Published

Every losing scalper we've ever looked at — including our own paper bots — dies the same quiet way. Not a blow-up, not a bad call on the market. They die by a thousand 0.05% cuts. Fees are the one cost that's certain, pays out on every single trade win or lose, and scales with how often you trade. Scalping is the strategy that trades the most. So let's stop hand-waving and do the math on a single round trip, then multiply it by a normal scalping day.

What one round trip actually costs

On a major futures exchange the taker fee is roughly 0.055% per side (Bybit and Binance USDT-perps both sit around there; some are 0.045–0.06%). A scalper crossing the spread to get filled pays taker on the way in and the way out:

That 0.11% is the wall every trade starts behind. Before the market moves a single tick in your favour, you are already down 0.11% of the position. To net nothing, price has to move 0.11% your way just to refund the house.

Leverage doesn't reduce the fee — it magnifies it

Here's the part that catches people. Fees are charged on notional (position size), not on your margin. Leverage inflates notional, so it inflates the fee relative to the money you actually put up. Say you post $100 of margin at 20x — that's $2,000 of notional:

So a "free" little scalp at 20x quietly costs 2.2% of your stake every time you open and close. Crank it to 50x and the same round trip eats 5.5% of margin. The leverage slider you thought was for upside is also a fee multiplier — the same point our 18-year leverage study made about volatility drag, except this drain is guaranteed, not probabilistic.

The break-even win rate fees force on you

This is the number that explains why so many "high win rate" scalpers still lose. Suppose you scalp for a 0.3% gross move each way — a typical small target. Subtract the 0.11% round-trip fee:

Your wins are smaller than your losses purely because of fees, even when the raw move is symmetric. Plug that into the break-even win rate formula — loss ÷ (win + loss):

0.41 ÷ (0.19 + 0.41) = 68.3%

You need to win 68 out of every 100 trades just to break even — and that's before funding. This is exactly why a 60% win rate quietly bleeds an account: it sounds like winning, but it's eight points under the line fees drew. (For the position-sizing side of why 60% still blows up, see our risk-of-ruin report.)

Now multiply by a day

A scalper doing 10 round trips a day at 20x pays 10 × 2.2% = 22% of margin in fees, every day, regardless of whether they won or lost. Over 20 trading days that's a notional fee bill of 440% of your margin churned through the exchange. You don't need a market crash to lose money at that pace — you just need to keep trading. The market is almost a spectator; the fee meter is the main event.

This matches what we found the hard way running thousands of scalping configs across two years of data: a high win rate looked great until fees were applied, and then fees, not the market, decided the result. The arithmetic above is why — it was never going to be otherwise.

What the math says to do instead

Three things fall straight out of the numbers:

None of this is exotic. It's the toll booth math, run once, cold. Most people never run it — and the calculators below let you run it on your own numbers in about thirty seconds.

Method: fees illustrated at 0.055% taker per side (round trip 0.11%), representative of Bybit/Binance USDT-perp taker tiers as of June 2026; your tier and any maker rebates differ. Funding is excluded and only makes the picture worse for held positions. Figures are arithmetic, not a backtest — but the relationship between fees, leverage and break-even win rate is exact.

Check your own numbers
Trade where the fees are lower
Trade on Bybit — lower fees, free to join → Binance → OKX → KuCoin → 📈 TradingView 🔒 NordVPN