SlideShare a Scribd company logo
1 of 59
Download to read offline
How deep generative models can help quants reduce
the risk of overfitting?
Talk at QuantUniversity Autumn School 2020
Gautier Marti
HKML Research
27 October 2020
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 1 / 59
Table of contents
1 Introduction
2 GANs explained
GANs Milestones & Major Achievements
How do GANs work?
3 Applications of GANs for Quants
Generating Synthetic Datasets to Avoid Strategy Overfitting
Generating Alternative Realistic Historical Paths
for Risk Estimation
4 Current State of the Art and Limitations
5 Conclusion and Questions?
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 2 / 59
Section 1
Introduction
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 3 / 59
Generative Adversarial Network’s Definition & Goal
(Deep) generative models are a broad class of models.
For the sake of clarity, we will only consider Generative Adversarial
Networks in this presentation.
Definition: Generative Adversarial Network (GAN)
A generative adversarial network (GAN) is a class of machine learning frame-
works where two neural networks (usually known as the generator and the
discriminator) contest with each other in a zero-sum game (i.e. one agent’s
gain is another agent’s loss).
Main purpose of a GAN: Generating realistic synthetic data
Given a training set, this technique learns to generate new data with the
same statistics as the training set.
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 4 / 59
Introduction (3 min. video)
https://www.youtube.com/watch?v=97B8tuHwLY0
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 5 / 59
Scope of the presentation
In this talk, we will focus on
doing a review of GANs success so far (mostly image generation),
explaining how Generative Adversarial Networks (GANs) work.
Then, we will present
applications of these models for quant researchers / traders use cases.
Finally, we shall briefly discuss
the current limitations and challenges to be overcome for a broader
adoption of these models in the industry.
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 6 / 59
Section 2
GANs explained
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 7 / 59
Subsection 1
GANs Milestones & Major Achievements
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 8 / 59
GANs: A relatively new Deep Learning model (2014)
Goodfellow, Ian, et al. “Generative adversarial nets.”
Advances in neural information processing systems. 2014.
Cited by 22354 papers as of 18 September 2020.
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 9 / 59
Text to Photo-realistic Image Synthesis (2016)
https://arxiv.org/pdf/1612.03242.pdf
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 10 / 59
Deepfakes (2017)
Ctrl Shift Face (YouTube channel)
https://www.youtube.com/channel/UCKpH0CKltc73e4wh0_pgL3g
You Won’t Believe What Obama Says In This Video!
https://www.youtube.com/watch?v=cQ54GDm1eL0
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 11 / 59
Edmond de Belamy (2018)
Created by 3 students
First artwork created using
Artificial Intelligence to be
featured in a Christie’s auction
Sold for USD 432,500
Signed
minG maxD Ex [log(D(x))] +
Ez [log(1 − D(G(z)))]
In French, “bel ami” means
“good fellow”, a pun-tribute to
Ian Goodfellow, the creator of
GANs
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 12 / 59
StyleGAN (2019) & StyleGAN2 (2020)
GANs can generate realistic fake human faces:
Give it a try: https://thispersondoesnotexist.com/
NVIDIA StyleGAN paper: https://arxiv.org/pdf/1812.04948.pdf
NVIDIA StyleGAN2 paper: https://arxiv.org/pdf/1912.04958.pdf
Remark. In December 2019, Facebook took down a network of accounts
with false identities, and mentioned that some of them had used profile
pictures created with artificial intelligence.
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 13 / 59
Speech2Face & Wav2Pix (2019)
GANs conditioned on speech of a person can output a realistic face with
its correct gender, ethnicity, and approximate age:
Relevant papers:
Wav2Pix: https://arxiv.org/pdf/1903.10195.pdf
Speech2Face: https://arxiv.org/pdf/1905.09773.pdf
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 14 / 59
GameGAN (2020)
GameGAN is able to learn Pac-Man dynamics and produce a visually
consistent simulation of the game:
NVIDIA paper https://arxiv.org/pdf/2005.12126.pdf
https://www.youtube.com/watch?v=4OzJUNsPx60
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 15 / 59
Subsection 2
How do GANs work?
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 16 / 59
A GAN basic architecture
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 17 / 59
GAN training – step by step tutorial
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 18 / 59
GAN training – step by step tutorial
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 19 / 59
GAN training – step by step tutorial
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 20 / 59
GAN training – step by step tutorial
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 21 / 59
GAN training – step by step tutorial
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 22 / 59
GAN training – step by step tutorial
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 23 / 59
GAN training – step by step tutorial
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 24 / 59
GAN training – step by step tutorial
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 25 / 59
GAN training – step by step tutorial
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 26 / 59
GAN training – step by step tutorial
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 27 / 59
GAN training – step by step tutorial
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 28 / 59
GAN training – step by step tutorial
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 29 / 59
GAN training – step by step tutorial
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 30 / 59
GAN training – step by step tutorial
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 31 / 59
GAN training – step by step tutorial
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 32 / 59
GAN training – step by step tutorial
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 33 / 59
GAN training – step by step tutorial
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 34 / 59
GAN training – step by step tutorial
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 35 / 59
GAN training – step by step tutorial
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 36 / 59
GAN training – step by step tutorial
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 37 / 59
GAN training – step by step tutorial
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 38 / 59
GAN training – step by step tutorial
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 39 / 59
Conditional Generative Adversarial Nets
Seminal paper: https://arxiv.org/pdf/1411.1784.pdf
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 40 / 59
Conditional Generative Adversarial Nets
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 41 / 59
Section 3
Applications of GANs for Quants
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 42 / 59
Subsection 1
Generating Synthetic Datasets to Avoid Strategy Overfitting
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 43 / 59
Strategy Overfitting
Scenario
In September 2020, a naive or unscrupulous strat presents to his
manager a new strategy: The strategy would have initiated a massive
short in February 2020, and would have bet big on a rally starting late
March 2020.
The strat sells the strategy to his manager as able to pick-up early
signals of market sell-offs and bounce-backs thanks to advanced
machine learning and alternative data (obviously, what else?).
The manager, excited, cannot wait but to deploy capital to it.
How to fight against this industry-wide fallacy?
Well-thought and carefully designed incentives schemes
(e.g. reward a sound research framework rather than impressive in-sample backtests)
Counterfactual thinking, realistic simulations of alternative paths
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 44 / 59
Strategy Overfitting – Simulations of Alternative Paths (1)
Some alternative data: Second-hand car market in Hong Kong
A tabular dataset X
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 45 / 59
Strategy Overfitting – Simulations of Alternative Paths (2)
econ quant idea: When inventory builds up (more people are selling their
cars than buying), then the HSI plummets over the next quarter.
math quant idea: My black-box ML algorithm predicts that HSI decreases
whenever two Ferraris are sold right after a Bugatti Veyron comes to
market, based on the given dataset.
Using GANs, we can sample new datasets ˆX ∼ X, or even ( ˆX, ˆy) ∼ (X, y),
where y can be the HSI performance or any variable we aim at predicting
(e.g. market+industry residualized returns of the respective carmakers).
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 46 / 59
Strategy Overfitting – Simulations of Alternative Paths (3)
Method 1: Strategy stability
Use a GAN to learn the distribution (X, y) of the tabular data
(cf. https://arxiv.org/pdf/1907.00503.pdf)
From the calibrated GAN, generate ˆX(n)
, ˆy(n)
N
n=1
synthetic datasets
Backtest the strategy on the N datasets, and collect perf. metrics
Analyze the perf. metrics; Discard the strategy if not stable.
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 47 / 59
Strategy Overfitting – Simulations of Alternative Paths (4)
Method 1bis: Find whether a strategy is robust and obvious given data
As the strat-quant team manager, provide each one of the N quants
with a GAN-generated copy of the original data.
They have to work on it independently.
Will they find the same strategy? With roughly the same parameters?
If so, the strategy is robust, but also likely to be crowded given the data.
Test it, and run it in production, on the original data.
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 48 / 59
Strategy Overfitting – Simulations of Alternative Paths (5)
In Koshiyama et al. (https://arxiv.org/pdf/1901.01751.pdf):
Generate N synthetic datasets:
Method 2: Strategy fine-tuning
Find the parameters which maximize the average perf. over the N datasets
Method 2bis: Strategy combination (ensemble models)
Fit a model on each of the N datasets
Predict using an ensemble of the N trained models
GAN vs. Stationary Bootstrap (for generating the N synthetic datasets)
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 49 / 59
Subsection 2
Generating Alternative Realistic Historical Paths
for Risk Estimation
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 50 / 59
Risk-based Portfolio Allocation (1)
Scenario
You want to systematically allocate capital to your strategies.
Literature claims that such or such method works better than all
others by providing a dubious backtest.
When you horse race the various allocation methods, results are not
stable across periods and universes.
How can we conclude anything useful? Realistic simulations!
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 51 / 59
CorrGAN: Sampling Realistic Financial Correlation
Matrices Using Generative Adversarial Networks
6 stylized facts of empirical
financial correlation matrices
Marchenko-Pastur-like distribution, except for:
a very large first eigenvalue,
a couple of other large eigenvalues
Perron-Frobenius property
positive correlations
Hierarchical structure of clusters
Scale-free property of the degrees in MST
Original paper:
https://arxiv.org/pdf/
1910.09504.pdf
Implemented in the MlFinlab
python package:
https://hudsonthames.
org/mlfinlab/
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 52 / 59
Risk-based Portfolio Allocation (2)
Method 3: Testing Portfolio Allocation on Alternative Historical Paths
Use a GAN to generate realistic synthetic correlation matrices
(cf. https://arxiv.org/pdf/1910.09504.pdf)
Generate time series verifying the correlation structure
Estimate portfolio weights (in-sample), measure out-of-sample risk
Analyze performance of the portfolio allocation methods
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 53 / 59
Risk-based Portfolio Allocation (3)
Question: Can we predict when a method (say the Hierarchical Risk
Parity (HRP)) outperforms another one (say naive risk parity)? And why?
Method 4: Understanding the outperformance of a method over another
Extract features from the underlying correlation matrix
Fit a ML model features → outperformance
(using train, validation, test datasets)
Verify if there is any good predictability
Explain the predictions as a function of features
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 54 / 59
Risk-based Portfolio Allocation (4)
Concrete example: HRP vs. naive risk parity
High values for the cophenetic correlation coefficient are characteristic of a strong
hierarchical structure. Thus, HRP outperforms naive risk parity when the
underlying DGP has a strong hierarchical structure (nested clusters).
cf. Jochen Papenbrock recent work for similar applications of Explainable AI (XAI) in finance (https://firamis.de/)
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 55 / 59
Section 4
Current State of the Art and Limitations
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 56 / 59
Current State of the Art and Limitations
This discussion is by nature technical, but we can highlight the following
limitations which are relevant for quants:
GANs can generate realistic tabular datasets (2019), but models
trained on synthetic data only were shown inferior to the ones trained
on the original data
We know how to GAN-generate realistic synthetic financial time series
(e.g. S&P 500 returns);
We know how to GAN-generate realistic synthetic financial correlation
matrices (of the S&P 500 constituents, for example);
However, we do not know yet how to GAN-generate realistic
multivariate financial time series, i.e. verifying both the time series
and the cross-sectional stylized facts
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 57 / 59
Section 5
Conclusion and Questions?
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 58 / 59
Conclusion and Questions?
Deep generative models and GANs in particular are an exciting new
technology.
In finance, they are actively researched in a few places but results are
not widely advertised.
We have to rely on top tech companies and academic labs to drive
the fundamental understanding and improvements of these models.
contact@hkml-research.com
Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 59 / 59

