The NIFTY 50 Inclusion Trade: Does It Still Work? | The Long & The Short Ep. 40

ELI5/TLDR

When a stock joins the Nifty 50, every index fund tracking the index is forced to buy it. The obvious idea: get in before they do and ride the wave. Zerodha tested every Nifty 50 inclusion from 2015 to 2025 (33 events) and found the edge is real but tiny — it shows up cleanly only in a two-week window right after the stock enters, then vanishes. The bigger-looking opportunity is to guess which stocks will be added 6-12 months early, but that turns out to be partly a statistical illusion and partly a crowded, hard game.

The Full Story

The mechanical reason the trade should exist

A stock index like the Nifty 50 is just a list of 50 companies. Funds that promise to “track” the index must hold exactly those 50, in the right proportions. So when the National Stock Exchange announces a new member, every passive fund and ETF is contractually obliged to go buy it — and to do so by a known deadline.

Every passive fund or ETF that tracks the Nifty 50 has to buy that stock… and they have to do it by a specific date. And all of that is known and hence there will be a predictable burst of buying.

Predictable forced buying is catnip for traders. If you know a wall of demand is coming, you buy ahead of it. This is the “inclusion trade,” and people have studied the American version (the S&P 500) for forty years.

What happened to the same trade in the US

It worked, then died. The bonus return from being added to the S&P 500 fell from about 7% in the 1990s to under 1% in the last decade. Once an edge is well-known, everyone front-runs it, and the front-running competes the profit away. Markets eat their own free lunches.

How a stock actually gets into the Nifty 50

NSE reviews the membership twice a year (March and September), using six months of data ending January 31 and July 31, with four weeks’ notice before changes take effect. To qualify, a stock must be in the futures-and-options segment, be liquid enough (an “impact cost” — how much a large order moves the price — of 0.5% or less), and be big enough. “Big” is measured by free-float market cap: only shares actually available to the public count, not the chunk locked up with founders. The 50 largest eligible names from the broader Nifty 500 make the cut. One wrinkle: it’s not pure arithmetic — an NSE committee gets the final say, so human judgment is in the loop.

The forward test: buy after inclusion

The study anchored everything to the effective date (T0 = the day the stock officially joins), then measured returns over six windows ahead. The standout result was two weeks:

A win rate of 78.8%… The mean return is 2.21% and the median is at 3.07%. When mean and median are both positive and close to each other, it tells you the result isn’t being distorted by a few big outliers.

Nearly four in five inclusions were up two weeks later, by similar amounts. The one-week window was a coin flip (54.5%), and everything past two weeks decayed: by six months the typical (median) stock was actually down 2.28%, and the spread of outcomes ballooned (standard deviation went from 5.6% at two weeks to nearly 30% at twelve months). The interpretation is clean: the forced buying gets absorbed over roughly two weeks, then it’s gone and you’re just holding an ordinary stock.

Traded mechanically — buy each inclusion, hold two weeks — ₹1 lakh in 2015 became ₹1.22 lakh by 2025, a 22.25% total return with a 14.3% worst drawdown. That sounds weak until you notice your money is invested only a few weeks a year; it sits idle the rest of the time. The return-to-pain ratio of 1.56 is decent for a simple event-driven strategy. The host’s verdict: real edge, modest size.

The backward test, and the trap inside it

Then he flipped the question: how do these stocks do before they’re added? The numbers look spectacular — up 58% on average in the 12 months prior, with a 93.9% win rate (every single stock that got in had risen the year before). Tempting to think: just predict the next entrant early and retire to Goa.

Except those numbers are almost guaranteed to look great, for a sneaky reason:

A stock gets added because it has grown large enough, which means the price has to go up for it to become eligible… The backward return and the selection criteria are essentially measuring the same thing. This is called sample truncation bias.

You only ever see the stocks that cleared the size bar — and clearing a market-cap bar requires the price to have gone up. So “stocks rise before inclusion” isn’t a discovery; it’s baked into how you chose the sample. You never get to see the stocks that nearly made it and didn’t.

So is there a real pre-inclusion edge?

The honest version of the question is: can you predict which stocks will be added, before the announcement, and is there profit left after everyone else tries the same thing? The methodology is public and rule-based, so the Nifty Next 50 acts as a “waiting room” — the names closing the free-float gap on the bottom of the Nifty 50 are the natural candidates. But three problems: you can identify the pool, not the exact stock or cycle (names hover on the boundary for years); the committee can override the rules; and you’re competing with everyone else reading the same public rankings, which prices the edge in early. He notes one Indian asset manager has built a large-cap fund around exactly this idea (he leaves the name as a comment-section teaser).

Key Takeaways

Index inclusion forces buying. Passive funds must hold every constituent at its index weight, so a new entrant triggers mandatory, deadline-bound buying — a predictable demand shock.
The US version of this trade decayed from ~7% (1990s) to <1% (2010s) as it got discovered and front-run. Edges erode once known.
For the Nifty 50 (33 inclusions, 2015-2025), the clean edge is a two-week window after the effective date: 78.8% win rate, ~2.2% mean / 3.1% median return.
Mean ≈ median is a quality check — when they agree, the result isn’t being driven by a few freak outliers.
Beyond two weeks the edge disappears. By six months the median inclusion was down ~2.3%, and outcome dispersion roughly quintupled. Forced buying is absorbed in ~2 weeks; after that it’s just a normal holding.
Nifty 50 rebalances twice a year (effective after March and September F&O expiry), using six-month data windows ending Jan 31 and Jul 31, with four weeks’ notice.
Eligibility filters: in the F&O segment, impact cost ≤ 0.5% (liquidity), and top-50 by free-float market cap within the Nifty 500. A committee retains final discretion.
Sample truncation bias is the lesson buried in the “stocks rise 58% before inclusion” stat: selection is based on size, size requires price rises, so the strong pre-returns are a property of the sample, not a tradable signal.
The Nifty Next 50 is the “waiting room” — a structured candidate list for the next inclusion cycle, since the index is built from the top 50 of the Nifty 500 by free float.
The strategy backtest used raw price returns, not benchmark-adjusted. To truly isolate the inclusion effect you’d compute excess return over the Nifty 50 itself — the host flags this as the missing rigor.

Claude’s Take

This is a careful, intellectually honest piece of work, which is rarer than it should be in retail-finance content. The standout moment is the author catching his own backward-returns test and explaining the truncation bias rather than selling the 58%-and-retire fantasy. That self-correction is the most valuable thing in the video — it’s a transferable lesson about any backtest where the selection criterion and the measured variable secretly overlap.

The honesty cuts both ways, though. The headline finding is a thin one: a ~2% median two-week pop that compounds to 22% over a decade because your capital is parked most of the year. He’s upfront that this isn’t spectacular, and he flags the biggest weakness himself — the returns aren’t benchmark-adjusted, so we don’t actually know whether inclusions beat the index or just rose with it. Until that excess-return test is run, “the edge is real” is more of a “the stocks went up” than a proven anomaly. Thirty-three events is also a small sample to hang win-rate percentages on.

Knocking a point off for the cliffhanger marketing tics (guess-the-fund-house, part-two-next-time) and for stopping right where the analysis gets interesting. But the reasoning is sound, the mechanics are explained well, and it respects the viewer’s intelligence. A 7: genuinely useful, methodologically self-aware, just not finished.