Backtesting is powerful — but only if done correctly. Many traders see impressive backtest numbers and trade the strategy live, only to lose money. Almost always, the reason is one of these eight mistakes.

Mistake 1 — Testing on Too Short a Period

Testing a strategy on just 6 months or 1 year of data is not enough. Markets go through different cycles — bull runs, bear phases, and sideways consolidation. A strategy that works in a bull market may fail badly in a bear market.

Fix: Always test on at least 3–5 years of data covering at least one bull and one bear cycle. For NSE, 2019–2024 includes the COVID crash and recovery — an excellent test period.

Mistake 2 — Ignoring Trading Costs

A strategy with 200 trades per year and ₹40 brokerage per trade loses ₹8,000 in brokerage alone — before STT, exchange charges, and slippage. Many strategies that look profitable before costs are losing strategies after costs.

Always set realistic brokerage (₹20 flat for discount brokers) and enable the STT and slippage settings in Momentum IQ before trusting any result.

Mistake 3 — Overfitting (Curve Fitting)

Overfitting means tuning your strategy parameters until they look perfect on historical data. The strategy becomes a perfect description of the past, not a predictor of the future.

Example: Testing RSI with periods 2, 3, 4... 50 and choosing RSI(7) because it gave 45% CAGR is overfitting. The RSI(7) happened to work on that data — it may not work going forward.

Fix: Decide your parameters based on logic, not search. Test RSI(14) because 14 is the standard, not because it gave the best backtest number.

Mistake 4 — Look-Ahead Bias

Look-ahead bias happens when your strategy uses information that would not have been available at the time of the trade. For example, using the closing price of a candle to trigger a trade on that same candle.

Momentum IQ's backtest engine executes trades on the open of the next candle after a signal is generated — this correctly avoids look-ahead bias.

Mistake 5 — Not Testing Out-of-Sample

Train your strategy on one period (2015–2020), then test it on data it has never seen (2021–2024). If it still works on the out-of-sample period, you have genuine edge. If it fails, the strategy was overfitted.

PeriodPurpose
2015–2020In-sample — develop and tune the strategy
2021–2024Out-of-sample — validate without touching parameters

Mistake 6 — Ignoring Max Drawdown

A strategy with 40% CAGR and 60% max drawdown sounds attractive until you realise it means your ₹1,00,000 could drop to ₹40,000 at its worst. Most traders cannot psychologically handle such drawdowns and exit at the worst possible time.

If you cannot tolerate the Max Drawdown in real money, do not trade the strategy live — regardless of the CAGR.

Mistake 7 — Survivorship Bias

NSE has many stocks that were delisted, went bankrupt, or were removed from indices. If you only test on stocks that exist today, you are testing on the survivors — which biases results upward.

Fix: Prefer testing on indices (NIFTY 50, NIFTY 500) rather than individual stock screens. Indices have defined inclusion/exclusion rules that reduce survivorship bias.

Mistake 8 — Over-Relying on a Single Strategy

No strategy works in all market conditions. Even the best backtested strategies have periods of underperformance lasting 6–18 months. Traders who put 100% of capital in one strategy often give up exactly when the strategy is about to recover.

Fix: Backtest 3–4 uncorrelated strategies and run them simultaneously with capital split between them. When one underperforms, others typically compensate.

Quick Mistake Checklist

  • ✓ Tested on 3+ years of data
  • ✓ Brokerage and charges included
  • ✓ Parameters chosen logically, not by searching for best results
  • ✓ Validated on out-of-sample period
  • ✓ Max Drawdown is acceptable for your risk tolerance
  • ✓ Tested on index or survivorship-bias-corrected data
  • ✓ Not the only strategy in your portfolio