More Related Content

Similar to How deep generative models can help quants reduce the risk of overfitting?

Part 2: The Art of Gathering Virtually: Frameworks for Tech Architecture & Ex...
Part 2: The Art of Gathering Virtually: Frameworks for Tech Architecture & Ex...Part 2: The Art of Gathering Virtually: Frameworks for Tech Architecture & Ex...
Part 2: The Art of Gathering Virtually: Frameworks for Tech Architecture & Ex...Kent Bye
 
Human-Centered AI: Scalable, Interactive Tools for Interpretation and Attribu...
Human-Centered AI: Scalable, Interactive Tools for Interpretation and Attribu...Human-Centered AI: Scalable, Interactive Tools for Interpretation and Attribu...
Human-Centered AI: Scalable, Interactive Tools for Interpretation and Attribu...polochau
 
International Conference on Data Science and Machine Learning (DSML 2020)
International Conference on Data Science and Machine Learning (DSML 2020) International Conference on Data Science and Machine Learning (DSML 2020)
International Conference on Data Science and Machine Learning (DSML 2020) ijdms
 
ECAI 2014 Tutorial on a behavioral analysis tool for agent-based simulations ...
ECAI 2014 Tutorial on a behavioral analysis tool for agent-based simulations ...ECAI 2014 Tutorial on a behavioral analysis tool for agent-based simulations ...
ECAI 2014 Tutorial on a behavioral analysis tool for agent-based simulations ...tamasmahr
 
