“Without a stop” sounds like a binary choice: either accept a fixed loss or let a position breathe. In an algorithm, it is more precise to ask which exit mechanism is doing the work. Does the position leave because price touched an ATR level, because the trend regime changed, because a time limit expired, or because the account could no longer carry the exposure?

This article tests that question on a real research EA. The numbers below are not a hypothetical illustration: they come from two fresh MT5 control replays. The only deliberate change is the trade-management layer.

The question tested Does the stop protect the strategy — or does it cut off the strategy’s winners before the regime has finished? The answer depends on the entire distribution: return, equity drawdown, trade count, holding time, costs and the size of the worst individual loss.

The advisor and the test protocol

The advisor is RegimeBreak 5-20-60. It estimates a daily direction from three momentum lookbacks, waits for an H4 breakout and sizes XAUUSD positions from an ATR-based distance. The test is long-only because that is the side used in the existing gold research on the site.

Instrument / period XAUUSD · H4

1 January 2024 through 31 December 2025. Starting balance: $100,000. Direction: long only.

Signal logic D1 5 / 20 / 60 + H4 breakout 12

Two of three momentum votes are required. Entry is a closed H4 breakout inside a valid long regime.

Risk basis 0.50% · ATR(14) × 1.8

Both variants use the same ATR distance to calculate volume. This isolates exit behaviour from a completely different position-sizing rule.

Deliberate change No initial broker SL/TP

The no-stop branch disables the initial SL, fixed TP, break-even and trailing. Opposite-regime exit and the maximum-holding input remain enabled.

Both replays used MT5’s 1 Minute OHLC model in the clean comparison. That matters: the existing published RegimeBreak page uses a different tick model and therefore its $14,232.03 result should not be mixed into this pairwise experiment. Same-model comparisons are the relevant evidence here.

The result at a glance

With stops+$13,941.54
Without initial SL/TP+$16,002.87
Profit difference+$2,061.33
Equity DD: stops2.52%
Equity DD: no initial SL/TP7.04%
Drawdown increase+4.52 pp
Controlled comparison of profit and maximum equity drawdown The no-initial-stop variant has a slightly higher net profit but nearly three times the maximum equity drawdown. NET PROFIT · USD16k12k8k0 $13.94k$16.00kstopsno initial SL/TP MAX EQUITY DRAWDOWN · %8%6%4%0 2.52%7.04%stopsno initial SL/TP
The headline return favours the no-stop branch, but the risk chart tells the more important prop-firm story: the same $100,000 account spends much more time and capital underwater.
MetricWith stopsNo initial SL/TPWhat changed
Net profit+$13,941.54+$16,002.87+$2,061.33
Return on $100k13.94%16.00%+2.06 pp
Closed trades8641−52.3%
Win rate50.0%53.7%Not the main story
Profit factor1.842.22Fewer, larger outcomes
Maximum equity DD2.52%7.04%+4.52 pp
Average holding time45h 04m170h 55m3.8× longer
Largest losing trade−$597.40−$1,944.023.3× larger
Largest winning trade+$828.12+$2,589.953.1× larger
Maximum loss streak66Same observed streak

The profit factor, trade count, holding time and individual trade values are calculated from the two exported deal logs, including commission and swap. Maximum equity drawdown is measured from the advisor’s exported equity snapshots. The raw control logs are kept with the project data: control deals and no-initial-SL/TP deals.

One trade shows why the average is misleading

Take the first long position in both control logs. The entry is identical: 4 March 2024 at 16:00, 0.28 lots of XAUUSD at 2091.73. After that, the exits tell two very different stories.

With stops

ATR target at 2118.38 is reached on the same day, after roughly 2 hours 39 minutes.

Net: +$744.52
Profit $746.20 less entry/exit commission.

Without initial SL/TP

The position stays open for seven days and exits at 2177.13 through the advisor’s non-price exit path.

Net: +$2,278.50
Profit $2,391.20 less commission and $111.02 swap.

That single example makes the no-stop result look attractive. The opposite side appears in the worst trade: a no-stop position held from 22 July to 29 July 2025 lost $1,944.02 after costs. In the stopped control run, the largest losing trade was $597.40. The no-stop variant is not “the same strategy with a better exit”; it is a strategy with a wider outcome distribution.

