Prescriptive AI·Decision systems
The Moment AI Stops Predicting and Starts Choosing
Where the real value of AI begins: when a model stops handing you a probability and starts making a decision you can actually act on — and defend.
Read on Medium →Bandits·Reinforcement learning
Thompson Sampling: How Recommender Systems Learn to Bet on What You'll Like
A hands-on look at Thompson Sampling — how recommenders balance exploring new options against exploiting what already works, and learn from every outcome.
Read on Medium →Causal ML·Quant finance
Using Market Stress as a Treatment: Ranking S&P 500 Stocks with Causal ML
Treating observable market stress as the treatment and forward market-adjusted alpha as the outcome — a leakage-aware causal-ML approach to ranking S&P 500 stocks.
Read on Medium →More on Medium
Follow along as I write up the ideas behind each system.