AML & ALMA: Project Overview
AML & ALMA: Project OverviewAML & ALMA: Project Overview
AML & ALMA: Project OvervieweProsima
 
International Conference on Data Mining and Machine Learning (DMML 2020)
International Conference on Data Mining and Machine Learning (DMML 2020)International Conference on Data Mining and Machine Learning (DMML 2020)
International Conference on Data Mining and Machine Learning (DMML 2020)dannyijwest
 
Models_and_Learning.pdf
Models_and_Learning.pdfModels_and_Learning.pdf
Models_and_Learning.pdfStarman Anoa
 
Ultimaker's Future Strategy
Ultimaker's Future StrategyUltimaker's Future Strategy
Ultimaker's Future StrategyMartijn Stolk
 
A Review on the Determinants of a suitable Chatbot Framework- Empirical evide...
A Review on the Determinants of a suitable Chatbot Framework- Empirical evide...A Review on the Determinants of a suitable Chatbot Framework- Empirical evide...
A Review on the Determinants of a suitable Chatbot Framework- Empirical evide...IRJET Journal
 
Introduction to Machine Learning
Introduction to Machine LearningIntroduction to Machine Learning
Introduction to Machine LearningDr. Radhey Shyam
 
Uses of ChatGPT in Marketing
Uses of ChatGPT in MarketingUses of ChatGPT in Marketing
Uses of ChatGPT in MarketingJoseArrunategui3
 
Uses of ChatGPT in Marketing
Uses of ChatGPT in MarketingUses of ChatGPT in Marketing
Uses of ChatGPT in MarketingJoseArrunategui3
 
International Conference on Data Mining and Machine Learning (DMML 2020)
International Conference on Data Mining and Machine Learning (DMML 2020)International Conference on Data Mining and Machine Learning (DMML 2020)
International Conference on Data Mining and Machine Learning (DMML 2020)dannyijwest
 
Call for Papers - International Conference on Machine Learning, NLP and Data ...
Call for Papers - International Conference on Machine Learning, NLP and Data ...Call for Papers - International Conference on Machine Learning, NLP and Data ...
Call for Papers - International Conference on Machine Learning, NLP and Data ...IJNSA Journal
 
Call for Paper - 3rd International conference on Big Data, Machine learning a...
Call for Paper - 3rd International conference on Big Data, Machine learning a...Call for Paper - 3rd International conference on Big Data, Machine learning a...
Call for Paper - 3rd International conference on Big Data, Machine learning a...mlaij
 
Towards new solutions for scientific computing: the case of Julia
Towards new solutions for scientific computing: the case of JuliaTowards new solutions for scientific computing: the case of Julia
Towards new solutions for scientific computing: the case of JuliaMaurizio Tomasi
 
Massive Data Analysis- Challenges and Applications
Massive Data Analysis- Challenges and ApplicationsMassive Data Analysis- Challenges and Applications
Massive Data Analysis- Challenges and ApplicationsVijay Raghavan
 
CALL FOR PAPERS - 8th International Conference on Computational Science and E...
CALL FOR PAPERS - 8th International Conference on Computational Science and E...CALL FOR PAPERS - 8th International Conference on Computational Science and E...
CALL FOR PAPERS - 8th International Conference on Computational Science and E...AIRCC Publishing Corporation
 

Similar to How deep generative models can help quants reduce the risk of overfitting? (20)

Part 2: The Art of Gathering Virtually: Frameworks for Tech Architecture & Ex...
Part 2: The Art of Gathering Virtually: Frameworks for Tech Architecture & Ex...Part 2: The Art of Gathering Virtually: Frameworks for Tech Architecture & Ex...
Part 2: The Art of Gathering Virtually: Frameworks for Tech Architecture & Ex...
 
Human-Centered AI: Scalable, Interactive Tools for Interpretation and Attribu...
Human-Centered AI: Scalable, Interactive Tools for Interpretation and Attribu...Human-Centered AI: Scalable, Interactive Tools for Interpretation and Attribu...
Human-Centered AI: Scalable, Interactive Tools for Interpretation and Attribu...
 
Impact of Industry 4.0.pptx
Impact of Industry 4.0.pptxImpact of Industry 4.0.pptx
Impact of Industry 4.0.pptx
 
International Conference on Data Science and Machine Learning (DSML 2020)
International Conference on Data Science and Machine Learning (DSML 2020) International Conference on Data Science and Machine Learning (DSML 2020)
International Conference on Data Science and Machine Learning (DSML 2020)
 
ECAI 2014 Tutorial on a behavioral analysis tool for agent-based simulations ...
ECAI 2014 Tutorial on a behavioral analysis tool for agent-based simulations ...ECAI 2014 Tutorial on a behavioral analysis tool for agent-based simulations ...
ECAI 2014 Tutorial on a behavioral analysis tool for agent-based simulations ...
 
Strategic market expansion path en
Strategic market expansion path enStrategic market expansion path en
Strategic market expansion path en
 
AML & ALMA: Project Overview
AML & ALMA: Project OverviewAML & ALMA: Project Overview
AML & ALMA: Project Overview
 
International Conference on Data Mining and Machine Learning (DMML 2020)
International Conference on Data Mining and Machine Learning (DMML 2020)International Conference on Data Mining and Machine Learning (DMML 2020)
International Conference on Data Mining and Machine Learning (DMML 2020)
 
Models_and_Learning.pdf
Models_and_Learning.pdfModels_and_Learning.pdf
Models_and_Learning.pdf
 
Ultimaker's Future Strategy
Ultimaker's Future StrategyUltimaker's Future Strategy
Ultimaker's Future Strategy
 
A Review on the Determinants of a suitable Chatbot Framework- Empirical evide...
A Review on the Determinants of a suitable Chatbot Framework- Empirical evide...A Review on the Determinants of a suitable Chatbot Framework- Empirical evide...
A Review on the Determinants of a suitable Chatbot Framework- Empirical evide...
 
Introduction to Machine Learning
Introduction to Machine LearningIntroduction to Machine Learning
Introduction to Machine Learning
 
Uses of ChatGPT in Marketing
Uses of ChatGPT in MarketingUses of ChatGPT in Marketing
Uses of ChatGPT in Marketing
 
Uses of ChatGPT in Marketing
Uses of ChatGPT in MarketingUses of ChatGPT in Marketing
Uses of ChatGPT in Marketing
 
International Conference on Data Mining and Machine Learning (DMML 2020)
International Conference on Data Mining and Machine Learning (DMML 2020)International Conference on Data Mining and Machine Learning (DMML 2020)
International Conference on Data Mining and Machine Learning (DMML 2020)
 
Call for Papers - International Conference on Machine Learning, NLP and Data ...
Call for Papers - International Conference on Machine Learning, NLP and Data ...Call for Papers - International Conference on Machine Learning, NLP and Data ...
Call for Papers - International Conference on Machine Learning, NLP and Data ...
 
