1h vs 4h: Is Crypto Day Trading Worth It After Fees? (A Backtest)
Faster charts feel like more opportunity: more setups, more trades, more chances to be right. So we ran the same crypto signal engine on the 1-hour and the 4-hour chart, with trading fees included, to see which one actually holds up. The answer surprised even us. On 1h, every version we tried lost money after fees. On 4h, the same rules kept a small, real edge. Here are the numbers, the reason behind them, and what they mean for the timeframe you trade.
One note before the numbers: this is our own engine, tested honestly and published with its weak spots. The results are hypothetical (simulated on past data, not real trades), and the edge we found on 4h is modest. If you're looking for a magic timeframe, this isn't that article. If you want to know why your fast-chart trading might be quietly losing to fees, keep reading.
How we tested it
We ported the exact logic of the ChartVerdict engine (the chart read, the setup score, the stop and the three targets) into a backtest and ran it on real Binance data:
- Coins: BTC, ETH, SOL, BNB and XRP.
- 4-hour data: November 2024 to September 2026, about 4,000 candles per coin.
- 1-hour data: January to September 2026, about 6,000 candles per coin.
- Fees: 0.2% round trip on every trade.
- No peeking: closed candles only, one open trade per coin at a time.
- Pessimistic on ties: if the stop and a target were both hit inside the same candle, we counted the stop first.
Results are in R, which means multiples of the amount you risked. An average of +0.10R means that, over many trades, you made back about 10% of your risk per trade on top of breaking even. Zero is breakeven. Anything below zero is a slow leak.
The results
Three things stand out.
1. On 1h, nothing survived fees. We tried 16 different entry filters on the 1-hour chart: higher score thresholds, trend filters, higher-timeframe agreement, combinations of them. Every single one ended between −0.05R and −0.14R per trade after fees. Some looked fine before fees. None did after.
2. On 4h, the same rules stayed positive. Taking every 4h signal and holding to the final target averaged +0.09R across 616 trades. That's small, but it's above zero after costs, which most "strategies" quietly aren't.
3. Filtering helped on 4h, not on 1h. On 4h, only taking setups that scored 60 or more and where the higher timeframe agreed lifted the result to about +0.12R. It stayed positive on all five coins and in both halves of the data. On 1h, the same filter just lost money more slowly.
| 4h entry filter (25% at TP1, 75% to TP3) | Trades | Win rate | Avg R after fees | Coins positive |
|---|---|---|---|---|
| Every signal | 616 | 51% | +0.06 | 3 of 5 |
| Score 60+ | 594 | 53% | +0.11 | 5 of 5 |
| Score 60+ and higher timeframe agrees | 491 | 53% | +0.12 | 5 of 5 |
| Score 60+ and all three timeframes agree | 455 | 54% | +0.10 | 4 of 5 |
Notice the win rates: low 50s. A coin flip, almost. The edge doesn't come from being right more often. It comes from the winners being a little bigger than the losers after costs, which is exactly the kind of edge fees can erase.
Why the 1-hour chart loses to fees
Fees are charged on the size of your position, but your result is measured against your stop. On a faster chart, stops sit closer to entry, so the same fee eats a much bigger slice of each trade's risk.
A simple illustration (round numbers, not taken from the backtest): say fees cost 0.2% of the position per round trip.
- On a 1h setup with a stop about 1% away, the fee equals 0.2R: a fifth of your risk is gone before the trade moves.
- On a 4h setup with a stop about 2.5% away, the same fee is about 0.08R.
Fee in R = fee % ÷ stop distance %. Halve your stop distance and you double what fees cost you per trade, while the edge per trade usually doesn't double with it.
Faster charts also trade more often, so a small negative number repeats more times per month. And shorter candles carry more noise relative to the move you're trying to catch, so stops get tagged by randomness more often. Put those together and a setup that's slightly positive on 4h can be slightly negative on 1h, with no change in the rules at all.
What this means for your trading
If you trade 1h or faster
You're not necessarily doing it wrong, but you need a bigger edge than the fees you pay. Check your real fee rate (maker vs taker, VIP tier, and funding if you hold perpetuals). Then work out your average stop distance and see what fees cost you in R. Many traders find their "break-even" system was really a slow loss. Using 1h for timing an entry inside a 4h setup is a different thing, and often a good one.
If you trade 4h
The edge is real but small, so it lives or dies on discipline. Size every trade from the stop, skip setups that don't meet the bar, and expect long stretches that look like a coin flip. In our test, banking 25% at the first target and leaving the stop where it was beat moving the stop to breakeven. Breakeven stops got knocked out by normal 4h noise before the bigger targets were reached. (More on that in risk-reward ratio.)
Whatever you trade
Count fees as part of the plan, not an afterthought. A strategy is only as good as its result after costs.
The limits of this test
- Crypto only. Five large coins, one exchange's data. Forex and stocks have different costs and behavior.
- Different periods. The 1h data covers January to September 2026; the 4h data covers November 2024 to September 2026. A longer 1h sample could look different.
- Simulated fills. Entries at our levels were counted when touched, which is slightly optimistic. Real slippage and partial fills aren't modeled.
- Small edges are fragile. +0.1R per trade can disappear with higher fees, worse execution, or a market that behaves differently from this period.
- Past results never guarantee future results. Treat this as evidence about fees and timeframes, not a promise about any strategy.
What we changed because of it
We built this backtest to test our own tool, and we changed the tool because of what it found. ChartVerdict now recommends 4h, labels 1h as lower-edge after fees, and only shows a trade when the setup scores 60+ and the higher timeframe agrees. Anything just below the bar shows as Near Miss, and the rest as Stand Aside. That means it tells you not to trade most of the time, which is the point. We also switched the exit plan to banking 25% at TP1 with the stop unchanged, then running the rest to TP3.
Key takeaways
- Same engine, same coins, fees included: 1h lost money (−0.05R to −0.14R per trade); 4h kept a small edge (about +0.1R).
- Fees cost more on faster charts because stops are tighter: fee in R = fee % ÷ stop distance %.
- The 4h edge came with a win rate in the low 50s, so it depends on discipline and sizing, not on being right.
- These are hypothetical results on one market period. Test your own costs before trusting any timeframe.
Check any coin on 4h in one tap.
ChartVerdict runs this engine on live candles and only shows a trade when it passes the same filter: score 60+ and the higher timeframe agreeing. Sign in free and your first 7 days are full Pro: unlimited Auto-Reads and the Setup Scanner across 28 coins + Gold.
⚡ Try ChartVerdict FreeEducational content only. Not financial advice. Backtest figures are hypothetical — simulated on past data, not real trades — and have inherent limitations, including the absence of real slippage and execution. Trading crypto, especially with leverage, involves substantial risk of loss and is not suitable for everyone. Past performance never guarantees future results. Trade only with money you can afford to lose.