// guide

Backtest metrics explained

Backtest metrics describe a strategy from four sides: how much it made, how much it hurt on the way, how much return it earned per unit of risk, and how much evidence stands behind the numbers. This guide defines each metric the way Janorium computes it, gives the thresholds the report uses, and walks through a real report where the metrics disagree.

7 min readUpdated 20 October 2026

Read metrics in groups, not one at a time

Every single metric can be made to look good. A strategy with a 90% win rate can lose money; one with a huge return can have been half-wiped-out along the way. Read the four groups together:

Group	The question	Main metrics
Return	Did it make money — and more than holding?	Total return, CAGR, buy and hold, alpha
Risk	How much did it hurt on the way?	Max drawdown, longest drawdown, Ulcer index
Risk-adjusted	Was the return worth the ride?	Sharpe, Sortino, Calmar
Trades and evidence	How does it make money, and how sure can we be?	Win rate, payoff ratio, profit factor, expectancy, number of trades

Return: The questionDid it make money — and more than holding?; Main metricsTotal return, CAGR, buy and hold, alpha
Risk: The questionHow much did it hurt on the way?; Main metricsMax drawdown, longest drawdown, Ulcer index
Risk-adjusted: The questionWas the return worth the ride?; Main metricsSharpe, Sortino, Calmar
Trades and evidence: The questionHow does it make money, and how sure can we be?; Main metricsWin rate, payoff ratio, profit factor, expectancy, number of trades

Return: total, CAGR and buy and hold

Total return — the gain or loss on the initial capital over the whole run, after fees, slippage and funding.
CAGR — the compound annual growth rate: the yearly return that would produce the same result. Use it to compare runs of different length.
Buy and hold — what simply holding the coin over the same window returned. It is the benchmark every strategy has to beat.
Alpha — total return minus buy and hold. Negative alpha means holding would have done better.

In crypto, buy and hold is a demanding benchmark: in 2020–2025 most large coins rose many times over. A strategy that "made +300%" while the coin made +800% did not add value — unless it did so with a much smaller drawdown.

Risk: max drawdown and how long it lasted

Max drawdown — the largest fall of equity from a previous peak, in percent. A 50% drawdown needs a +100% gain just to get back.
Longest drawdown — the longest stretch from a peak until it was exceeded again. Depth is one kind of pain; two years underwater is another.
Ulcer index — the quadratic mean of the drawdown series: it punishes drawdowns that are both deep and long.

The report rates max drawdown below 10% as shallow, 10–25% as moderate, 25–40% as deep and 40% or more as severe. These are judgements about pain, not about the strategy: leverage and position size move drawdown directly.

Sharpe, Sortino and Calmar: return per unit of risk

Sharpe ratio — the average daily return divided by the standard deviation of daily returns, annualised with √365 because crypto trades every day. Janorium uses no risk-free rate.
Sortino ratio — the same, but only downside volatility counts, so a strategy is not penalised for big up days.
Calmar ratio — CAGR divided by max drawdown: yearly return earned per unit of worst-case pain.

Metric	Losing	Weak	Good	Strong
Sharpe	below 0	0 to 1	1 to 2	2 and above
Sortino	below 0	0 to 1.5	1.5 to 3	3 and above
Calmar	below 0	0 to 0.5	0.5 to 1	1 and above

Sharpe: Losingbelow 0; Weak0 to 1; Good1 to 2; Strong2 and above
Sortino: Losingbelow 0; Weak0 to 1.5; Good1.5 to 3; Strong3 and above
Calmar: Losingbelow 0; Weak0 to 0.5; Good0.5 to 1; Strong1 and above

Win rate, payoff ratio, profit factor and expectancy

Win rate — the share of trades that closed with a profit.
Payoff ratio — the average win divided by the average loss.
Profit factor — gross profit divided by gross loss. Above 1 the winners outweigh the losers.
Expectancy — the average profit or loss per trade: win rate × average win + loss rate × average loss.

Win rate means nothing on its own. The break-even win rate is 1 / (1 + payoff ratio): with winners twice the size of losers, winning one trade in three already breaks even, while a strategy that wins 60% of the time with winners a third the size of its losers loses money. Trend-following strategies usually have low win rates and large payoffs; mean-reversion strategies the opposite.

