By Shakti Tiwari · Nifty Options Trader & XGBoost Expert
This is the Options AI research vertical — focused entirely on Nifty 50 and Bank Nifty derivative models. It is a separate, topically-focused property from the main hub, built so Indian index traders get deep, repeatable method instead of tips.
The method is the same five-layer pipeline documented across Shakti Tiwari's research: Data Engine → Feature Engineering → Predictor (XGBoost/LightGBM) → Risk Filter → Executor. Here we go deeper on index-specific behaviour.
Nifty 50 and Bank Nifty are not S&P 500. They have different expiry mechanics (weekly expiry, Thursday settlement), different OI concentration, and a retail-heavy participant base that creates repeatable sentiment patterns in the Option Chain. A model trained on US equities transfers poorly. Index-specific features matter.
Both XGBoost and LightGBM work. For index data (larger row counts, more strikes), LightGBM's histogram splits give faster iteration. The predictor outputs P(direction | features) per strike; the filter then decides action.
import lightgbm as lgb
model = lgb.LGBMClassifier(n_estimators=400, learning_rate=0.05,
max_depth=5, subsample=0.85, colsample_bytree=0.8)
model.fit(X_train, y_train)Bank Nifty spot 52,000, ATM 52,000, IV 17%, PCR 0.92, India VIX 14, DTE 6, OI buildup +ve at 52,200 CE. Features: max_pain_distance 0.2%, vix_z −0.3. Predictor P(up)=0.64. Filter: band ✅, DTE ✅, VIX ✅, max_pain 0.2% ❌ (too close to pin) → SIZE REDUCED, not blocked. This nuance is what separates a framework from a signal.
🌐 optiontradingwithai.in · 📝 Dev.to · 🐦 X · 📘 Book
Educational only — not investment advice. SEBI-registered research rules apply; verify everything before acting.
← Back to Main Hub