Reinforcement Learning for Finance with Dr. Yves J. Hilpisch

In this episode of ODSC’s Ai X Podcast, Dr. Yves J. Hilpisch, founder and CEO of The Python Quants (http://tpq.io), and founder and CEO of The AI Machine (http://aimachine.io), joins us to discuss reinforcement learning for finance. Yves is also the author of the book "Reinforcement Learning for Finance” and has a diploma in Business Administration and a Ph.D. in Mathematical Finance. Yves is also an adjunct professor for Computational Finance at the Miami Herbert Business School. Show Topics: Overview of The Python Quants The speaker's new book, “Reinforcement Learning for Finance” and why the focus on reinforcement learning Dynamic time problems Markov decision processes Key types of reinforcement learning models Deep Q-Learning (DQL) and how it relates to Q-Learning How deep Q-Learning be applied to financial contexts, such as trading strategies or portfolio management Issues associated with using static historical time series data for training DQL agents in finance End-of-day data vs tick data Adding white noise to historical time series data to improve the training of DQL agents Key differences between the noisy time series data and the simulated time series data approaches Generative Adversarial Networks (GANs) utility for generating synthetic financial time series data GANs’ advantages over traditional Monte Carlo simulations in generating financial data How to check the quality of synthetic data The role of Kolmogorov-Smirnov (KS) test in evaluating the synthetic data generated by GANs How the chapter compare the effectiveness of GAN-generated data to real financial data The primary goal of the trading agent The role of buy bots The role of agentic AI Topic analysis and sentiment analysis Overview of the “Researchers Find AI Model Outperforms Human Stock Forecasters ‘Financial Statement Analysis with Large Language Models’” paper Yves’ session at ODSC Europe SHOW NOTES Monte Carlo Simulation in Finance: https://www.investopedia.com/articles/investing/112514/monte-carlo-simulation-basics.asp Python Quants: https://home.tpq.io/ Certificate in Python for Finance: https://home.tpq.io/certificate/ Markov decision process: https://en.wikipedia.org/wiki/Markov_decision_process Black Scholes model: https://www.investopedia.com/terms/b/blackscholes.asp Deep Q Learning: https://www.tensorflow.org/agents/tutorials/0_intro_rl Backtesting: https://www.investopedia.com/terms/b/backtesting.asp Model collapse: https://en.wikipedia.org/wiki/Model_collapse GANS: https://en.wikipedia.org/wiki/Generative_adversarial_network Black Swan Events: https://www.investopedia.com/terms/b/blackswan.asp Kamograve Smirnov test: https://en.wikipedia.org/wiki/Kolmogorov%E2%80%93Smirnov_test Delta Hedging: https://www.investopedia.com/terms/d/deltahedging.asp Hedging strategies: https://www.investopedia.com/trading/hedging-beginners-guide/ Option Replication: https://www.cfainstitute.org/en/membership/professional-development/refresher-readings/option-replication-put-call-parity Geometric Brownian motion: https://en.wikipedia.org/wiki/Geometric_Brownian_motion Jump Diffusion: https://en.wikipedia.org/wiki/Jump_diffusion Heston model: https://en.wikipedia.org/wiki/Heston_model Bates Mode: https://en.wikipedia.org/wiki/Stochastic_volatility_jump Gain Fallacy (A loss of 70% requires a 300% gain to break even): https://www.rgbcapitalgroup.com/preserving-capital Prime Brokers: https://www.investopedia.com/terms/p/primebrokerage.asp Algorithmic trading: https://www.investopedia.com/articles/active-trading/101014/basics-algorithmic-trading-concepts-and-examples.asp Financial statement analysis, with large language models: https://arxiv.org/pdf/2407.17866