Stop-loss placement: nine rules tested
Nine Bitcoin stop rules returned from -0.42% to +278.97% in one simulation. See how stop-based sizing and fills shaped the comparison.
This August 2026 simulation applied nine stop rules to one BTCUSDT 4-hour EMA signal stream. Recorded returns ranged from −0.42% to +278.97%, and maximum drawdown ranged from 12.4% to 70.6%. The entry logic and intended 2% risk budget stayed fixed, but stop-based position sizing, exits, fill prices, fees, and final trade counts changed with the stop.
That makes this an experiment about the combined stop-and-sizing rule, not a claim that moving one line on a chart independently caused the return spread.
The protocol
- Strategy: EMA 20/50 crossover on BTCUSDT, 4-hour candles. Long on the cross up, short on the cross down, and an opposite cross always closes the trade if the stop hasn't already.
- Window: April 1, 2020 to August 1, 2026.
- Sizing: 2% of equity risked per trade, sized off the stop distance. This detail turns out to run the whole experiment, more on it below.
- Costs: the artifact's “PRODUCT” model: intra-bar fills on 1-minute data, order-book fills where we have book coverage, dynamic slippage, funding.
- The nine stops: fixed at 1%, 2.5%, 5%, 10%, and 20% from entry; at 1.5× and 3× ATR; a 2.5% stop that trails; and a 2.5% stop that jumps to break-even once the trade is +5%.
Every run produced 258 to 260 round trips from the same signal-generation rule. The artifact does not attribute the small count differences to one cause.
| Method field | Historical record |
|---|---|
| Starting capital | $10,000 |
| Engine revision | fa5c5f10, labelled “lookback-invariant ema/rsi” |
| Exact venue | Unknown in the result artifact |
| Exact fee, dynamic-slippage, and L2-coverage values | Unknown in the result artifact |
| Selection | All nine configurations compared on the same full window |
| Final untouched OOS period | None recorded |
CME's position-sizing lesson explains why stop distance and account risk jointly determine quantity. Investor.gov also notes that a stop price is a trigger, not a guaranteed fill. Those mechanics are distinct from the historical performance values below.
The results
| Stop | Return | Max drawdown | Sharpe | Win rate | Stopped out |
|---|---|---|---|---|---|
| Fixed 1% | +91.02% | 70.6% | 0.42 | 13.1% | 85% of trades |
| Fixed 2.5% | +166.99% | 41.3% | 0.51 | 22.4% | 59% |
| Fixed 5% | +12.44% | 40.0% | 0.19 | 25.2% | 28% |
| Fixed 10% | -0.42% | 29.8% | 0.06 | 26.4% | 4% |
| Fixed 20% | +6.15% | 12.4% | 0.17 | 26.7% | 0% |
| 1.5× ATR | +218.81% | 52.1% | 0.51 | 20.5% | 69% |
| 3× ATR | +95.92% | 37.0% | 0.43 | 25.6% | 28% |
| 2.5% trailing | +7.96% | 19.4% | 0.16 | 37.3% | 94% |
| 2.5% + break-even at +5% | +278.97% | 37.9% | 0.61 | 22.0% | 66% |
The rows challenge several simple rules of thumb, while remaining one in-sample historical comparison.
Tight isn't safe and wide isn't rich
The intuition says a tight stop protects you and a wide stop gives the trade room. The table says the relationship isn't even monotone.
The 1% stop got stopped out on 85% of its trades and won only 13% of them, yet still made +91% while recording 70.6% drawdown. The 10% stop made approximately nothing over six years. And the 20% stop, wide enough that it never fired once, beat the 10% stop while drawing down only 12.4%.
One mechanism is position sizing. Before leverage or exposure caps, risking 2% of equity across a 1% stop produces a position ten times the quantity of risking the same amount across a 10% stop. Tight stops therefore increase size and can exit more often, while wider stops reduce size. The aggregate artifact does not isolate that mechanism from the changed exits, fees, and fills.
For this signal and sizing rule, stop distance changed exposure as well as exit placement. The table cannot identify a universally best distance, and it should not be used to select 2.5% for another strategy without a separate test.
Win rate can mislead
Look at the trailing stop. It has the best win rate in the table, 37.3%, and one of the worst returns, +7.96%.
A trailing stop ratchets toward price as the trade moves in its favor. In this run, 94% of exits were recorded as stop exits. The result is consistent with the trail cutting winners during pullbacks, but the aggregate artifact does not preserve each path needed to prove that mechanism trade by trade.
Meanwhile the 1% stop won 13% of the time and recorded a much higher return. Win rate measures how often you're right, not how much being right is worth, so it cannot rank these exit rules on its own.
What the ATR stops add
A fixed percent does not adapt to changing 4-hour volatility. An ATR-based stop scales its distance with the recent true range, becoming tighter in quieter samples and wider when that measure rises.
In this table, 1.5× ATR returned +218.81%, the highest return among the plain fixed or ATR stops, with a 0.51 Sharpe and 52.1% drawdown. The 3× ATR row returned +95.92% with 37.0% drawdown. The result shows a trade-off in this sample; it does not establish that the multiple caused the difference or will transfer to another period.
The highest-return configuration
The highest-return row adds a rule: start with the plain 2.5% stop, and once the trade is up 5%, move the stop to the entry price. Nothing else changes. +278.97%, drawdown 37.9%, and the best Sharpe and profit factor in the table.
Unlike the trailing stop, the break-even rule moves once and then leaves the stop at entry. A companion detail artifact records 42 of its 172 stop exits within 0.5% of break-even. That is consistent with converting some reversals from full stop losses into scratches, while fees and slippage remain.
Only one break-even trigger was tested, on the same development window used to compare every row. The historical prop-firm EMA study used the same mechanic, but a later recorded engine check did not reproduce its original trade count. The two dated simulations are reasons to test the rule, not independent confirmation of a live effect.
What to do with this
Report sizing with the stop. Under risk-based sizing, stop distance helps set position size. An exit comparison that omits this coupling leaves out a major reason its equity curves differ.
Compare fixed and volatility-scaled rules deliberately. ATR adapts distance to its lookback's measured range; a fixed percentage does not. Neither is a forecast of the next move.
Do not evaluate an exit rule by win rate alone. Read return, drawdown, and what fraction of exits the stop is taking. In this table, the 94%-stop-exit row also had the highest win rate and one of the lowest returns.
Treat break-even as another parameterized rule. The one tested trigger had the highest return here, but it still needs prespecified evaluation on later data and sensitivity checks around the trigger.
Honest caveats
One strategy family, asset, window, and sizing rule cannot establish how the same stops behave for another system. The +5% break-even trigger is one tested value. Every row was inspected on the same full period, with no recorded final untouched OOS test, so the highest row deserves the same overfitting scrutiny as any selected maximum.
No stop rule here is a finished strategy recommendation. The evidence supports the observed range under this historical protocol and the coupling between distance and size. AlphaProve's public risk-management reference defines the sizing and stop settings needed to reproduce the design on a new period before drawing a new conclusion.