Call for Paper - 3rd International conference on Big Data, Machine learning a...
Call for Paper - 3rd International conference on Big Data, Machine learning a...Call for Paper - 3rd International conference on Big Data, Machine learning a...
Call for Paper - 3rd International conference on Big Data, Machine learning a...
 
Towards new solutions for scientific computing: the case of Julia
Towards new solutions for scientific computing: the case of JuliaTowards new solutions for scientific computing: the case of Julia
Towards new solutions for scientific computing: the case of Julia
 
Massive Data Analysis- Challenges and Applications
Massive Data Analysis- Challenges and ApplicationsMassive Data Analysis- Challenges and Applications
Massive Data Analysis- Challenges and Applications
 
CALL FOR PAPERS - 8th International Conference on Computational Science and E...
CALL FOR PAPERS - 8th International Conference on Computational Science and E...CALL FOR PAPERS - 8th International Conference on Computational Science and E...
CALL FOR PAPERS - 8th International Conference on Computational Science and E...
 

More from Gautier Marti

Using Large Language Models in 10 Lines of Code
Using Large Language Models in 10 Lines of CodeUsing Large Language Models in 10 Lines of Code
Using Large Language Models in 10 Lines of CodeGautier Marti
 
What deep learning can bring to...
What deep learning can bring to...What deep learning can bring to...
What deep learning can bring to...Gautier Marti
 
A quick demo of Top2Vec With application on 2020 10-K business descriptions
A quick demo of Top2Vec With application on 2020 10-K business descriptionsA quick demo of Top2Vec With application on 2020 10-K business descriptions
A quick demo of Top2Vec With application on 2020 10-K business descriptionsGautier Marti
 
Autoregressive Convolutional Neural Networks for Asynchronous Time Series
Autoregressive Convolutional Neural Networks for Asynchronous Time SeriesAutoregressive Convolutional Neural Networks for Asynchronous Time Series
Autoregressive Convolutional Neural Networks for Asynchronous Time SeriesGautier Marti
 
Some contributions to the clustering of financial time series - Applications ...
Some contributions to the clustering of financial time series - Applications ...Some contributions to the clustering of financial time series - Applications ...
Some contributions to the clustering of financial time series - Applications ...Gautier Marti
 
Clustering CDS: algorithms, distances, stability and convergence rates
Clustering CDS: algorithms, distances, stability and convergence ratesClustering CDS: algorithms, distances, stability and convergence rates
Clustering CDS: algorithms, distances, stability and convergence ratesGautier Marti
 
Clustering Financial Time Series using their Correlations and their Distribut...
Clustering Financial Time Series using their Correlations and their Distribut...Clustering Financial Time Series using their Correlations and their Distribut...
Clustering Financial Time Series using their Correlations and their Distribut...Gautier Marti
 
A closer look at correlations
A closer look at correlationsA closer look at correlations
A closer look at correlationsGautier Marti
 
Clustering Financial Time Series: How Long is Enough?
Clustering Financial Time Series: How Long is Enough?Clustering Financial Time Series: How Long is Enough?
Clustering Financial Time Series: How Long is Enough?Gautier Marti
 
Optimal Transport vs. Fisher-Rao distance between Copulas
Optimal Transport vs. Fisher-Rao distance between CopulasOptimal Transport vs. Fisher-Rao distance between Copulas
Optimal Transport vs. Fisher-Rao distance between CopulasGautier Marti
 
On Clustering Financial Time Series - Beyond Correlation
On Clustering Financial Time Series - Beyond CorrelationOn Clustering Financial Time Series - Beyond Correlation
On Clustering Financial Time Series - Beyond CorrelationGautier Marti
 
Optimal Transport between Copulas for Clustering Time Series
Optimal Transport between Copulas for Clustering Time SeriesOptimal Transport between Copulas for Clustering Time Series
Optimal Transport between Copulas for Clustering Time SeriesGautier Marti
 
On the stability of clustering financial time series
On the stability of clustering financial time seriesOn the stability of clustering financial time series
On the stability of clustering financial time seriesGautier Marti
 
Clustering Random Walk Time Series
Clustering Random Walk Time SeriesClustering Random Walk Time Series
Clustering Random Walk Time SeriesGautier Marti
 
On clustering financial time series - A need for distances between dependent ...
On clustering financial time series - A need for distances between dependent ...On clustering financial time series - A need for distances between dependent ...
On clustering financial time series - A need for distances between dependent ...Gautier Marti
 

More from Gautier Marti (15)

Using Large Language Models in 10 Lines of Code
Using Large Language Models in 10 Lines of CodeUsing Large Language Models in 10 Lines of Code
Using Large Language Models in 10 Lines of Code
 
What deep learning can bring to...
What deep learning can bring to...What deep learning can bring to...
What deep learning can bring to...
 
A quick demo of Top2Vec With application on 2020 10-K business descriptions
A quick demo of Top2Vec With application on 2020 10-K business descriptionsA quick demo of Top2Vec With application on 2020 10-K business descriptions
A quick demo of Top2Vec With application on 2020 10-K business descriptions
 
Autoregressive Convolutional Neural Networks for Asynchronous Time Series
Autoregressive Convolutional Neural Networks for Asynchronous Time SeriesAutoregressive Convolutional Neural Networks for Asynchronous Time Series
Autoregressive Convolutional Neural Networks for Asynchronous Time Series
 
Some contributions to the clustering of financial time series - Applications ...
Some contributions to the clustering of financial time series - Applications ...Some contributions to the clustering of financial time series - Applications ...
Some contributions to the clustering of financial time series - Applications ...
 
Clustering CDS: algorithms, distances, stability and convergence rates
Clustering CDS: algorithms, distances, stability and convergence ratesClustering CDS: algorithms, distances, stability and convergence rates
Clustering CDS: algorithms, distances, stability and convergence rates
 
Clustering Financial Time Series using their Correlations and their Distribut...
Clustering Financial Time Series using their Correlations and their Distribut...Clustering Financial Time Series using their Correlations and their Distribut...
Clustering Financial Time Series using their Correlations and their Distribut...
 