Metric	Weak	Good	Strong
Profit factor	below 1 unprofitable, 1 to 1.3 thin	1.3 to 1.8	1.8 and above
Payoff ratio	below 1 small wins, 1 to 1.5 balanced	1.5 to 2	2 and above

Profit factor: Weakbelow 1 unprofitable, 1 to 1.3 thin; Good1.3 to 1.8; Strong1.8 and above
Payoff ratio: Weakbelow 1 small wins, 1 to 1.5 balanced; Good1.5 to 2; Strong2 and above

How much evidence: trades, periods, coins

Number of trades. Every ratio above is an average, and averages over a dozen trades are noise. Look for dozens to hundreds.
Several periods. Check bull, bear and sideways stretches — the report breaks results down by market regime.
Several coins. A rule that works on one coin and fails on its neighbours has probably found that coin’s history, not a pattern.
Shape versus scale. Doubling every position doubles return and drawdown but leaves Sharpe, Sortino and profit factor unchanged. Judge the strategy by the shape metrics, the bet size by the scale ones — leverage never makes a strategy better.

A real report where the metrics disagree

The BTC run from the golden-cross study: hold BTC on Binance spot while the 50-day SMA is above the 200-day SMA, 2020-01-01 → 2026-01-01, with fees and slippage.

Metric	Value	How the report reads it	What it really says
Total return	+529.12%	—	Made money…
Buy and hold	+859.56%	—	…but holding made more: alpha −330.45 points.
CAGR	40.08%	—	Strong in isolation, weak next to the benchmark.
Max drawdown	57.11%	severe	More than half of the account lost at the worst point, and the run ended below its peak.
Sharpe · Sortino	0.96 · 1.48	weak · weak	Not much return per unit of volatility.
Calmar	0.70	good	Decent return per unit of drawdown.
Profit factor	4.95	strong	Looks superb…
Win rate · payoff	66.67% · 2.48	— · strong	…on only six trades.
Trades	6	—	Not enough evidence for any of the trade metrics above.

Total return: Value+529.12%; How the report reads it—; What it really saysMade money…
Buy and hold: Value+859.56%; How the report reads it—; What it really says…but holding made more: alpha −330.45 points.
CAGR: Value40.08%; How the report reads it—; What it really saysStrong in isolation, weak next to the benchmark.
Max drawdown: Value57.11%; How the report reads itsevere; What it really saysMore than half of the account lost at the worst point, and the run ended below its peak.
Sharpe · Sortino: Value0.96 · 1.48; How the report reads itweak · weak; What it really saysNot much return per unit of volatility.
Calmar: Value0.70; How the report reads itgood; What it really saysDecent return per unit of drawdown.
Profit factor: Value4.95; How the report reads itstrong; What it really saysLooks superb…
Win rate · payoff: Value66.67% · 2.48; How the report reads it— · strong; What it really says…on only six trades.
Trades: Value6; How the report reads it—; What it really saysNot enough evidence for any of the trade metrics above.

Read alone, the profit factor says "excellent strategy". Read together, the metrics say: lost to holding, deep drawdown, a handful of trades. That is why the study went on to test 19 coins instead of one — and found that the rule beat holding on only 3 of the 12 that rose.

Cheat sheet: questions to ask any backtest

Did it beat buy and hold — or at least match it with a much smaller drawdown?
Would I have kept running it through the max drawdown and the longest one?
Is the Sharpe in a believable range for the number of trades behind it?
Do win rate and payoff ratio together give a profit factor comfortably above 1?
Are there enough trades, coins and periods to call it evidence?
Does it survive doubled costs and a different start date?

To run your own backtest and get all of these numbers, follow how to backtest a crypto strategy. For the bigger picture — what a backtest is and where it misleads — see what is backtesting.

Frequently asked questions

What is a good Sharpe ratio for a crypto strategy?

In a backtest, a Sharpe ratio between 1 and 2 is good and above 2 is strong; below 1 is weak. Treat values above 2–3 with suspicion: they often come from too few trades, missing costs or overfitting.

What is a good max drawdown?

Below 10% is shallow, 10–25% moderate, 25–40% deep and above 40% severe. What matters most is whether you would have kept the strategy running through it — and that leverage makes it deeper.

What does profit factor mean?

Profit factor is gross profit divided by gross loss. Above 1 the strategy made money; 1.3–1.8 is good and above 1.8 strong — provided there are enough trades behind it.

Is a high win rate good?

Not by itself. A strategy can win 70% of its trades and lose money if the losers are much larger than the winners. Read win rate together with the payoff ratio: the break-even win rate is 1 / (1 + payoff ratio).

What is the difference between Sharpe and Sortino?

Sharpe divides return by all volatility; Sortino only by downside volatility. A strategy with large upside swings gets a better Sortino than Sharpe.

What is the Calmar ratio?

The Calmar ratio is the compound annual growth rate divided by the max drawdown. Above 1 means the strategy earned more per year than its worst drawdown.

Test your own idea

Create a free account, start from a template or from your own rules, and read your first report.

Create a free account Browse the leaderboard