Reading the backtest metrics that matter.

Net profit is the first thing everyone looks at, and the easiest to misread. TradingTune shows it both as a percent return and as an absolute amount. The percent is what lets you compare runs and assets fairly, because an absolute figure depends entirely on the position size you happened to test with.

The number that actually matters is net profit relative to a benchmark. If a strategy returns 40 percent over a stretch where simply buying and holding the asset returned 60 percent, the strategy underperformed despite a healthy looking total. Always ask what the market did over the exact same dates before you call a return good. Our worked examples line every tuned strategy up against buy and hold for precisely this reason.

Win rate is the share of trades that closed in the green, and it is the metric most likely to lead you astray. A strategy can win 90 percent of its trades and still lose money, if the occasional losers are far larger than the frequent winners. The reverse is just as common: a strategy that wins barely 40 percent of the time can be highly profitable when its winners dwarf its losers.

So win rate only means something paired with the size of wins versus losses. That is exactly what profit factor, average trade, and expectancy capture. Treat a high win rate as a question, not an answer: it tells you how often the strategy is right, never how much being right is worth.

Profit factor is gross profit divided by gross loss across all trades. A value of 1 means the strategy broke even before costs; below 1 it lost money, and above 1 it made money. The further above 1 the figure sits, the more room the edge has to absorb a rough patch before it turns unprofitable.

Profit factor neatly fixes the blind spot in win rate, because it accounts for the size of wins and losses rather than just their count. A modest win rate with a profit factor comfortably above 1 is a far stronger result than a sky-high win rate sitting barely above break-even. Read it as a measure of margin for error.

Two strategies can earn the same return while one rides a smooth equity curve and the other lurches wildly to get there. Risk-adjusted metrics separate them. The Sharpe ratio measures return per unit of volatility, rewarding smoother gains. Sortino refines the idea by counting only downside volatility, on the logic that upside swings are not what keeps traders awake at night.

As a rough guide, a Sharpe under 1 is weak, around 1 is decent, above 2 is strong, and anything extreme deserves suspicion rather than celebration. The crucial caveat: both ratios are easy to inflate with very few trades, where a couple of lucky outcomes masquerade as consistency. Never read a Sharpe without glancing at the trade count beside it.

Max drawdown is the deepest fall from an equity peak to the trough that follows, before a new peak is made. It answers the question that decides whether you can actually trade a strategy: how much pain would you have had to sit through to collect its returns. A 30 percent drawdown means your account was, at its worst, down nearly a third from its high water mark.

This is where a glowing net profit can quietly hide real danger. A run that doubled your money but suffered a 70 percent drawdown along the way is one most people would have abandoned long before the recovery, and a touch more leverage would have meant ruin. Read net profit and max drawdown as a pair: the return is only worth the drawdown you could realistically have survived.

Every other metric is only as trustworthy as the number of trades it rests on. Trade count is your sample size. With just a handful of trades, a dazzling win rate, profit factor, and Sharpe can all be pure luck, the statistical equivalent of flipping heads a few times in a row and declaring the coin rigged.

There is no universal threshold, but a few dozen trades is a bare minimum to take a result seriously, and triple digits is far more reassuring. Too few trades is a genuine red flag, not a minor footnote. When you scan an optimization run, let trade count gate everything else: a top row with too little activity behind it has not earned a place on your shortlist.

Average trade is net profit divided by trade count: the typical result of putting the strategy on once. Expectancy expresses the same idea from the win and loss side, blending how often you win with how much you win and lose, to give the average amount you can expect each trade to add. A positive expectancy is the mathematical heart of a real edge.

Average trade matters for a practical reason too. If the typical trade is small, real-world commissions, spread, and slippage can swallow the edge whole, even when the raw backtest looks profitable. A strategy whose average trade comfortably clears its trading costs is far more likely to survive contact with a live market.

No single metric ranks a strategy. The skill is reading them as a system, where each number checks the others. Net profit is only good relative to buy and hold. Win rate is only meaningful next to profit factor. Sharpe and a high return only count once the max drawdown looks survivable. And every one of them is only as solid as the trade count underneath it.

A trustworthy run is rarely the one with the single highest number in any column. It is the one that holds up in all of them at once: a sensible return that beats the benchmark, a profit factor with margin to spare, a drawdown you could have lived through, and enough trades to believe it was not luck. That all-round quality is what survives out of sample, the theme of our guide to avoiding overfitting.

The results table is sortable and filterable on every metric, which turns this whole framework into a quick, repeatable workflow rather than a row-by-row squint. A practical pass looks like this:

• Filter out rows below a sensible trade count first, so noise never reaches your shortlist.
• Filter again on max drawdown, dropping anything deeper than you could realistically have held.
• Now sort the survivors by net profit, profit factor, or Sharpe to see which robust runs also lead on return.
• Eyeball the neighbours of the top rows. A strong result surrounded by other strong results is far more believable than a lone spike.

Once a run earns your trust, apply its best result in one click, then use the multi-asset retest to see whether those same metrics hold up on related symbols. Filtering and sorting do not just find the biggest number, they find the result that is good across the board, which is the one worth taking forward.

See these metrics on real charts in our worked strategy backtests compared to buy and hold, where every tuned run is measured against a plain benchmark over the same dates. If a term here was unfamiliar, the trading and optimization glossary defines each one plainly.

To understand how the choice of search shapes the results you are reading, browse the seven optimization methods TradingTune uses. And to make sure a great-looking row is a real edge rather than a curve fit, read the companion guide to avoiding overfitting.

New here? Start with the getting started walkthrough to see this live results table fill in for the first time, then learn the common backtesting mistakes that make these metrics lie so you know what to discount as you read.