A closer look at correlations
A closer look at correlationsA closer look at correlations
A closer look at correlations
 
Clustering Financial Time Series: How Long is Enough?
Clustering Financial Time Series: How Long is Enough?Clustering Financial Time Series: How Long is Enough?
Clustering Financial Time Series: How Long is Enough?
 
Optimal Transport vs. Fisher-Rao distance between Copulas
Optimal Transport vs. Fisher-Rao distance between CopulasOptimal Transport vs. Fisher-Rao distance between Copulas
Optimal Transport vs. Fisher-Rao distance between Copulas
 
On Clustering Financial Time Series - Beyond Correlation
On Clustering Financial Time Series - Beyond CorrelationOn Clustering Financial Time Series - Beyond Correlation
On Clustering Financial Time Series - Beyond Correlation
 
Optimal Transport between Copulas for Clustering Time Series
Optimal Transport between Copulas for Clustering Time SeriesOptimal Transport between Copulas for Clustering Time Series
Optimal Transport between Copulas for Clustering Time Series
 
On the stability of clustering financial time series
On the stability of clustering financial time seriesOn the stability of clustering financial time series
On the stability of clustering financial time series
 
Clustering Random Walk Time Series
Clustering Random Walk Time SeriesClustering Random Walk Time Series
Clustering Random Walk Time Series
 
On clustering financial time series - A need for distances between dependent ...
On clustering financial time series - A need for distances between dependent ...On clustering financial time series - A need for distances between dependent ...
On clustering financial time series - A need for distances between dependent ...
 

Recently uploaded

ASML's Taxonomy Adventure by Daniel Canter
ASML's Taxonomy Adventure by Daniel CanterASML's Taxonomy Adventure by Daniel Canter
ASML's Taxonomy Adventure by Daniel Cantervoginip
 
Generative AI for Social Good at Open Data Science East 2024
Generative AI for Social Good at Open Data Science East 2024Generative AI for Social Good at Open Data Science East 2024
Generative AI for Social Good at Open Data Science East 2024Colleen Farrelly
 
9711147426✨Call In girls Gurgaon Sector 31. SCO 25 escort service
9711147426✨Call In girls Gurgaon Sector 31. SCO 25 escort service9711147426✨Call In girls Gurgaon Sector 31. SCO 25 escort service
9711147426✨Call In girls Gurgaon Sector 31. SCO 25 escort servicejennyeacort
 
NLP Data Science Project Presentation:Predicting Heart Disease with NLP Data ...
NLP Data Science Project Presentation:Predicting Heart Disease with NLP Data ...NLP Data Science Project Presentation:Predicting Heart Disease with NLP Data ...
NLP Data Science Project Presentation:Predicting Heart Disease with NLP Data ...Boston Institute of Analytics
 
Identifying Appropriate Test Statistics Involving Population Mean
Identifying Appropriate Test Statistics Involving Population MeanIdentifying Appropriate Test Statistics Involving Population Mean
Identifying Appropriate Test Statistics Involving Population MeanMYRABACSAFRA2
 
毕业文凭制作#回国入职#diploma#degree澳洲中央昆士兰大学毕业证成绩单pdf电子版制作修改#毕业文凭制作#回国入职#diploma#degree
毕业文凭制作#回国入职#diploma#degree澳洲中央昆士兰大学毕业证成绩单pdf电子版制作修改#毕业文凭制作#回国入职#diploma#degree毕业文凭制作#回国入职#diploma#degree澳洲中央昆士兰大学毕业证成绩单pdf电子版制作修改#毕业文凭制作#回国入职#diploma#degree
毕业文凭制作#回国入职#diploma#degree澳洲中央昆士兰大学毕业证成绩单pdf电子版制作修改#毕业文凭制作#回国入职#diploma#degreeyuu sss
 
PKS-TGC-1084-630 - Stage 1 Proposal.pptx
PKS-TGC-1084-630 - Stage 1 Proposal.pptxPKS-TGC-1084-630 - Stage 1 Proposal.pptx
PKS-TGC-1084-630 - Stage 1 Proposal.pptxPramod Kumar Srivastava
 
RS 9000 Call In girls Dwarka Mor (DELHI)⇛9711147426🔝Delhi
RS 9000 Call In girls Dwarka Mor (DELHI)⇛9711147426🔝DelhiRS 9000 Call In girls Dwarka Mor (DELHI)⇛9711147426🔝Delhi
RS 9000 Call In girls Dwarka Mor (DELHI)⇛9711147426🔝Delhijennyeacort
 
Indian Call Girls in Abu Dhabi O5286O24O8 Call Girls in Abu Dhabi By Independ...
Indian Call Girls in Abu Dhabi O5286O24O8 Call Girls in Abu Dhabi By Independ...Indian Call Girls in Abu Dhabi O5286O24O8 Call Girls in Abu Dhabi By Independ...
Indian Call Girls in Abu Dhabi O5286O24O8 Call Girls in Abu Dhabi By Independ...dajasot375
 
Effects of Smartphone Addiction on the Academic Performances of Grades 9 to 1...
Effects of Smartphone Addiction on the Academic Performances of Grades 9 to 1...Effects of Smartphone Addiction on the Academic Performances of Grades 9 to 1...
Effects of Smartphone Addiction on the Academic Performances of Grades 9 to 1...limedy534
 
科罗拉多大学波尔得分校毕业证学位证成绩单-可办理
科罗拉多大学波尔得分校毕业证学位证成绩单-可办理科罗拉多大学波尔得分校毕业证学位证成绩单-可办理
科罗拉多大学波尔得分校毕业证学位证成绩单-可办理e4aez8ss
 
Advanced Machine Learning for Business Professionals
Advanced Machine Learning for Business ProfessionalsAdvanced Machine Learning for Business Professionals
Advanced Machine Learning for Business ProfessionalsVICTOR MAESTRE RAMIREZ
 
Machine learning classification ppt.ppt
Machine learning classification  ppt.pptMachine learning classification  ppt.ppt
Machine learning classification ppt.pptamreenkhanum0307
 
Statistics, Data Analysis, and Decision Modeling, 5th edition by James R. Eva...
Statistics, Data Analysis, and Decision Modeling, 5th edition by James R. Eva...Statistics, Data Analysis, and Decision Modeling, 5th edition by James R. Eva...
Statistics, Data Analysis, and Decision Modeling, 5th edition by James R. Eva...ssuserf63bd7
 
Learn How Data Science Changes Our World
Learn How Data Science Changes Our WorldLearn How Data Science Changes Our World
Learn How Data Science Changes Our WorldEduminds Learning
 
