Drawdown, APR and why both matter

A backtest runs a strategy over past prices with the exact settings you chose and reports what would have happened. The headline is the return, but a return on its own hides the part that decides whether you could actually have earned it.

Total return and APR

Total return is what the strategy made or lost over the test window, as a share of the money put in, after taker fees. It sits beside buy and hold: what you would have made by buying on the first day and doing nothing. That comparison comes first. A strategy that made 8% while holding made 20% did its job badly, however good 8% sounds.

APR stretches the same pace to a year: (1 + return) to the power of 365 divided by the number of days, minus 1. It lets you compare a 60-day test with a 300-day one. It is a projection, not a result. A 10% gain over 60 days becomes an APR of about 79%, and nothing promises the next 300 days will look like those 60. The shorter the window, the more an APR flatters.

Maximum drawdown

Maximum drawdown is the largest fall in account value from a peak to a later low, during the test, measured on full equity. It answers a question the return does not: how bad did it get on the way?

Two strategies can end with the same return and very different paths. One climbs steadily; the other falls 35% before recovering. In real money the second is the one people stop at the bottom, which turns a temporary drawdown into a permanent loss. The backtest page warns you when the deepest dip was larger than everything the run made.

Win rate, and why it misleads alone

Win rate is the share of exits that closed above their average cost, before fees. It says how often, not how much, and it means different things for different strategies.

• For a trend follower, a low rate can be a good run: many small losses, a few large wins.
• For a grid, the rate runs high by design, one small gain per step. It cannot show the position left holding when the range breaks.
• For a martingale, a high rate can hide the one stuck cycle that matters.

That is why the tooltip beside the win rate on each strategy page reads differently by strategy type. Read it with the drawdown, every time.

Where a backtest stops being a guide

A backtest is the best evidence you can get before risking money, and it still differs from live trading in known ways.

• Fees. The backtest charges standard taker fees. Your account may pay a little less, and a grid's resting orders usually pay maker rates, but the fee is never zero.
• Slippage. Live orders can fill worse than the price the bar shows, especially on thin pairs and at the US open for stock tokens.
• Bar boundaries. A daily bar on OKX closes at 16:00 UTC, on Binance at midnight UTC. The same strategy on the same coin can score differently depending on where the day is cut. For a daily-bar result, A9's nightly shortlist re-runs it with the day cut at six different hours and keeps it only if it makes money at the typical cut and at most of them.
• Settings that are too exact. Auto-tune and the nightly shortlist also re-run the chosen settings with each one nudged by 20%, and report what the neighbours earned. If the result falls apart, the headline leaned on lucky values.
• Short history. Tokenized stocks listed in mid-2026, so most have only months of data.

A simple way to read a result

1. Compare the return with buy and hold over the same window.
2. Read the maximum drawdown and ask whether you would sit through it in real money.
3. If the settings came from Auto-tune, check the neighbours' result for a big gap.
4. Only then look at the APR and the win rate.