Quant investing in India - testing a stock score on the Nifty 50
Quantooo Lab ยท 2026-10-22
Most quant research you will find online uses US data. We wanted to know whether the same method behaves differently in India, so we ran it on the Nifty 50 with Indian data sources and the same safeguards.
Can quant methods be used for Indian stocks?
Yes. The method does not care which market it runs on, as long as the data is point-in-time: each company result, price and economic figure must be used only from the moment it became public. India makes one part of this easier than most markets, because NSE filings carry an exact broadcast time, down to the minute.
What data did we use for the Nifty 50?
- Prices for the 50 index members and the Nifty 50 index.
- Company results from NSE filings (XBRL and integrated filings), used from their exact broadcast time, with units normalised across formats. About four years of history.
- Market backdrop from India VIX, the share of stocks in uptrends, and Indian interest rates.
- No news tone: we found no free, licensed headline feed per Indian company, so India scores average four parts instead of five, rather than using made-up or unlicensed data.
How was the India score tested?
Exactly like the US version: development data from 2020, walk-forward evaluation, random controls, every weighting counted, and the final twelve months sealed until the rules were frozen. The score was frozen on 20 October 2026 with equal weights, because no alternative weighting beat them significantly.
What did the test show?
On the sealed holdout, the India score's rank correlation with next-month returns relative to the Nifty 50 was +0.009, with a t-statistic of 0.21. That is no skill: indistinguishable from random rankings. It matches the US result, where the score also showed no reliable ability to rank stocks. The full report is on the India research page.
One warning applies to India in particular. Our simulated India history uses today's Nifty 50 members back to 2020. Companies usually join the index because they grew, so every simulated India portfolio, including the equal-weight benchmark, looks better than it could have in real time. That is survivorship bias, and we flag it on the India portfolios page.
What does this mean for quant investing in India?
That large, well-followed Indian stocks are as hard to rank with simple public signals as large US stocks. It does not mean quant methods are useless in India. It means a claim of an edge needs the same strict testing in India as anywhere else, and many claims you see online would not survive it.
How can you follow the India results live?
The India scores are published and fingerprinted every trading day, and a one-year live challenge started on 21 October 2026: two rule-based portfolios against the Nifty 50 and a random benchmark. Follow it on the India portfolios page, or start from the India home page.
Open questions for students and researchers
Questions without a single right answer. Pick one and share your reasoning in the discussion below.
- NSE broadcast timestamps make point-in-time fundamentals easier in India. What other market-specific data quirks would a researcher need to handle?
- With 50 stocks, how much statistical power does a ranking test have compared with 100 or 500 stocks? What does that imply for our India result?
- India has no free per-company news feed in our setup. How could you build a news-tone signal for Indian stocks legally and point-in-time?
Debate: three perspectives
An illustrative debate written by the author to lay out the main viewpoints. The three voices are roles, not real people. Your view is welcome in the discussion below.
The Sceptic: Indian large caps are dominated by institutions and heavily researched. Expecting simple public signals to work there is unrealistic.
The Practitioner: India's mid and small caps are a different market, with more dispersion and less coverage. That is where a careful researcher should look next.
The Academic: Cross-market replication is valuable precisely because conditions differ. A method that fails consistently in two markets tells you more than a success in one.
Which view is closest to yours, and what evidence would change your mind?