Many models. One book.
IndiQuant is a hedge fund in the making, trading Indian equities. Independent researchers build models on anonymised NSE data. We score every submission against what the market did, weight the models that hold up by their record, and trade the combined signal as one risk-managed book.
Rounds, scoring and the meta-model run every NSE session. Paper trading of the book starts next, and live capital follows within the next few months.
- rounds, each NSE session
- Daily
- equities only
- NSE
- round window
- 19:00–09:00 IST
- prediction target
- 20-session
- portfolio rebalance
- Weekly
Recognition
Ranked #2 Top Company in Research by F6S, August 2026.
How the fund works
Only skill gets scored.
Every NSE session, contributors send a prediction for each name in the round. Once the 20-session outcome is in, each submission gets a score. The meta-model weights contributors by those scores, and the portfolio layer turns the result into a weekly, cost-aware book.
Backtests don't count here. A model's rank and its weight both come from out-of-sample results.
- Scored on
- Realised returns
- Weighted by
- Track record
- Traded as
- One weekly book
The thesis
Why build a fund this way.
Breadth
Many independent models.
One research desk tends to find the same kind of signal again and again. Many people working independently on the same data find different ones, and the meta-model uses that spread.
Evidence
Weight is earned.
A model's say in the book follows its scored, out-of-sample record. Backtests and job titles count for nothing.
India
One market, studied closely.
NSE equities on a point-in-time universe, with Indian trading costs priced into the portfolio from the start.
For contributors
From first model to a scored track record.
Free to join. No fee, no deposit. You bring the model and the compute. We supply the data, the scoring and a place in the meta-model if your record earns it.
01
Discover
Read the rules and download the round's dataset.
02
Build
Engineer features, train, iterate. Any method is fine.
03
Submit
Send one prediction per name before 09:00 IST.
04
Improve
Scores come back once the 20-session outcome is known.
05
Contribute
Models with a strong record carry more weight in the meta-model.
06
Grow
Your record builds round by round, under a public handle.
The platform
Where contributors work.
This site describes IndiQuant. Datasets, models and submissions all live on the platform.
01
Download the dataset
Anonymised, cross-sectional market data. No tickers, no company names. Only features and a target.
02
Build your model
Use whatever tools and methods you like, on your own machine. We only see the predictions.
03
Submit predictions
Submit a score per name before the round locks at 09:00 IST. Scoring runs against realised market outcomes.
Contributor rankings
The board fills as rounds resolve.
First scores expected late October 2026
No model is ranked yet. Rounds began on 21 September 2026 and each is scored about 21 trading sessions after it opens; a model joins the board after 20 resolved rounds, so the first ranked rows arrive around late November 2026.
Become a contributorThe stack
From anonymised data to a risk-managed book.
Six stages, with a signed hand-off between each, and all six run every session.
Anonymised data
Contributors get features and a target. No tickers, no company names.
Contributor models
Up to 3 models each, any method, 10 submission attempts per round.
Scoring
Correlation with the realised 20-session return, reported with a standard error.
Meta-model
Signals weighted by each model's scored record. Near-duplicates are discounted.
Portfolio
Rebalanced weekly by an optimiser that prices trading costs and holds sector, liquidity and turnover limits.
Execution
Runs in a separate, locked-down environment with its own credentials. Live order routing stays off until it is deliberately switched on.
Vision
An Indian equities fund built from the best models we can find, wherever they come from.
- Now
- The full pipeline, every session
- Next
- Live capital, within the next few months
- Then
- A fund for eligible investors
Join the platform
Because intuition is not a strategy.
If you can build a model that scores, there's room for it in the book. Investors and partners who want to follow our progress can write to us.
A company-funded research-grant programme is planned; it starts only after counsel's opinion, the research gate and the contributor agreement are in place, and it is discretionary. No grant has been paid.