How to Backtest a Strategy

There are two kinds of people who skip backtesting: people who’ve never lost real money on a strategy that looked great in their head, and people who’ve already learned that lesson the expensive way. Let’s get you into the first group without the expensive part.

Running the backtest

Open Backtests, pick a strategy template, a pair, a date range and a timeframe, and run it. TradeFIQ replays your strategy on real historical candles, with fees included, and shows you every trade it would have taken. How far back you can go and how fine a timeframe you can use depend on your plan.

The mistake almost everyone makes first

Overfitting. It sounds technical, but the idea is simple: you tweak a strategy until it fits the past perfectly, and in doing so you teach it the past instead of the market. The classic example is testing on a date range that includes one huge, obvious move. You keep adjusting the entry until it “catches” that move. The result looks brilliant and is fitted to one event that will never repeat exactly.

The fix isn’t complicated, it’s just discipline: test on a range you haven’t stared at yet, and resist the urge to keep tweaking parameters until the backtest looks perfect. A strategy that’s been tuned to nail one specific historical period usually falls apart the moment real, unseen data shows up.

Picking a date range that actually means something

A backtest over two weeks tells you almost nothing, markets can look calm or wild for two weeks purely by chance. You want a range long enough to include a few genuinely different conditions, a trending stretch, some chop, ideally something volatile. If your strategy only performs during one type of market, that’s important information, but you need enough range in the test to actually find that out.

Reading the results without fooling yourself

Total P&L is the number everyone looks at first, and it’s the least useful one on its own. A strategy that made money through one enormous lucky trade and lost steadily everywhere else is not the same as one that ground out consistent small wins, even if the final number looks identical.

Look instead at:

• Win rate alongside the size of wins and losses. A 40% win rate can still be very profitable if wins are big and losses are small.
• Max drawdown, the worst peak-to-trough dip along the way. This is the number that tells you whether you could stomach running the strategy live.
• Why each trade closed. Every simulated trade shows whether it hit take-profit, stop-loss, an exit signal or a timeout. If nearly everything ends on the stop, your exits need work.
• Consistency across the whole range, not just the total. A strategy that’s flat for months and then spikes once is riskier than it looks.

Why your live results might differ anyway

Even a well-built backtest won’t perfectly match live trading, and that’s expected, not a sign anything’s broken. Slippage, latency, and real order-book conditions all play a role that a clean historical simulation can’t fully capture. We go into this properly in our piece on paper trading versus live trading, worth a read before you flip the switch.

The actual takeaway

A good backtest doesn’t tell you a strategy will definitely work. It tells you whether the logic behaves sensibly across a range of real conditions, and whether the drawdowns are something you could genuinely live with. That’s a lower bar than “will this make money,” and it’s a far more honest one.