputs and function values:[yf∗

]∼ N

(0 ,

[K(X,X) + σ2

nI K(X,X∗)K(X∗, X) K(X∗, X∗)

])

8 Machine Learning for Quantitative Finance

Page 12: Machine Learning for Quantitative Finance€¦ · Machine Learning for Quantitative Finance: Fast Derivative Pricing, Hedging and Fitting, Quantitative Finance, forthcoming. 22 Machine

Posterior distribution

I Condition on the observations:

f∗|X∗, X,y ∼ N(µ,Σ

)with

µ = K(X∗, X)[K(X,X) + σ2

nI]−1y

Σ = K(X∗, X∗)−K(X∗, X)[K(X,X) + σ2

nI]−1

K(X,X∗)

9 Machine Learning for Quantitative Finance

Page 13: Machine Learning for Quantitative Finance€¦ · Machine Learning for Quantitative Finance: Fast Derivative Pricing, Hedging and Fitting, Quantitative Finance, forthcoming. 22 Machine

Kernel function

Squared exponential kernel function

k(x,x′) = σ2f exp

(−|x− x′|2

2`2

)

with hyperparameters σf and `:I σ2

f = signal varianceI ` = length-scale parameter

→ Hyperparameters (including σn) are estimated from the trainingdata, usually with MLE.

10 Machine Learning for Quantitative Finance

Page 14: Machine Learning for Quantitative Finance€¦ · Machine Learning for Quantitative Finance: Fast Derivative Pricing, Hedging and Fitting, Quantitative Finance, forthcoming. 22 Machine

Mean function

Often set to zero, but can be modelled using basis functions h(x).

g(x) = f(x) + h(x)Tβ ∼ GP (h(x)Tβ, k(x,x′))

whereI f(x) ∼ GP (0, k(x,x′))I β should be estimated from the training data

Common choice:h(x) = (1, x, x2)

11 Machine Learning for Quantitative Finance

Page 15: Machine Learning for Quantitative Finance€¦ · Machine Learning for Quantitative Finance: Fast Derivative Pricing, Hedging and Fitting, Quantitative Finance, forthcoming. 22 Machine

Application set-up

I Construct a training set:

product, market andmodel parameters model price

sample n randomcombinations xi

compute ncorresponding prices yi

time-consuming method

I Fit a Gaussian process regression (GPR) model.

I Fast prediction of new model prices.

12 Machine Learning for Quantitative Finance

Page 16: Machine Learning for Quantitative Finance€¦ · Machine Learning for Quantitative Finance: Fast Derivative Pricing, Hedging and Fitting, Quantitative Finance, forthcoming. 22 Machine

Pricing European call options

Training set:

Product/market VG Heston

K ∈ [40%, 160%] σ ∈ [0.05, 0.45] κ ∈ [1.4, 2.6]T ∈ [11M, 1Y ] ν ∈ [0.55, 0.95] ρ ∈ [−0.85,−0.55]r ∈ [1.5%, 2.5%] θ ∈ [−0.35,−0.05] θ ∈ [0.45, 0.75]q ∈ [0%, 5%] η ∈ [0.01, 0.1]

v0 ∈ [0.01, 0.1]

→ sample n values of each parameter→ calculate n FFT-based model prices

13 Machine Learning for Quantitative Finance

Page 17: Machine Learning for Quantitative Finance€¦ · Machine Learning for Quantitative Finance: Fast Derivative Pricing, Hedging and Fitting, Quantitative Finance, forthcoming. 22 Machine

Pricing European call options

I Fit the GPR model

K, T , r − q,model parameters

FFTmodel price

GPR

I Construct a test set:• Similarly as training set• Slightly smaller parameter intervals

14 Machine Learning for Quantitative Finance

Page 18: Machine Learning for Quantitative Finance€¦ · Machine Learning for Quantitative Finance: Fast Derivative Pricing, Hedging and Fitting, Quantitative Finance, forthcoming. 22 Machine

Out-of-sample prediction

(a) Variance Gamma (b) Heston

→ model trained on 10 000 points, tested on 100 000 points.