What actually changed in the trade distribution?

Monthly equity snapshots for the controlled replays The no-initial-stop curve finishes higher but has deeper mid-period setbacks and a more irregular path. EQUITY SNAPSHOTS · $100K START$118k$109k$100k with stopsno initial SL/TP
Monthly last-observation equity snapshots from the two control logs. The no-stop line finishes higher, but its setbacks are deeper and its path is less forgiving.

1. The stop is also a signal-frequency control

With a stop and a target, a position leaves quickly whenever the market invalidates the entry or reaches the intended reward. That frees the advisor to evaluate another breakout. Without the initial SL/TP, a position remains in the book until a regime or time exit arrives. The number of trades fell from 86 to 41, even though the signal logic was unchanged.

This is why comparing only the final profit is dangerous. The no-stop version did not simply improve every trade. It traded less often, held longer and concentrated more of the result in a few large winners and losers.

2. Position sizing does not cap the loss

In this experiment, the ATR distance was retained for volume calculation. At entry, the position size still represented roughly 0.50% risk to that hypothetical distance. But because no broker stop was placed, that number was no longer a hard loss limit. It was only a sizing reference.

Important distinction Risk used for sizing ≠ risk guaranteed by the exit A 0.50% sizing rule can produce a 1.00%, 2.00% or larger realised loss when the position has no price-based stop and the market moves against it before a time or regime exit.

3. Swap becomes part of the exit decision

The first no-stop winner paid $111.02 in swap while it was held for seven days. That cost did not erase the trade, but it changed the economics. On a less favourable symbol, broker or holding period, carrying cost can turn the “let it run” idea into a slow leak. Stop-loss research should therefore include swap, not only the price difference between entry and exit.

What this means for a prop account

The no-stop return looks better in an unconstrained personal-account ranking. For a prop challenge, the drawdown and path usually matter more than the extra 2.06 percentage points of final return.

  • The stopped control used 2.52% maximum equity drawdown, leaving a much larger buffer to a typical 5–6% account limit.
  • The no-initial-stop control reached 7.04% maximum equity drawdown in the exported equity observations — already beyond many challenge limits.
  • The no-stop worst closed loss was $1,944.02, more than three times the stopped worst loss.
  • Seven-day holdings create exposure to rollover, weekend gaps, swap changes and a longer period in which a prop rule can be breached.

There is one more practical caution: a broker stop can fail to cap risk perfectly during a gap or a fast market. That is a reason to stress-test execution, not a reason to pretend that the absence of a stop removes risk. The no-stop result demonstrates the opposite: the loss limit becomes less explicit and the account carries more path risk.

Case study 2: AAPL 3days_down

The second experiment asks the same question with a very different strategy. The signal is a simple daily mean-reversion rule: buy AAPL after three consecutive lower closes, then exit when the completed close is above the previous day’s high. The only variable is the initial fixed stop at 0%, 1%, 2%, 3%, 4% or 5% below entry.

Instrument / period AAPL.NAS · D1

1 January 2022 through 22 April 2026. Starting balance: $10,000. Long only.

Position model 100% notional allocation

One position at a time, sized to the full current balance as notional exposure. Maximum leverage remains 2×.

Entry / exit 3 lower closes → close > previous high

The daily signal and discretionary exit are identical across all six runs.

Stop grid No stop, then 1%–5%

Stops are real MT5 stop orders below the entry price. The no-stop branch relies only on the daily exit rule.

Same signal, six different trade distributions

The final balances are read directly from MT5. Win rate, profit factor and closed-balance drawdown are derived from the executed deal logs using the observed $0.02/share/deal commission convention. The stop grid is not monotonic: a tighter stop creates more exits and re-entries, but does not automatically reduce account-level drawdown.

VariantFinal balanceReturnTradesWin rate*PF*Closed DD*Stop hits
No stop$13,893.55+38.94%5962.71%2.0118.86%0
1% stop$13,653.15+36.53%10239.22%1.5517.47%58
2% stop$11,503.66+15.04%8447.62%1.2120.49%37
3% stop$13,314.95+33.15%7558.67%1.4623.46%21
4% stop$12,215.00+22.15%7160.56%1.3124.12%16
5% stop$13,232.69+32.33%6563.08%1.5421.98%8

* Final balances are direct MT5 results. Trade count, win rate, profit factor and closed DD are log-derived; swap is not present in the tester logs. The AAPL runs use 100% notional allocation, so these drawdowns should not be read as a recommended prop-account sizing.

