Appearance

Personalize your experience

Mode

Accent Color

Direction

Performance Metrics

Backtesting

What each figure on the results page is computed from, what it hides, and the order to read them in so the important ones are not drowned out by the return.

Read them in this order

The page is not ordered by importance, and the return is the most prominent number on it. That is exactly backwards. Reading in this order stops you from evaluating a on the one figure that says least about whether it is usable.

1
1. Trade count
Below 30, stop reading — nothing else on the page is meaningful. This is a gate, not a metric.
2
2. Max drawdown
Decides whether the is holdable. A 60% return through a 45% is a strategy almost nobody actually sits through.
3
3. Profit factor and expectancy
Whether there is an edge, and how it is per trade. Independent of how much capital was deployed.
4
4. Win rate with payoff ratio
Always together. Either one alone is uninterpretable.
5
5. Sharpe or Sortino
Return per unit of risk, which is what makes two strategies comparable.
6
6. Total return
Last. It is the consequence of everything above plus the position size you happened to choose.

P&L metrics

Metric	Type	Description
totalPnl	currency	Net profit after costs. Depends on position size, so it does not across strategies.
totalPnlPercent	%	Net profit against initial capital. Comparable only if the risk taken was similar.
annualisedReturn	%	Return scaled to a year. Beware extrapolating from a short test — three good months annualised is a fiction.
averageWin	currency	Mean profit on winning .
averageLoss	currency	Mean loss on losing . against your : if it is much larger, stops are slipping or gapping.
profitFactor	ratio	Gross profit / gross loss. 1.0 is break-even. Above 1.5 suggests a real edge; far above 3 on a sample usually suggests overfitting rather than brilliance.
expectancy	currency	Average P&L per trade. The single most practical number here: multiply by expected to project, and if it is negative no amount of helps.

how they are computed

1const wins   = trades.filter(t => t.pnl > 0);2const losses = trades.filter(t => t.pnl < 0);34const grossProfit = wins.reduce((s, t) => s + t.pnl, 0);5const grossLoss   = Math.abs(losses.reduce((s, t) => s + t.pnl, 0));67const profitFactor = grossLoss === 0 ? Infinity : grossProfit / grossLoss;89const winRate    = wins.length / trades.length;10const averageWin = grossProfit / wins.length;11const averageLos = grossLoss / losses.length;1213// expectancy — what one trade is worth on average14const expectancy = winRate * averageWin - (1 - winRate) * averageLos;1516// equivalently, and more usefully for reasoning:17const payoff   = averageWin / averageLos;18const breakEven = 1 / (1 + payoff);      // the win rate you need

Caution:A of Infinity means zero losing , which in practice means the sample is too or something is wrong. It is never good news.

Win rate and payoff, together

Neither number means anything alone, and quoting one without the other is the most common way performance gets misrepresented — including to oneself.

Payoff ratio	Type	Description
1 : 1	break-even at 50%	Costs push the real requirement to roughly 53–55%. A thin margin for error.
2 : 1	break-even at 33%	Two losses in three is still profitable. Comfortable.
3 : 1	break-even at 25%	Where most trend-following strategies . Uncomfortable to run, robust to a bad streak.
1 : 2	break-even at 67%	Achievable for mean reversion. Fragile: one gapped stop erases four wins.
1 : 4	break-even at 80%	Even an 80% win rate only breaks even. Almost never worth running.

★

Pro tip:The asymmetry is worth internalising. A high win rate with a poor payoff feels better — most are pleasant — and is structurally more fragile, because the rare loss is enough to undo a long run of wins. A low win rate with a good payoff feels worse and breaks less.