专业一比一美国俄亥俄大学毕业证成绩单pdf电子版制作修改
专业一比一美国俄亥俄大学毕业证成绩单pdf电子版制作修改专业一比一美国俄亥俄大学毕业证成绩单pdf电子版制作修改
专业一比一美国俄亥俄大学毕业证成绩单pdf电子版制作修改yuu sss
 
RABBIT: A CLI tool for identifying bots based on their GitHub events.
RABBIT: A CLI tool for identifying bots based on their GitHub events.RABBIT: A CLI tool for identifying bots based on their GitHub events.
RABBIT: A CLI tool for identifying bots based on their GitHub events.natarajan8993
 
While-For-loop in python used in college
While-For-loop in python used in collegeWhile-For-loop in python used in college
While-For-loop in python used in collegessuser7a7cd61
 
Building on a FAIRly Strong Foundation to Connect Academic Research to Transl...
Building on a FAIRly Strong Foundation to Connect Academic Research to Transl...Building on a FAIRly Strong Foundation to Connect Academic Research to Transl...
Building on a FAIRly Strong Foundation to Connect Academic Research to Transl...Jack DiGiovanna
 
Call Us ➥97111√47426🤳Call Girls in Aerocity (Delhi NCR)
Call Us ➥97111√47426🤳Call Girls in Aerocity (Delhi NCR)Call Us ➥97111√47426🤳Call Girls in Aerocity (Delhi NCR)
Call Us ➥97111√47426🤳Call Girls in Aerocity (Delhi NCR)jennyeacort
 

Recently uploaded (20)

ASML's Taxonomy Adventure by Daniel Canter
ASML's Taxonomy Adventure by Daniel CanterASML's Taxonomy Adventure by Daniel Canter
ASML's Taxonomy Adventure by Daniel Canter
 
Generative AI for Social Good at Open Data Science East 2024
Generative AI for Social Good at Open Data Science East 2024Generative AI for Social Good at Open Data Science East 2024
Generative AI for Social Good at Open Data Science East 2024
 
9711147426✨Call In girls Gurgaon Sector 31. SCO 25 escort service
9711147426✨Call In girls Gurgaon Sector 31. SCO 25 escort service9711147426✨Call In girls Gurgaon Sector 31. SCO 25 escort service
9711147426✨Call In girls Gurgaon Sector 31. SCO 25 escort service
 
NLP Data Science Project Presentation:Predicting Heart Disease with NLP Data ...
NLP Data Science Project Presentation:Predicting Heart Disease with NLP Data ...NLP Data Science Project Presentation:Predicting Heart Disease with NLP Data ...
NLP Data Science Project Presentation:Predicting Heart Disease with NLP Data ...
 
Identifying Appropriate Test Statistics Involving Population Mean
Identifying Appropriate Test Statistics Involving Population MeanIdentifying Appropriate Test Statistics Involving Population Mean
Identifying Appropriate Test Statistics Involving Population Mean
 
毕业文凭制作#回国入职#diploma#degree澳洲中央昆士兰大学毕业证成绩单pdf电子版制作修改#毕业文凭制作#回国入职#diploma#degree
毕业文凭制作#回国入职#diploma#degree澳洲中央昆士兰大学毕业证成绩单pdf电子版制作修改#毕业文凭制作#回国入职#diploma#degree毕业文凭制作#回国入职#diploma#degree澳洲中央昆士兰大学毕业证成绩单pdf电子版制作修改#毕业文凭制作#回国入职#diploma#degree
毕业文凭制作#回国入职#diploma#degree澳洲中央昆士兰大学毕业证成绩单pdf电子版制作修改#毕业文凭制作#回国入职#diploma#degree
 
PKS-TGC-1084-630 - Stage 1 Proposal.pptx
PKS-TGC-1084-630 - Stage 1 Proposal.pptxPKS-TGC-1084-630 - Stage 1 Proposal.pptx
PKS-TGC-1084-630 - Stage 1 Proposal.pptx
 
RS 9000 Call In girls Dwarka Mor (DELHI)⇛9711147426🔝Delhi
RS 9000 Call In girls Dwarka Mor (DELHI)⇛9711147426🔝DelhiRS 9000 Call In girls Dwarka Mor (DELHI)⇛9711147426🔝Delhi
RS 9000 Call In girls Dwarka Mor (DELHI)⇛9711147426🔝Delhi
 
Indian Call Girls in Abu Dhabi O5286O24O8 Call Girls in Abu Dhabi By Independ...
Indian Call Girls in Abu Dhabi O5286O24O8 Call Girls in Abu Dhabi By Independ...Indian Call Girls in Abu Dhabi O5286O24O8 Call Girls in Abu Dhabi By Independ...
Indian Call Girls in Abu Dhabi O5286O24O8 Call Girls in Abu Dhabi By Independ...
 
Effects of Smartphone Addiction on the Academic Performances of Grades 9 to 1...
Effects of Smartphone Addiction on the Academic Performances of Grades 9 to 1...Effects of Smartphone Addiction on the Academic Performances of Grades 9 to 1...
Effects of Smartphone Addiction on the Academic Performances of Grades 9 to 1...
 
科罗拉多大学波尔得分校毕业证学位证成绩单-可办理
科罗拉多大学波尔得分校毕业证学位证成绩单-可办理科罗拉多大学波尔得分校毕业证学位证成绩单-可办理
科罗拉多大学波尔得分校毕业证学位证成绩单-可办理
 
Advanced Machine Learning for Business Professionals
Advanced Machine Learning for Business ProfessionalsAdvanced Machine Learning for Business Professionals
Advanced Machine Learning for Business Professionals
 
Machine learning classification ppt.ppt
Machine learning classification  ppt.pptMachine learning classification  ppt.ppt
Machine learning classification ppt.ppt
 
Statistics, Data Analysis, and Decision Modeling, 5th edition by James R. Eva...
Statistics, Data Analysis, and Decision Modeling, 5th edition by James R. Eva...Statistics, Data Analysis, and Decision Modeling, 5th edition by James R. Eva...
Statistics, Data Analysis, and Decision Modeling, 5th edition by James R. Eva...
 
Learn How Data Science Changes Our World
Learn How Data Science Changes Our WorldLearn How Data Science Changes Our World
Learn How Data Science Changes Our World
 
专业一比一美国俄亥俄大学毕业证成绩单pdf电子版制作修改
专业一比一美国俄亥俄大学毕业证成绩单pdf电子版制作修改专业一比一美国俄亥俄大学毕业证成绩单pdf电子版制作修改
专业一比一美国俄亥俄大学毕业证成绩单pdf电子版制作修改
 
RABBIT: A CLI tool for identifying bots based on their GitHub events.
RABBIT: A CLI tool for identifying bots based on their GitHub events.RABBIT: A CLI tool for identifying bots based on their GitHub events.
RABBIT: A CLI tool for identifying bots based on their GitHub events.
 
