The Complete Backtesting Checklist: 15 Things to Verify Before Going Live

You’ve built a strategy, backtested it, and the results look great. Before you risk real money, run through this checklist. Each item represents a common failure point that separates strategies that work on paper from strategies that work in practice. I’ve learned most of these the hard way.

Data Quality

1. Is your test period long enough?

Three months of bull market will make any long strategy look profitable. That’s not your strategy working — that’s the trend doing the heavy lifting. Test on at least 2-3 years of data. You need to see how your strategy handles trending markets, ranging markets, high volatility crashes, and the boring sideways periods in between.

2. Does your data include different market regimes?

Check the dates of your test period. Did it cover a major correction? A prolonged sideways phase? If your entire test window was a bull run, your results are meaningless for predicting forward performance. On StratBase.ai, you can access up to 5 years of data across crypto, forex, and stocks — enough to capture multiple full market cycles.

3. Are you using the right timeframe?

A strategy designed for 4-hour charts shouldn’t be tested on daily candles. The signals are completely different. Make sure your backtest timeframe matches the timeframe you’ll actually trade on. If you plan to scalp on 1-minute charts, test on 1-minute data — not 15-minute approximations.

Cost Realism

4. Are real exchange fees included?

This is where most paper profits evaporate. A strategy making 500 trades per year with 0.04% taker fee each way gives up 40% of your capital to the exchange annually (500 × 0.08%). If your average profit per trade is 0.15%, fees just ate more than half of it.

Use the exact fee schedule of your exchange. The difference between Binance maker (0.02%) and taker (0.04%) fees matters when compounded over hundreds of trades.

5. Is funding rate accounted for?

If you’re trading perpetual futures, funding rate is a hidden cost that most backtests ignore. It’s charged every 8 hours and can be positive or negative. During trending markets, funding can reach 0.1% per interval — that’s 0.3% per day just for holding a position.

6. Have you considered slippage?

Your backtest assumes perfect execution at the candle close price. Reality gives you slippage, partial fills, and requotes. For liquid pairs like BTC/USDT on major exchanges, slippage is minimal. For smaller altcoins or during high-volatility events, it can wipe out your edge entirely.

Strategy Robustness

7. Does it work on more than one instrument?

A strategy that only profits on BTC/USDT during a specific 6-month window isn’t capturing a market pattern — it’s fitting to noise. Test the same logic on at least 3-5 similar instruments. It doesn’t need to be profitable on all of them, but if it only works on one, you’re likely overfitting.

8. Are parameters robust or fragile?

Change each parameter by 10-20%. Does the strategy still work? If switching RSI from 14 to 12 turns your +80% annual return into a -15% loss, your parameters are fragile. A robust strategy shows gradually changing results when you adjust parameters — not cliff-edge dropoffs.

9. Is your entry logic realistic?

Be honest about execution. Can you actually enter at the signal price? If your strategy requires catching the exact bottom of a wick, you won’t replicate that in live trading. Strategies that use candle close prices for signals are more realistic than those relying on intra-candle extremes.

Risk Management

10. Is there a stop loss?

No stop loss means unlimited downside risk. One flash crash or black swan event can erase months of gains in minutes. Every strategy needs a defined maximum loss per trade. Use ATR (Average True Range) to set stops that match the instrument’s volatility rather than arbitrary percentages.

11. Is the risk/reward ratio acceptable?

Minimum viable ratio is 1:1.5. If your stop loss is 2%, your take profit should be at least 3%. Even with a 60% win rate, a 1:1 risk/reward barely breaks even after fees. The math is unforgiving.

12. Can you survive the maximum drawdown?

Look at the equity curve, not just the final number. A strategy that makes +100% but has a 50% drawdown means your account was cut in half at some point. Most traders psychologically break at 25-30% drawdown and close positions at the worst possible time. If the max drawdown exceeds what you can stomach, reduce position size or find a different strategy.

Statistical Validity

13. Are there enough trades?

A strategy with 8 trades over 2 years tells you nothing statistically. You need at least 30-50 trades for the results to be meaningful, and 100+ trades for real confidence. If your strategy only triggers once a month, consider testing on additional instruments to increase the sample size.

14. Is the win rate paired with the right metrics?

A 90% win rate sounds amazing until you realize the average winner is $10 and the average loser is $200. Win rate alone is meaningless. Always pair it with average win/loss ratio and profit factor. A strategy with 40% win rate and 3:1 reward-to-risk is far superior to 80% win rate with 0.3:1.

15. Would you trust this strategy with real money tomorrow?

This is the final gut check. After reviewing all metrics, looking at the equity curve, checking drawdowns, and accounting for real fees — would you wire money to your exchange account and turn this on right now? If there’s hesitation, identify what’s causing it and fix it before going live.

The Bottom Line

A backtest isn’t proof that a strategy works. It’s a filter that eliminates strategies that definitely don’t work. If your strategy passes all 15 checks on this list, it has earned the right to be tested with real capital — starting with the smallest position size your exchange allows.

The goal isn’t finding a perfect strategy. The goal is finding one that’s honest enough to survive contact with the real market. That starts with an honest backtest.

Further Reading

• RSI on Investopedia
• Drawdown on Investopedia
• Binance