Trading the untradable index
Backtesting fills on the NIFTY spot index, a number you cannot buy or sell, lets a strategy trade at prices no real instrument ever offered.
The trap
The NIFTY 50 index is a calculated number, not something you can trade. When a backtest buys and sells “NIFTY” at the index’s own 1-min or 5-min high, low and close, it assumes fills at prices that may never have existed on any tradeable instrument. Real trades happen in futures, options or a synthetic future (a long call plus a short put at the same strike, which behaves like a future).
How it fools you
The index is a weighted average of 50 stocks that do not all tick at the same instant. Within a bar, that lag between constituents can print a high or low that no tradeable contract ever reached. Futures, which trade as a single instrument, do not show those extremes.
Any rule that triggers when price touches a level inherits these phantom prints. A breakout rule enters on a high that was never there. A tight target gets “hit” that was never really available. The result is edge that only exists on paper.
A real example
We re-ran three NIFTY intraday strategies over exactly the same dates, once on the index and once on futures, the prices you could actually have traded. A synthetic future tracks the near-month future closely and costs less to trade.
Opening range breakout (full post): go with the first break of the first 30 minutes’ range, stop and target at the same distance. October 2022 to July 2026, the best setting from our original sweep, entries at the range level.
| Traded on | Gross per trade | Net per trade | Total, net |
|---|---|---|---|
| NIFTY index (1.5 pts cost) | +3.4 pts | +1.9 pts | +1,379 pts |
| NIFTY futures (2 pts cost) | +1.4 pts | −0.6 pts | −435 pts |
Same rules, same days: the index more than doubled the gross edge, and a small profit became a small loss.
Inside-bar breakout: a 15-minute inside bar, then a 1-minute break of the last 30 minutes’ range in the direction of the hourly trend. June 2023 to June 2026, entry at the next 1-minute open, stop assumed hit first when a bar touches both. Gross points per trade, before about 2 points of cost:
| Target | Index | Futures |
|---|---|---|
| 1.5× risk | +3.99 (+2,899 total) | −0.37 (−267 total) |
| 1.25× risk | +3.47 | +1.22 |
| 1.0× risk | +1.99 | +1.17 |
On the index it looked like a small, steady edge, positive every year. On futures it was gone.
Supertrend(10,2) (full post): close-based signals read from 30-minute index bars, so it suffers less, but the trading vehicle still decides the verdict. October 2022 to July 2026:
| Trades filled on | Cost per round trip | Total | t |
|---|---|---|---|
| NIFTY index | 3 pts | +10,954 pts | 2.07 |
| NIFTY synthetic futures | 2 pts | +8,448 pts | 1.60 |
| NIFTY futures | 5 pts | +6,874 pts | 1.30 |
On the index it passes our test of being “beyond luck” (a t of about 2). On anything you can actually trade, it doesn’t.
The distortion doesn’t always flatter. For strategies with a tight intrabar stop, phantom spikes on the index can trigger stops that futures never hit, so the index can understate a result too. Spot is not reliably optimistic or pessimistic; it is just wrong.
How to spot it
- Your fills reference the index’s own high, low or close.
- The edge comes from entries or exits that touch a level, rather than close beyond it.
- The strategy works on the index but not on the same dates in futures.
- Your “NIFTY” price file turns out to be the index, not a futures contract.
How we guard against it
Every intraday index result here is re-run with fills on a tradeable series (near-month or back-adjusted continuous futures) at realistic costs, and the spot-only figure is never the headline. See Methodology: costs and no look-ahead bias.
Was this useful?
More to say, or spotted a mistake? Write to [email protected]. We read everything.