Every listed Indian company’s directors, key management and their immediate relatives must disclose their trades in the company’s own shares. The filings are public, timestamped, free, and arrive within two trading days. They are also asymmetric in a way that most people reading them get backwards.
The asymmetry
Ask why a senior executive might sell shares in their own company.
They are buying a house. They are paying for a child’s education. They are diversifying a portfolio that is 80% one stock. They have an ESOP tranche vesting and a tax bill attached. They are getting divorced. Somewhere down the list: they think the stock is expensive.
Now ask why the same executive might buy shares on the open market, with post-tax money, in a company where they already have most of their financial and professional life invested — and where the purchase must be publicly disclosed and creates a six-month window in which they cannot profitably reverse it.
The list is much shorter. Essentially: they think it is cheap.
That asymmetry is the whole signal. Sales are noisy because the motive space is large. Purchases are informative because the motive space is nearly a point.
The evidence
Lakonishok and Lee’s 2001 study of US insider filings found that companies with heavy insider buying outperformed those with heavy insider selling by roughly 4.8% over the following twelve months, with the effect concentrated in smaller companies. Their conclusion on the sell side was that it carried little predictive content — consistent with the motive argument above.
Jeng, Metrick and Zeckhauser reached a similar conclusion with a different method: insider purchase portfolios earned abnormal returns of around 6% annually, while sale portfolios earned close to nothing abnormal.
The size concentration matters for how you use it. The effect is largest where analyst coverage is thin, which is what you would expect if the mechanism is an information gap rather than a risk premium. In a heavily covered large cap, the insider knows less relative to the market than in a midcap that three analysts follow.
How the Indian disclosure works
SEBI’s Prohibition of Insider Trading Regulations, 2015, at Regulation 7(2)(a), require designated persons and their immediate relatives to disclose trades to the company within two trading days of any transaction, or series of transactions, exceeding ₹10 lakh in value over a calendar quarter. The company then files with the exchanges, which publish the disclosure.
Two features make this usable that would not be obvious:
The window is short. Two trading days from transaction to company, then onward to the exchange. Compared to a quarterly shareholding pattern that can be four months stale, this is nearly real-time.
The contra-trade restriction has teeth. A designated person who buys cannot sell within six months, and vice versa. That materially raises the cost of a cosmetic purchase intended to signal confidence — you are locked in.
The exclusions matter too. ESOP exercises, gifts and inter-se transfers between promoters are disclosed but are not open-market purchases, and treating them as buy signals will produce noise. An open-market purchase is a different act from receiving stock you were granted.
Building it as an event strategy
This is the template that cannot be built with a calendar rebalance, and understanding why generalises to a whole class of signals.
Most fields carry a value on every trading day — a close price, a market cap, a moving average. An insider disclosure is not like that. It exists on the day it was filed and does not exist on any other day. Sample the universe on the last Friday of the month and you will see disclosures filed on that Friday and nothing else, missing the other twenty sessions entirely.
The fix is a rule that scans a window rather than a point: on each rebalance, look back N days and keep securities that had a qualifying event somewhere in that range, exposing how long ago it happened for downstream ranking.
Window length is a real trade-off. Too short and the two-day filing lag consumes most of it. Too long and it stops being an event signal and becomes a stale tilt — a purchase from four months ago tells you what someone thought about a price that no longer exists. Forty-five days with a weekly rebalance is a reasonable middle: long enough to catch the filings, short enough that positions turn over as the information ages.
What will go wrong
Small purchases are noise. A director buying ₹2 lakh of a ₹15,000 crore company has told you nothing meaningful. Ranking on the rupee value of the purchase, rather than treating every disclosure as equal, concentrates the portfolio in the ones where the insider actually committed something.
The signal concentrates in illiquid names. The size effect that makes the strategy work also points it at companies where your fills are worse and the exchange is more likely to have surveillance restrictions in place. A turnover floor and a surveillance exclusion are not optional garnish here — they are what keeps the backtest connected to executable trades.
Clustering matters more than any single filing. One director buying is one person’s opinion. Four insiders buying in the same fortnight is a different kind of evidence. A simple value-ranked screen treats these the same; if you extend this template, distinguishing them is the highest-value change to make.
The base rate is unforgiving. Insiders are not right about the short term, and this is not a timing signal. Lakonishok and Lee’s outperformance figure is over twelve months. A strategy holding for four weeks is running a much noisier version of the same bet.
Try it
The Insider Buying Follow-Through template scans weekly for open-market insider purchases disclosed in the previous 45 days, filters for rupee turnover and excludes stocks under GSM surveillance, and holds the twenty largest purchases equally weighted.
The comparison that teaches most: run it with the window at 15 days, then at 90, and watch turnover and returns move in opposite directions. Somewhere between the two is where the filing lag stops eating the signal and staleness has not yet set in — and finding that point yourself is more instructive than accepting the default.
Further reading
- Promoter pledging: the risk that turns a decline into a collapse reads a different mandatory disclosure about the same people
- Bulk deals vs block deals vs shareholding patterns sorts out which filing covers which kind of transaction, and on what lag
Glossary: event study, abnormal return, SAST disclosure, point-in-time data.