Backtesting replays a strategy over historical data to estimate how it would have performed. Done honestly, it separates a genuine edge from a story — and it's the single cheapest way to avoid paying tuition to the market.
Test net of the full cost stack
An edge that looks great on paper can vanish once you subtract brokerage, STT, exchange & SEBI charges, GST, stamp duty and slippage. A useful backtest models all of it — India's F&O costs are not trivial.
The overfitting trap
Tune enough parameters on one sample and anything looks profitable. Guard against it: keep an out-of-sample holdout the search never sees, prefer strategies that work across many instruments (breadth), and be suspicious of results that are too clean.
Backtests do not guarantee future results. They estimate a distribution of outcomes under past conditions — regimes change. Use them to size conviction, not to promise returns.
What to look at
- Profit factor and net P&L after costs
- Win rate and average win vs average loss
- Max drawdown — can you actually sit through it?
- Out-of-sample vs in-sample — does the edge hold on unseen data?
The House of Trading lab backtests on real candles net of the full Indian F&O charge stack, with an out-of-sample check — so you see the edge before you trade it.