While-For-loop in python used in college
While-For-loop in python used in collegeWhile-For-loop in python used in college
While-For-loop in python used in college
 
Building on a FAIRly Strong Foundation to Connect Academic Research to Transl...
Building on a FAIRly Strong Foundation to Connect Academic Research to Transl...Building on a FAIRly Strong Foundation to Connect Academic Research to Transl...
Building on a FAIRly Strong Foundation to Connect Academic Research to Transl...
 
Call Us ➥97111√47426🤳Call Girls in Aerocity (Delhi NCR)
Call Us ➥97111√47426🤳Call Girls in Aerocity (Delhi NCR)Call Us ➥97111√47426🤳Call Girls in Aerocity (Delhi NCR)
Call Us ➥97111√47426🤳Call Girls in Aerocity (Delhi NCR)
 

How deep generative models can help quants reduce the risk of overfitting?

  • 1. How deep generative models can help quants reduce the risk of overfitting? Talk at QuantUniversity Autumn School 2020 Gautier Marti HKML Research 27 October 2020 Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 1 / 59
  • 2. Table of contents 1 Introduction 2 GANs explained GANs Milestones & Major Achievements How do GANs work? 3 Applications of GANs for Quants Generating Synthetic Datasets to Avoid Strategy Overfitting Generating Alternative Realistic Historical Paths for Risk Estimation 4 Current State of the Art and Limitations 5 Conclusion and Questions? Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 2 / 59
  • 3. Section 1 Introduction Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 3 / 59
  • 4. Generative Adversarial Network’s Definition & Goal (Deep) generative models are a broad class of models. For the sake of clarity, we will only consider Generative Adversarial Networks in this presentation. Definition: Generative Adversarial Network (GAN) A generative adversarial network (GAN) is a class of machine learning frame- works where two neural networks (usually known as the generator and the discriminator) contest with each other in a zero-sum game (i.e. one agent’s gain is another agent’s loss). Main purpose of a GAN: Generating realistic synthetic data Given a training set, this technique learns to generate new data with the same statistics as the training set. Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 4 / 59
  • 5. Introduction (3 min. video) https://www.youtube.com/watch?v=97B8tuHwLY0 Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 5 / 59
  • 6. Scope of the presentation In this talk, we will focus on doing a review of GANs success so far (mostly image generation), explaining how Generative Adversarial Networks (GANs) work. Then, we will present applications of these models for quant researchers / traders use cases. Finally, we shall briefly discuss the current limitations and challenges to be overcome for a broader adoption of these models in the industry. Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 6 / 59
  • 7. Section 2 GANs explained Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 7 / 59
  • 8. Subsection 1 GANs Milestones & Major Achievements Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 8 / 59
  • 9. GANs: A relatively new Deep Learning model (2014) Goodfellow, Ian, et al. “Generative adversarial nets.” Advances in neural information processing systems. 2014. Cited by 22354 papers as of 18 September 2020. Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 9 / 59
  • 10. Text to Photo-realistic Image Synthesis (2016) https://arxiv.org/pdf/1612.03242.pdf Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 10 / 59
  • 11. Deepfakes (2017) Ctrl Shift Face (YouTube channel) https://www.youtube.com/channel/UCKpH0CKltc73e4wh0_pgL3g You Won’t Believe What Obama Says In This Video! https://www.youtube.com/watch?v=cQ54GDm1eL0 Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 11 / 59
  • 12. Edmond de Belamy (2018) Created by 3 students First artwork created using Artificial Intelligence to be featured in a Christie’s auction Sold for USD 432,500 Signed minG maxD Ex [log(D(x))] + Ez [log(1 − D(G(z)))] In French, “bel ami” means “good fellow”, a pun-tribute to Ian Goodfellow, the creator of GANs Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 12 / 59
  • 13. StyleGAN (2019) & StyleGAN2 (2020) GANs can generate realistic fake human faces: Give it a try: https://thispersondoesnotexist.com/ NVIDIA StyleGAN paper: https://arxiv.org/pdf/1812.04948.pdf NVIDIA StyleGAN2 paper: https://arxiv.org/pdf/1912.04958.pdf Remark. In December 2019, Facebook took down a network of accounts with false identities, and mentioned that some of them had used profile pictures created with artificial intelligence. Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 13 / 59
  • 14. Speech2Face & Wav2Pix (2019) GANs conditioned on speech of a person can output a realistic face with its correct gender, ethnicity, and approximate age: Relevant papers: Wav2Pix: https://arxiv.org/pdf/1903.10195.pdf Speech2Face: https://arxiv.org/pdf/1905.09773.pdf Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 14 / 59
  • 15. GameGAN (2020) GameGAN is able to learn Pac-Man dynamics and produce a visually consistent simulation of the game: NVIDIA paper https://arxiv.org/pdf/2005.12126.pdf https://www.youtube.com/watch?v=4OzJUNsPx60 Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 15 / 59
  • 16. Subsection 2 How do GANs work? Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 16 / 59
  • 17. A GAN basic architecture Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 17 / 59
  • 18. GAN training – step by step tutorial Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 18 / 59
  • 19. GAN training – step by step tutorial Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 19 / 59
  • 20. GAN training – step by step tutorial Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 20 / 59
  • 21. GAN training – step by step tutorial Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 21 / 59
  • 22. GAN training – step by step tutorial Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 22 / 59
  • 23. GAN training – step by step tutorial Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 23 / 59
  • 24. GAN training – step by step tutorial Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 24 / 59
  • 25. GAN training – step by step tutorial Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 25 / 59
  • 26. GAN training – step by step tutorial Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 26 / 59
  • 27. GAN training – step by step tutorial Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 27 / 59
  • 28. GAN training – step by step tutorial Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 28 / 59
  • 29. GAN training – step by step tutorial Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 29 / 59
  • 30. GAN training – step by step tutorial Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 30 / 59
  • 31. GAN training – step by step tutorial Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 31 / 59
  • 32. GAN training – step by step tutorial Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 32 / 59
  • 33. GAN training – step by step tutorial Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 33 / 59
  • 34. GAN training – step by step tutorial Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 34 / 59
  • 35. GAN training – step by step tutorial Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 35 / 59
  • 36. GAN training – step by step tutorial Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 36 / 59
  • 37. GAN training – step by step tutorial Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 37 / 59
  • 38. GAN training – step by step tutorial Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 38 / 59
  • 39. GAN training – step by step tutorial Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 39 / 59
