Testing Your Strategy (Backtesting)

You’ve built your strategy. Now comes one of the most important steps in algo trading: Backtesting — simply put, testing your logic on past market data.

Think of it like a dress rehearsal before the live show. It tells you: “If I had used this strategy earlier, how would it have performed?”

Let’s break it down step by step.

Why Backtest?

Before you risk real money, you need answers like:

• Does this strategy actually work?
• How many trades does it take in a day/week?
• Is it mostly profitable or loss-making?
• What’s the biggest loss it ever took?

Backtesting gives you a clear picture — without any real-world risk.

What You’ll Need:

• Access to historical data (this is usually built into the platform you use)
• A strategy builder (dropdown-based or rule-based interface)
• Defined logic (from Chapter 3.1)

Steps to Backtest

1. Choose Your Symbol
   Example: Pick a stock or index like Nifty, Bank Nifty, or a stock you follow.

2. Select the Timeframe
   For intraday: 5-min or 15-min For swing: Daily or hourly charts

3. Set Entry & Exit Rules
   

Example:

a. Entry: When 5 EMA crosses above 20 EMA b. Exit: 2% profit or 1% stop-loss OR 5 EMA crosses below 20 EMA

4. Set Capital Allocation Let’s say ₹10,000 per trade. This helps calculate profits/losses.

5. Run the Backtest The system will scan the historical data, simulate all matching trades, and give you results.

Key Metrics to Watch

After running the backtest, look at:

• Total Trades: How many were taken?
• Win Rate: Percentage of winning trades
• Average Profit/Loss: Per trade result
• Maximum Drawdown: Worst loss from peak to bottom
• Profit Factor: Total profit ÷ Total loss (above 1 is good)

These numbers help you decide whether to:

• Keep the strategy as-is
• Modify the entry/exit
• Ditch it completely

Common Mistakes to Avoid

• Don’t overfit the strategy just to get perfect past results. That’s called curve fitting and it often fails in real trading.
  
• Make sure the data used is clean and includes all market conditions — trending, sideways, volatile.
  

Final Thought:

Backtesting doesn’t guarantee success — but skipping it guarantees trouble.

It’s your safety net, your feedback loop, and your teacher. One step closer to going live with your first algo!