RealPnL
DEMO Binance account / since start -0.46% / 4 open / last trade: TRIM SOL long · -0.10% 9 h ago / updated 09:25 UTC

Research ·

How much do fees cost a trend bot? The same rules at six cost levels

About 688 orders a year, trading 45 times the account's value. Every 0.01% of cost per trade takes roughly half a point a year off the return, and at 0.25% per trade the weak year 2025 goes negative.

A backtest can pick any cost per trade it likes. Ours uses 0.07%: Binance's standard taker fee of 0.05% plus 0.02% for slippage. Here's what happens if that assumption is wrong, and how much of it you can control.

How much the bot trades

In the backtest of the live rules, January 2021 to September 2026, the bot sent 3,944 orders, about 688 a year:

Kind of orderCountTypical size (% of account)
New position80710.1%
Close7926.9%
Trim (make smaller)1,3163.7%
Add (make bigger)1,0292.7%

Stop-loss exits are extra; they're orders on the exchange, not sent at the 4-hour close.

Added up, the orders trade about 45 times the account's value per year. Most of that isn't new trades: it's resizing. As prices and volatility move, the bot keeps each position near its target size. At 0.07% per unit traded, all of this costs roughly 3% of the account a year.

Six cost levels

0%: +50%+50%0%: +10%+10%0%0.04%: +48%+48%0.04%: +8%+8%0.04%0.07%*: +46%+46%0.07%*: +7%+7%0.07%*0.10%: +44%+44%0.10%: +6%+6%0.10%0.15%: +41%+41%0.15%: +3%+3%0.15%0.25%: +34%+34%0.25%: -1%-1%0.25%
Return per year, 2021–20262025 alone
Cost per unit tradedReturn per yearSharpeWorst drop2025
0% (impossible)50.4%1.80−17.7%10.2%
0.04% (maker fees, no slippage)47.7%1.72−18.1%8.4%
0.07% (what we assume)45.7%1.67−18.5%7.0%
0.10%43.8%1.61−19.0%5.6%
0.15%40.6%1.52−20.1%3.4%
0.25%34.4%1.33−22.8%−0.9%

Hypothetical backtest of the live rules, funding not included (funding is here).

The rule of thumb: every 0.01% of cost per trade costs about half a point of return a year. Over six years that's modest. In a thin year it decides everything: at 0.25% per trade, 2025 goes from a small gain to a loss.

Is 0.07% realistic?

So far, yes. On Binance's demo platform the bot's fills landed within ±5 basis points (0.05%) of the real market price at the same minute, and the one new position in that window filled 74% as a maker order. Exits were the problem: waiting for a maker fill on an exit cost 9–22 basis points, so exits now go straight to the market. The full measurement.

What you control if you copy

The bot's own lever on costs is how often it resizes: it only resizes a position when the target moves more than 30% away from what's held. We tested wider gaps and decided the savings weren't worth it.