Project H
Public research log

Project H. Where AI meets Indian F&O.

A research engine that reads machine learning, market structure and the news cycle together — and holds every idea to an honest bar before it trades. This is the public log: the thinking, not the positions.

AI-drivenNews-awareIndian F&OEvent-driven
How it thinks

A disciplined stack, held to an honest bar.

The edge isn't any single model — it's the discipline of combining many independent signals, refusing to trust what can't be measured, and letting risk lead. The specifics stay private; the principles don't.

Many weak signals, one view

No single model gets to decide. Independent, loosely-correlated signals are combined into one conviction — diversification at the level of ideas, not just names.

Honest validation

Nothing is trusted until it survives out-of-sample, corrected for how many things were tried. A good-looking backtest is treated as a hypothesis, not a result.

Risk before return

Sizing, stop discipline and defined-risk structures come first. The P&L is simply what's left after the risk has been respected.

News as signal, not noise

The news cycle is read for what actually moves price — filtered for materiality and novelty, and attributed to the right name rather than the loudest headline.

Featured

Latest entry

Cover image for The AI/ML approach to options selection — why we picked ranking over classification
News-led·

The AI/ML approach to options selection — why we picked ranking over classification

Picking the right machine-learning objective for short-horizon options trading is non-obvious. Classification asks 'will this stock go up?' — the wrong question. Ranking asks 'which of these 146 stocks will outperform the others?' — the right one. Why the distinction matters and how it changes everything downstream.

Harish Subramanian
Harish Subramanian