AAPL final balance by stop size No stop and a one percent stop finish highest, while the two percent stop is the weakest. Closed-balance drawdown is not monotonic. FINAL BALANCE · $10K STARTNO STOP1%2%3%4%5% $13.89k$13.65k$11.50k$13.31k$12.22k$13.23k DD is not monotonic
The 2% stop is the weakest result in this sample. More protection did not automatically mean less closed-balance drawdown because a stop changes trade frequency and causes the strategy to re-enter during continued weakness.

A concrete path difference: the first AAPL trade

Both the no-stop and 1% stop runs buy 58 shares at 172.391 on 7 January 2022. The no-stop branch exits on 12 January at 176.589 for approximately +$241.16 after estimated commission. The 1% branch is stopped on 10 January at 169.925 for approximately −$145.35, then later re-enters. This is the mechanism behind 102 trades with a 1% stop versus 59 without one.

What this case actually says For AAPL 3days_down, a fixed stop changes trade frequency more clearly than it improves account safety. No stop and 1% stop had the best final balances, but the 17.47%–18.86% closed DD range is still far too large to call 100% allocation prop-safe.

The 2% stop is especially instructive: it looks like a compromise between no stop and a very tight stop, yet it finishes at only $11,503.66 with a 1.21 profit factor. The 3%–5% branches recover much of the return, but their closed drawdowns rise as high as 24.12%. In this strategy, “wider stop” is not a free improvement either.

Putting both studies together

These experiments do not produce one universal rule. They expose two different jobs that a stop can perform.

StrategyWhat no stop didWhat the stop didPractical conclusion
RegimeBreak · XAUUSDProfit +$16,002.87, equity DD 7.04%, worst loss −$1,944.02.Profit +$13,941.54, equity DD 2.52%, worst loss −$597.40.Stops materially reduce path risk and are defensible for a prop-account risk architecture.
3days_down · AAPLBest balance $13,893.55, 59 trades, closed DD 18.86%.1% is close at $13,653.15, but 102 trades; 2%–5% results are unstable and DD remains high.Stops do not rescue 100% allocation. Lower exposure and an account-level breaker are the next controls to test.

The distinction is important. RegimeBreak’s stop is acting as a loss limiter around a breakout system whose no-stop positions can remain open for days and accumulate large adverse excursions. In AAPL 3days_down, the stop often interrupts a mean-reversion trade before the recovery exit, then creates a new entry while weakness continues. It is therefore changing the signal’s effective holding-period distribution, not simply “protecting” the same trade.

Combined conclusion

For RegimeBreak, the stop is worth the return sacrifice in this sample: giving up $2,061.33 of net profit reduces maximum equity drawdown by 4.52 percentage points and cuts the largest losing trade by roughly 69%. For AAPL 3days_down, the data do not justify saying that a fixed stop improves the strategy: no stop and 1% finish best, 2% is worst, and every variant remains exposed to a large drawdown at 100% notional allocation. The correct next test is lower AAPL allocation with a portfolio-level loss limit, while keeping the RegimeBreak stop as the baseline risk control.

A better design question than “stop or no stop”

The useful engineering question is: what should invalidate this trade, and how quickly must the account respond? For this EA, the control results suggest a layered design:

LayerPurposeTestable rule
Initial price stopLimit damage when the breakout fails quickly.ATR distance chosen before looking at the result.
Position sizeKeep the initial loss tolerable for the account.Calculate volume from tick value, contract size and stop distance.
Regime exitLeave when the daily directional premise disappears.Close on the defined opposite-regime condition.
Time exitPrevent capital from being trapped indefinitely.Stress-test 10, 20, 30 and 40 H4 bars.
Portfolio circuit breakerProtect the prop account from correlated damage.Pause or reduce risk before the firm’s hard limit.

A stop can be too tight, too wide or simply mismatched to the strategy’s holding period. But this experiment does not support replacing it with a vague promise that “the trend will eventually recover.” If the strategy needs more room, test a wider stop, smaller position size, volatility-conditioned risk or a different invalidation rule. Measure the full distribution every time.