Backtesting

Backtesting is the process of applying a trading strategy to historical market data to see how it would have performed in the past. Before risking real money, traders and quantitative analysts backtest their strategies to evaluate their viability, understand the risk profile, and refine the rules. A strategy that performs well in a backtest has a higher chance of working in live markets than one that has never been tested.

What Is Backtesting?

In backtesting, you apply your buy and sell rules to historical price data and calculate the theoretical trades that would have been made and the resulting profits and losses. Modern backtesting tools handle this automatically given a set of rules or code.

What Backtesting Measures

– **Total return**: how much the strategy would have gained over the test period
– **Win rate**: percentage of trades that were profitable
– **Maximum drawdown**: the largest peak-to-trough decline
– **Sharpe ratio**: return per unit of risk
– **Average win and average loss**: used to calculate expectancy
– **Number of trades**: total trades and frequency

Limitations of Backtesting

**Overfitting (curve fitting)**: a strategy optimised excessively on historical data “fits” that specific data set but fails on new data because it has memorised noise rather than real patterns. Signs of overfitting include too many rules, exceptional historical results with poor forward performance.

**Survivorship bias**: using only current stocks in a backtest ignores companies that went bankrupt or were delisted. This biases results upward.

**Look-ahead bias**: inadvertently using data that would not have been available at the time of the trade (e.g., end-of-day prices to enter at the open).

**Slippage and costs**: real-world trading involves bid-ask spread, brokerage, STT, and slippage that reduce returns compared to theoretical backtest results.

Tools for Backtesting in India

– **Amibroker**: popular for technical strategy backtesting
– **Python (pandas, backtrader, vectorbt)**: for quantitative strategies
– **Zerodha Streak**: simplified backtesting for retail traders on Indian markets
– **TradingView Pine Script**: for strategy testing on charting platform

Practical Example

Rajan codes a simple crossover strategy in Python: buy Nifty when the 50-day MA crosses above the 200-day MA; sell when it crosses below. He backtests this from 2010 to 2023 on historical Nifty data. The result shows 60% win rate, 22% CAGR, and 28% maximum drawdown. He then forward tests it on new data to check if performance is consistent.

Key Takeaways

– Backtesting validates a trading strategy on historical data before deploying real capital
– Key metrics include total return, win rate, maximum drawdown, and expectancy
– Overfitting, survivorship bias, and look-ahead bias are the main risks in backtesting
– Always account for transaction costs and slippage in backtests
– A good backtest must be followed by forward testing (paper trading or small live capital) to confirm results