Risk metrics

Metric	Type	Description
maxDrawdown	%	Largest peak-to-trough equity decline. The number that determines whether you can actually run the , as distinct from whether it is profitable.
drawdownDuration	days	How long the worst trough lasted. Frequently more decisive than depth — a 15% over eight months is harder to hold than a 25% one over three weeks.
sharpeRatio	ratio	Excess return divided by total volatility. Above 1 good, above 2 excellent. Penalises upside volatility as though it were risk, which is its main flaw.
sortinoRatio	ratio	Like , but only downside deviation in the denominator. Fairer to strategies with wins and small losses, which is most trend-following.
calmarRatio	ratio	Annualised return / . Return per unit of worst-case pain. Above 1 acceptable, above 3 strong.
ulcerIndex	index	Depth and of drawdowns combined. Captures the experience of holding a better than alone.
recoveryFactor	ratio	Net profit / . How much return was earned per unit of drawdown endured.
maxConsecutiveLosses	count	The longest losing streak. Ask honestly whether you would still be running the at the end of it — most abandonment happens here rather than at the low.

Drawdown deserves its own attention

is measured from the running peak, so it is the loss you would have experienced had you started at the worst possible moment. That is the definition, because that is the experience that ends strategies.

drawdown series

1let peak = equity[0];2const drawdown = equity.map(v => {3  peak = Math.max(peak, v);4  return (v - peak) / peak;        // 0 at a new high, negative below5});67const maxDrawdown = Math.min(...drawdown);89// duration matters as much as depth — measure the longest run below a peak10let longest = 0;11let current = 0;12for (const d of drawdown) {13  current = d < 0 ? current + 1 : 0;14  longest = Math.max(longest, current);15}

Depth tells you the capital requirement. A 30% means the must be funded to survive losing 30% without intervention.
tells you the psychological requirement, and it is the one people underestimate. Eight months below a previous high is eight months of wondering whether the is broken.
Frequency tells you what is normal. Several 10% drawdowns is a different proposition from one 30% , even at the same total return.

★

Pro tip: the question concretely before deploying: if this loses maxDrawdown over drawdownDuration starting on day one, will I still be running it at the end? If the honest answer is no, the is unsuitable regardless of its return — and the response is smaller size, not more optimism.

Advanced metrics

Metric	Type	Description
systemQualityNumber	ratio	Expectancy divided by the of trade P&L, scaled by the of . Combines edge, consistency and sample size into one figure.
annualisedVolatility	%	of returns, annualised. Useful for comparing against a benchmark.
skewness	number	Asymmetry of the return . Positive means a long tail — occasional wins, which is the desirable .
kurtosis	number	Tail weight. Above 3 means fatter tails than normal, so extreme moves are more likely than the volatility figure implies. Read alongside .
monthlyReturns	grid	P&L by month. The fastest way to see whether a year’s result came from two months.
averageDuration	bars	Mean holding . Far shorter than intended usually means the stop sits inside normal noise.

ℹ

Note:The monthly returns grid is under-rated. A with a 24% annual return where twenty came from one month is not a 24% strategy — it is a strategy that caught one move, and the aggregate metrics cannot tell you that.

Thresholds worth holding to

Metric	Type	Description
Trades	> 30, ideally > 100	A gate before anything else is read.
Profit factor	> 1.5	Below 1.3 the edge is likely inside your cost uncertainty.
Max drawdown	< 20–30%	Depends on temperament and capital. Be conservative: the lived experience is worse than the chart.
Sharpe	> 1.0	Above 3 on a retail warrants suspicion rather than celebration.
Calmar	> 1.0	Return should at least match the worst .
Expectancy	positive after costs	Non-negotiable. A negative expectancy cannot be by position sizing.

Caution:These are filters, not targets. Tuning a until it clears them is precisely how overfitting happens. Use them to reject, never as an optimisation .

Appearance

Performance Metrics

Read them in this order

P&L metrics

Win rate and payoff, together

Risk metrics

Drawdown deserves its own attention

Advanced metrics

Thresholds worth holding to

See also

Related articles

Appearance

Read them in this order

P&L metrics

Win rate and payoff, together

Risk metrics

Drawdown deserves its own attention

Advanced metrics

Thresholds worth holding to

See also

Related articles

Related to Performance Metrics