  • 40. Conditional Generative Adversarial Nets Seminal paper: https://arxiv.org/pdf/1411.1784.pdf Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 40 / 59
  • 41. Conditional Generative Adversarial Nets Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 41 / 59
  • 42. Section 3 Applications of GANs for Quants Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 42 / 59
  • 43. Subsection 1 Generating Synthetic Datasets to Avoid Strategy Overfitting Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 43 / 59
  • 44. Strategy Overfitting Scenario In September 2020, a naive or unscrupulous strat presents to his manager a new strategy: The strategy would have initiated a massive short in February 2020, and would have bet big on a rally starting late March 2020. The strat sells the strategy to his manager as able to pick-up early signals of market sell-offs and bounce-backs thanks to advanced machine learning and alternative data (obviously, what else?). The manager, excited, cannot wait but to deploy capital to it. How to fight against this industry-wide fallacy? Well-thought and carefully designed incentives schemes (e.g. reward a sound research framework rather than impressive in-sample backtests) Counterfactual thinking, realistic simulations of alternative paths Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 44 / 59
  • 45. Strategy Overfitting – Simulations of Alternative Paths (1) Some alternative data: Second-hand car market in Hong Kong A tabular dataset X Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 45 / 59
  • 46. Strategy Overfitting – Simulations of Alternative Paths (2) econ quant idea: When inventory builds up (more people are selling their cars than buying), then the HSI plummets over the next quarter. math quant idea: My black-box ML algorithm predicts that HSI decreases whenever two Ferraris are sold right after a Bugatti Veyron comes to market, based on the given dataset. Using GANs, we can sample new datasets ˆX ∼ X, or even ( ˆX, ˆy) ∼ (X, y), where y can be the HSI performance or any variable we aim at predicting (e.g. market+industry residualized returns of the respective carmakers). Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 46 / 59
  • 47. Strategy Overfitting – Simulations of Alternative Paths (3) Method 1: Strategy stability Use a GAN to learn the distribution (X, y) of the tabular data (cf. https://arxiv.org/pdf/1907.00503.pdf) From the calibrated GAN, generate ˆX(n) , ˆy(n) N n=1 synthetic datasets Backtest the strategy on the N datasets, and collect perf. metrics Analyze the perf. metrics; Discard the strategy if not stable. Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 47 / 59
  • 48. Strategy Overfitting – Simulations of Alternative Paths (4) Method 1bis: Find whether a strategy is robust and obvious given data As the strat-quant team manager, provide each one of the N quants with a GAN-generated copy of the original data. They have to work on it independently. Will they find the same strategy? With roughly the same parameters? If so, the strategy is robust, but also likely to be crowded given the data. Test it, and run it in production, on the original data. Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 48 / 59
  • 49. Strategy Overfitting – Simulations of Alternative Paths (5) In Koshiyama et al. (https://arxiv.org/pdf/1901.01751.pdf): Generate N synthetic datasets: Method 2: Strategy fine-tuning Find the parameters which maximize the average perf. over the N datasets Method 2bis: Strategy combination (ensemble models) Fit a model on each of the N datasets Predict using an ensemble of the N trained models GAN vs. Stationary Bootstrap (for generating the N synthetic datasets) Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 49 / 59
  • 50. Subsection 2 Generating Alternative Realistic Historical Paths for Risk Estimation Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 50 / 59
  • 51. Risk-based Portfolio Allocation (1) Scenario You want to systematically allocate capital to your strategies. Literature claims that such or such method works better than all others by providing a dubious backtest. When you horse race the various allocation methods, results are not stable across periods and universes. How can we conclude anything useful? Realistic simulations! Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 51 / 59
  • 52. CorrGAN: Sampling Realistic Financial Correlation Matrices Using Generative Adversarial Networks 6 stylized facts of empirical financial correlation matrices Marchenko-Pastur-like distribution, except for: a very large first eigenvalue, a couple of other large eigenvalues Perron-Frobenius property positive correlations Hierarchical structure of clusters Scale-free property of the degrees in MST Original paper: https://arxiv.org/pdf/ 1910.09504.pdf Implemented in the MlFinlab python package: https://hudsonthames. org/mlfinlab/ Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 52 / 59
  • 53. Risk-based Portfolio Allocation (2) Method 3: Testing Portfolio Allocation on Alternative Historical Paths Use a GAN to generate realistic synthetic correlation matrices (cf. https://arxiv.org/pdf/1910.09504.pdf) Generate time series verifying the correlation structure Estimate portfolio weights (in-sample), measure out-of-sample risk Analyze performance of the portfolio allocation methods Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 53 / 59
  • 54. Risk-based Portfolio Allocation (3) Question: Can we predict when a method (say the Hierarchical Risk Parity (HRP)) outperforms another one (say naive risk parity)? And why? Method 4: Understanding the outperformance of a method over another Extract features from the underlying correlation matrix Fit a ML model features → outperformance (using train, validation, test datasets) Verify if there is any good predictability Explain the predictions as a function of features Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 54 / 59
  • 55. Risk-based Portfolio Allocation (4) Concrete example: HRP vs. naive risk parity High values for the cophenetic correlation coefficient are characteristic of a strong hierarchical structure. Thus, HRP outperforms naive risk parity when the underlying DGP has a strong hierarchical structure (nested clusters). cf. Jochen Papenbrock recent work for similar applications of Explainable AI (XAI) in finance (https://firamis.de/) Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 55 / 59
  • 56. Section 4 Current State of the Art and Limitations Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 56 / 59
  • 57. Current State of the Art and Limitations This discussion is by nature technical, but we can highlight the following limitations which are relevant for quants: GANs can generate realistic tabular datasets (2019), but models trained on synthetic data only were shown inferior to the ones trained on the original data We know how to GAN-generate realistic synthetic financial time series (e.g. S&P 500 returns); We know how to GAN-generate realistic synthetic financial correlation matrices (of the S&P 500 constituents, for example); However, we do not know yet how to GAN-generate realistic multivariate financial time series, i.e. verifying both the time series and the cross-sectional stylized facts Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 57 / 59
  • 58. Section 5 Conclusion and Questions? Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 58 / 59
  • 59. Conclusion and Questions? Deep generative models and GANs in particular are an exciting new technology. In finance, they are actively researched in a few places but results are not widely advertised. We have to rely on top tech companies and academic labs to drive the fundamental understanding and improvements of these models. contact@hkml-research.com Gautier Marti (HKML Research) Applications of GANs for Quants 27 October 2020 59 / 59