Take today’s Nifty 500 constituent list, pull 15 years of price history for those names, and backtest a strategy on it. The result will look better than anything you could have actually earned, and the reason is what is missing from the list.
YESBANK, DHFL, RCOM, JETAIRWAYS and UNITECH were all large, widely held, index-eligible companies at some point in the last 15 years. Each then collapsed, was restructured, or left the exchange. A universe assembled from today’s list never buys any of them. A strategy running in 2016 would have bought several.
Why it inflates results
The bias works in two directions at once. Companies that failed are removed from the sample, so the average return of what remains is higher than the average return of what was actually investable. And the survivors are, by definition, the ones that made it through, so the drawdowns in your sample are the survivable ones.
The effect is largest exactly where retail strategies concentrate.
| Screen type | Why survivorship hits harder |
|---|---|
| Value and low PE | Distressed companies screen as cheap. A survivor-only universe removes the ones whose cheapness was correct. |
| High dividend yield | A collapsing price inflates trailing yield. Delisted names are removed before you ever see them. |
| Small and micro cap | The delisting and suspension rate is highest here, so the deletion is largest. |
| Mutual fund tables | Poorly performing schemes get merged into better ones, and their history goes with them. |
The fix: as-of universes
The correct construction is to rebuild the universe as it stood on each rebalance date, including securities that no longer trade. Historical universes on saral.money retain delisted stocks, so a backtest can hold a name that later stopped existing and take the loss the way you would have.
Caveats
Keeping delisted names is necessary but not sufficient. You also need a defensible assumption about what happens to the position when trading stops. A suspended stock that eventually pays nothing is a different result from one marked at its last traded price, and the gap between those two assumptions is entirely your judgement.
Check how any backtest you rely on, including your own, treats a position that cannot be exited.