FEATURED / ML
XGBoost for 15-minute NIFTY
An accessible research framework for features, time order, walk-forward validation and costs.
Public work
Show the data. Show the limits.
Selected experiments in NIFTY/options ML and local AI systems. The recurring themes are real data, leakage control, walk-forward validation and a refusal to turn a backtest into a promise.
Read the field note
METHOD / RISK
Data leakage in trading AI
Why future-looking features silently make a model look smarter than it is, and how to audit the pipeline.
BUILD / LOCAL AI
Local RAG chatbot
A practical walkthrough for a grounded trading-research chatbot using Ollama and Termux.
Research shelf
More from the notebook.
Start with a question, then inspect the data and the validation design before trusting the result.
Options education
NSE option chain, OI, PCR and Greeks
Practical explanations of open interest, put-call ratio and Greeks for NIFTY learners.
Market behaviour
IV crush in weekly expiry
Why implied-volatility collapse can change an option buyer's outcome even when direction looks right.
Project map
Code, dashboards and research systems
Explore the featured repositories behind the public options research and AI trading work.