Blog / 30 Sep 2026
How to read a backtest: twelve numbers, and the order to read them in
Total return is the number people look at first and the one least able to tell you whether a strategy is any good. A reading order, worked through on a real example.
A backtest hands you a dozen numbers at once, and most people look at one: the total return. That is the number least able to tell you whether a strategy is any good. Here is an order to read them in, using the example that opens when you first load StrategyLab: a 20 / 50 SMA crossover on two years of daily Bitcoin candles.
1. Compare it with doing nothing
The example turned $10,000 into $10,723.08, a total return of +7.23%. On its own that sounds fine. Buying on the first day and holding returned +37.37% over the same candles, with the same fee and slippage. The strategy trailed the simplest alternative by 30 points.
This is why StrategyLab puts buy & hold in the first sentence of every result. A strategy does not have to beat it to be worth using, but you should know what you are giving up and what you are getting in exchange.
2. Look at the worst stretch, not the end point
Max drawdown is the largest fall from a high point to a later low. The example’s was −32.22%; buy & hold’s was −53.08%. That is the exchange: the crossover made much less, and its worst moment was much less bad, because it spent only 45% of the time in the market.
Ask yourself honestly whether you would have kept following the rules at the bottom of that fall. A strategy you abandon at its low point earns its drawdown and none of its recovery.
3. Count the trades
The example made 9 trades. Four won and five lost. With nine trades, one lucky entry changes everything. StrategyLab prints a warning in this case: “Only 9 trades. That is too few to tell skill from luck.” Take it literally. As a rough working rule, a result built on fewer than about thirty trades tells you very little, and more is better.
4. Then the ratios
- Win rate (44% here) is the share of trades that made money. It means little alone: trend strategies often win less than half the time and still profit, because the wins are bigger than the losses.
- Average win / loss ($947 against $613) is the other half of that story. Multiply each by how often it happens and you have the strategy’s edge, if any.
- Profit factor (1.24) is money made on winners divided by money lost on losers. Above 1 is a profit. Just above 1, on nine trades, is noise.
- Sharpe (0.26) compares return with how bumpy the ride was. Above 1 is decent. Sortino (0.40) is the same idea but only counts downward moves as risk.
- Yearly growth, or CAGR (+3.56%) is the steady yearly rate that gives the same total. On a test shorter than a year it exaggerates, so check the dates.
5. Check the months
The monthly returns table shows where the result came from. In the example, a single month (August 2026, +25.0%) accounts for more than the whole two-year gain. Remove it and the strategy lost money. A result that depends on one month is a result about that month.
A short checklist
- Did it beat buy & hold? If not, did it give you a smaller drawdown in return?
- Could you have sat through the max drawdown?
- Are there enough trades for the numbers to mean anything?
- Does the profit come from many trades and months, or from one?
- Does it still work on data you did not tune it on?
The last question needs its own tools, and its own article. For the first four, open the example and read the tiles in this order.
The figures in this article are simulated results from one example: real BTC-USD daily candles from Coinbase Exchange, 1 Oct 2024 to 29 Sep 2026, $10,000 starting cash. Hypothetical performance has limits and is not a guarantee of future returns. Educational use only, not investment advice. See the risk disclosure.