The same strategy: you made $311 gross, and paid $2,983 in fees
Same market conditions, same $10,000 principal: MACD strategy gross profit +$311, fees $2,983—fees are 9.6 times the gross profit. Your net value wasn’t killed by market moves; it was eaten by the fee rate.
First, let’s lay out the backtest settings clearly: BTC/USDT, 1 hour, from 2024-01-16 to 2026-08-22, for a total of 22,786 candles. For each trade, use 10% of equity; charge 0.1% fees on both sides (buy and sell), and use ATR for a stop-loss.
Holding the same period (HODL) is +81.19%, and $10,000 becomes $18,119.
The six strategies’ statements, sorted by number of trades:
• Turtle trading: 70 trades, fees $142, net +$55 (the only positive result among the six)
• Grid trading: 118 trades, fees $237, net −$381
• ICT/SMC: 188 trades, fees $364, net −$495
• AI trend (simulation): 294 trades, fees $573, net −$598
• RSI mean reversion: 1,267 trades, fees $2,255, net −$2,181
• MACD golden cross & death cross: 1,714 trades, fees $2,983, net −$2,671
The pattern is crystal clear: fees and number of trades are almost linearly related, and they have nothing to do with returns. RSI win rate is 38%, MACD win rate is 30.5%—both far higher than Turtle’s 12.9%—but the net result is the worst. All the win-rate advantage gets converted into fees by high-frequency trading.
Three checks you can use immediately:
1️⃣ Calculate turnover cost first, then talk about optimizing entries. Per round-trip trade, fees are approximately 0.2% × position ratio; with a 10% position, each trade consumes about 0.02% of principal. For 1,000 trades, that’s 20%.
2️⃣ Put the fee rate into the backtest code—don’t deduct it afterward. Deducing it afterward only gives you self-comforting conclusions like: “The strategy is good, it’s just that fees are high.”
3️⃣ Add an accounting identity assertion in the backtest: gross profit − fees must equal the change in ending equity. If it doesn’t match, your backtest is doing self-deception.
There’s one more variable not accounted for this time: slippage. All the numbers above only include fees. In real order books, the execution price of market orders will differ slightly from the backtest price; the higher the frequency, the more you get eaten—so the real-world negatives will be even uglier than what’s shown here.
Add one more item to the checklist: print “gross profit” and “fees” as two separate columns for each trade, then see which column is larger. If the fees column is larger than gross profit, what you need to optimize is frequency—not entries.
Reducing trade frequency is the cheapest way to improve net returns; it matters far more than tuning parameters.
How many trades does your strategy execute in a year? Drop the number in the comments and I’ll estimate how much you’ll lose to fees.
$BTC #crypto
Same market conditions, same $10,000 principal: MACD strategy gross profit +$311, fees $2,983—fees are 9.6 times the gross profit. Your net value wasn’t killed by market moves; it was eaten by the fee rate.
First, let’s lay out the backtest settings clearly: BTC/USDT, 1 hour, from 2024-01-16 to 2026-08-22, for a total of 22,786 candles. For each trade, use 10% of equity; charge 0.1% fees on both sides (buy and sell), and use ATR for a stop-loss.
Holding the same period (HODL) is +81.19%, and $10,000 becomes $18,119.
The six strategies’ statements, sorted by number of trades:
• Turtle trading: 70 trades, fees $142, net +$55 (the only positive result among the six)
• Grid trading: 118 trades, fees $237, net −$381
• ICT/SMC: 188 trades, fees $364, net −$495
• AI trend (simulation): 294 trades, fees $573, net −$598
• RSI mean reversion: 1,267 trades, fees $2,255, net −$2,181
• MACD golden cross & death cross: 1,714 trades, fees $2,983, net −$2,671
The pattern is crystal clear: fees and number of trades are almost linearly related, and they have nothing to do with returns. RSI win rate is 38%, MACD win rate is 30.5%—both far higher than Turtle’s 12.9%—but the net result is the worst. All the win-rate advantage gets converted into fees by high-frequency trading.
Three checks you can use immediately:
1️⃣ Calculate turnover cost first, then talk about optimizing entries. Per round-trip trade, fees are approximately 0.2% × position ratio; with a 10% position, each trade consumes about 0.02% of principal. For 1,000 trades, that’s 20%.
2️⃣ Put the fee rate into the backtest code—don’t deduct it afterward. Deducing it afterward only gives you self-comforting conclusions like: “The strategy is good, it’s just that fees are high.”
3️⃣ Add an accounting identity assertion in the backtest: gross profit − fees must equal the change in ending equity. If it doesn’t match, your backtest is doing self-deception.
There’s one more variable not accounted for this time: slippage. All the numbers above only include fees. In real order books, the execution price of market orders will differ slightly from the backtest price; the higher the frequency, the more you get eaten—so the real-world negatives will be even uglier than what’s shown here.
Add one more item to the checklist: print “gross profit” and “fees” as two separate columns for each trade, then see which column is larger. If the fees column is larger than gross profit, what you need to optimize is frequency—not entries.
Reducing trade frequency is the cheapest way to improve net returns; it matters far more than tuning parameters.
How many trades does your strategy execute in a year? Drop the number in the comments and I’ll estimate how much you’ll lose to fees.
$BTC #crypto