On 2 February 2022 Meta lost $230bn in a day. I gave a language model one careful prompt and ran it 360 times on that print. It called the direction right 60 times out of 60 — and produced 13 different trades. Here are the nine pitfalls, and the skill that fixes them. The post One Prompt, One Stock, 360 Runs: How an LLM Gets a Trade Right and Still Can t Be Trusted With It appeared first on Jiri…
TL;DR: For a year the standard critique of LLM-designed strategies was the model can t actually do the math — the backtest is hallucinated. That critique is now dead. ChatGPT, Claude and Gemini all execute real, sandboxed Python with pandas and Matplotlib (OpenAI, Anthropic, Google). The code runs. The Sharpe is computed, not written. And that... The post The LLM Ran the Backtest. That s Exactly…
TL;DR — Robo stock-picking is forecast to grow roughly 600% by 2029, and a ChatGPT-picked basket is genuinely beating the UK s most popular funds. But most AI beats the market backtests are contaminated by look-ahead bias that lives inside the model s weights, not in your dataset — and here is the paradox almost nobody has... The post Your LLM Doesn t Pick Stocks — It Remembers Them appeared first…
I took the viral "make medallion fund" prompt literally and backtested it. $1 became $0.02. Here's where LLM alpha actually lives. The post Make Medallion Fund : I Ran the Viral One-Prompt Strategy. $1 Became $0.02. appeared first on Jiri Pik .
TL;DR: The best AI coding assistant for finance in 2026 isn t a single tool — it s a two-layer stack. I run GitHub Copilot for the inner loop (inline completions, Visual Studio s .NET firepower, GitHub-native PR approvals) and Claude Opus 4.8 for the outer loop (multi-file reasoning, refactors, debugging the math). Copilot has the better hands... The post The Quant s Guide to AI Coding Assistants:…
I cut my portfolio's max drawdown from −34% to −24% with reinforcement learning — without coding a single sell rule. Full PyTorch code and charts inside. The post How Reinforcement Learning Cut My Portfolio Drawdown From −34% to −24% — Without Me Telling It How appeared first on Jiri Pik .
On 5 Feb 2018, XIV lost 96% in 50 minutes. A LASSO stop from my book would have exited hours earlier. Full backtest in C# on QuantConnect. The post Volmageddon & The 90-Second Rule: The $0.01 Stop That Would Have Saved $3 Billion appeared first on Jiri Pik .
Modern traders need a chest of strategies, not a single play. I break down how CNNs are transforming technical pattern detection — with 97% accuracy and a 1.8 Sharpe ratio from our head-and-shoulders overlay. The post Every Trader Needs a Strategy Chest — AI Is How You Fill It appeared first on Jiri Pik .
I'm guest-lecturing via Zoom at Boston University on AI trading with my co-authors — here's what we're covering and why it matters. The post Guest Lecture at Boston University — Teaching AI Trading from Singapore via Zoom appeared first on Jiri Pik .
I coded Charlie Munger's quality investing philosophy in QuantConnect C# — 10% CAGR over 11 years. Here's what worked and what didn't. The post Charlie Munger s Quality Investing — Coded in QuantConnect C# appeared first on Jiri Pik .