Break-Even Win Rate: The Formula Most Traders Get Wrong

At a 0.30% gross target with a symmetric stop and taker fills on both sides, you need a 68.3% win rate to break even. The number most fee calculators return for the same setup is closer to 54%, and the gap is one modelling error.

The error

The common approach subtracts the fee from winning trades and leaves losing trades at their nominal size. That is wrong in an obvious way once stated: a losing trade is also two fills, and both are charged. The fee applies to every round trip regardless of outcome.

Written out, with gross target T and round-trip fee F, both in percent:

win  = T − F
loss = T + F
break-even p = (T + F) / (2T)

A win comes in smaller than the target because fees are taken out of it. A loss comes in larger than the stop because fees are added on top. Both effects push in the same direction, which is why the correct answer sits so far above the fee-free 50%.

What it looks like at real fee levels

Bybit's standard schedule puts a taker round trip at 0.11% and a maker round trip at 0.04%.

Gross target	Fee-free	Taker (F = 0.11%)	Maker (F = 0.04%)
0.10%	50.0%	impossible	70.0%
0.20%	50.0%	77.5%	60.0%
0.30%	50.0%	68.3%	56.7%
0.50%	50.0%	61.0%	54.0%
1.00%	50.0%	55.5%	52.0%

The first row is not a rounding artefact. At a 0.10% target the taker round trip of 0.11% exceeds the entire move you are trying to capture, so there is no win rate that produces a positive expectancy. The strategy is arithmetically dead before execution quality, slippage or discipline enter the picture.

Why this is worse than it looks

Read the table by column and the pattern is that fees matter most exactly where traders assume they matter least — at small targets and high frequency, which is the profile of most systematic intraday strategies.

A trader running a 0.30% target with a 62% win rate has what looks like a strong edge. Fee-free, that is a comfortably profitable system. At taker fees it is losing money, and the equity curve will take months to make that obvious because the drift is small per trade and the variance is not.

This is the same mechanism behind a backtest that survives the walk-forward and dies live: if the simulation charged fees only on winners, or charged one fill instead of two, the live result will be worse by a margin that grows with trade count. We ran the same arithmetic over a real trade log in 50,000 arb bot trades.

Two levers, and what each is worth

Take the 0.30% target row. Break-even is 68.3% at taker, 56.7% at maker. Moving fills to the maker side buys back 11.6 percentage points of required win rate — more than any realistic improvement to the signal itself.

The second lever is the rate. Break-even at 0.30% is a function of F, so anything that lowers the effective round-trip cost moves the required win rate down along the same curve. Volume tiers do that, and are unreachable for most accounts. A rebate does it without a volume threshold: what a trading rebate is, and the terms for Bybit accounts are on our Bybit page.

Check your own numbers

Take your median gross target and your actual round-trip fee — the executed one, from the fee column of your trade history, not the schedule you assume applies. Run (T + F) / (2T), or drop your own volume, taker share and target into the fee calculator and let it compute both the break-even and the rebate-adjusted version. Compare to your realised win rate over the last few hundred trades.

If the two are within a few points of each other, you do not have a signal problem. You have a cost problem, and it is the cheaper of the two to fix.