15 Machine Learning for Quantitative Finance

Page 19: Machine Learning for Quantitative Finance€¦ · Machine Learning for Quantitative Finance: Fast Derivative Pricing, Hedging and Fitting, Quantitative Finance, forthcoming. 22 Machine

Performance summary

VG Heston

Size of training set 5000 10 000 20 000 5000 10 000 20 000

In-sample predictionMAE 0.0016 0.0017 0.0013 0.0036 0.0027 0.0033AAE 2.5763e-04 1.9627e-04 1.4747e-04 5.2260e-04 4.0347e-04 3.4524e-04

Out-of-sample predictionMAE 0.0028 0.0022 0.0016 0.0060 0.0054 0.0048AAE 2.2508e-04 1.6942e-04 1.2828e-04 5.8991e-04 4.4112e-04 3.6623e-04

Speed-up × 30 × 15 × 7 × 40 × 20 × 10

withMAE = max {|ECF F T (i)− ECGP R(i)|, i = 1, . . . , n}

AAE = 1n

n∑i=1

|ECF F T (i)− ECGP R(i)|

16 Machine Learning for Quantitative Finance

Page 20: Machine Learning for Quantitative Finance€¦ · Machine Learning for Quantitative Finance: Fast Derivative Pricing, Hedging and Fitting, Quantitative Finance, forthcoming. 22 Machine

Pricing American options

I Put options with strike K and maturity T

I Use binomial tree model (daily steps) with volatility

σ ∈ [0.05, 0.55]

K, T , r, q, σ binomial treemodel price

GPR

17 Machine Learning for Quantitative Finance

Page 21: Machine Learning for Quantitative Finance€¦ · Machine Learning for Quantitative Finance: Fast Derivative Pricing, Hedging and Fitting, Quantitative Finance, forthcoming. 22 Machine

Out-of-sample performance

MAE 0.0086

AAE 9.1684e-04

Speed-up ×70

→ model trained on 10 000 points, tested on 100 000 points.

18 Machine Learning for Quantitative Finance

Page 22: Machine Learning for Quantitative Finance€¦ · Machine Learning for Quantitative Finance: Fast Derivative Pricing, Hedging and Fitting, Quantitative Finance, forthcoming. 22 Machine

Pricing barrier options

I Down-and-out barrier put options with barrier level H, strike Kand maturity T with

H ∈ [55%, 99%]

I Use Monte Carlo simulation, according to Heston’s model

H, K, T , r, q,κ, ρ, θ, η, v0

MC Hestonmodel price

GPR

19 Machine Learning for Quantitative Finance

Page 23: Machine Learning for Quantitative Finance€¦ · Machine Learning for Quantitative Finance: Fast Derivative Pricing, Hedging and Fitting, Quantitative Finance, forthcoming. 22 Machine

Out-of-sample performance

MAE 0.0086

AAE 6.7386e-04

Speed-up ×5850

→ model trained and tested on 10 000 points.

20 Machine Learning for Quantitative Finance

Page 24: Machine Learning for Quantitative Finance€¦ · Machine Learning for Quantitative Finance: Fast Derivative Pricing, Hedging and Fitting, Quantitative Finance, forthcoming. 22 Machine

Conclusion

I Time-consuming pricing methods

I Gaussian process regression• Matrix inversion• Hyperparameter optimization

I Apply GPR on existing methods• Speed-up of several orders of magnitude• Some trade-off with accuracy

21 Machine Learning for Quantitative Finance

Page 25: Machine Learning for Quantitative Finance€¦ · Machine Learning for Quantitative Finance: Fast Derivative Pricing, Hedging and Fitting, Quantitative Finance, forthcoming. 22 Machine

Thank you!

More information:

De Spiegeleer, J., Madan, D. B., Reyners, S. and Schoutens, W. (2018),Machine Learning for Quantitative Finance: Fast Derivative Pricing, Hedgingand Fitting, Quantitative Finance, forthcoming.

22 Machine Learning for Quantitative Finance