Backtest Truth, Part 3: Why Every Trade Needs a Stop
Tests with no stop showed drawdowns of about 35% to 48%. One rule held a single open trade for 355 days and ignored 21 opposite signals.
On this page
This is education only and not financial advice. Trading involves risk, and you can lose money.
A stop loss is a planned exit if the trade goes against you. It is not a prediction that you will be wrong. It is a limit on how much one idea is allowed to hurt you. Part 3 is what the tests showed when that limit was removed.
A trend rule that never had a stop
In one test, a 4-hour trend rule stayed in the market at all times. It flipped only when the averages crossed, and it had no stop. It made a modest gain, but the drawdown was about 35%. “Modest gain” with a 35% drawdown means the path down was large even though the ending number was not a total loss. For a beginner, the path is the part that empties the account before the ending number arrives.
The same family of rules, with a stop, a trailing exit, and 2% risk, had a smaller drawdown: about 18% in 1 year and 24% in 2 years. A trailing exit is a stop that moves as the trade goes your way, so a winner is not left without a planned exit either. The comparison is not “stops make you rich.” It is “the version with a stop and a small risk cap had a smaller drawdown than the version that stayed in the market with no stop.”
Part 5 explains what 2% risk means. Here it is enough to see that adding a stop and that risk cap changed the depth of the drop.
The rule that would only exit in profit
A stricter version was worse. This rule also had no stop. It would close and flip only if the exit would lock in a profit after fees. If the new signal was not yet a profit, the cross was ignored and the position was held.
Over 1 year it lost about 37%, with a 48% drawdown. It closed only one trade, then opened a long in late November 2025 that was still open, and still not in profit, 355 days later. The test ignored 21 opposite signals. The worst open loss was about 36%.
Read the shape of that, not as a tip, but as a warning about the rule. An opposite signal is the test saying “the averages now point the other way.” This rule heard that 21 times and stayed, because leaving would not yet lock a profit after fees. The one closed trade could look neat. The open trade sat there 355 days, with a worst open loss of about 36%.
Over 2 years the same stuck long appeared after a few closed winners, with 38 ignored signals and a drawdown of about 46%. More time did not fix the habit. It gave the rule more signals to ignore.
A 100% win rate that was a trap
The closed-trade win rate was 100%. That number is a trap. The rule only counted exits that were already profits, and it refused to count the large open loss. A report can look perfect while the account is stuck.
Win rate is the share of closed trades that made money. If you only close the winners, the share becomes 100% by construction. The loser is still there. It is just not allowed into the report. Anyone reading “100% wins” without asking “what is still open?” is reading a filter, not a result.
Those 36–48% drawdowns were in a conservative all-in, one-times test. The same path would have been far more dangerous with leverage. Leverage means controlling a larger position than the cash you put up. This series does not walk through leverage settings. It only records that a drop of that size, already harsh at one-times size, would have been far more dangerous if the position had been larger than the cash.
What a stop is for
A stop does not guarantee a profit. It keeps one bad hold from becoming a year-long problem, and it gives you a clear risk size before you enter. Without that line, a test can stay in a trade for 355 days, ignore 21 or 38 signals, and still print a perfect win rate on the trades it agreed to close.
If a test has no stop, the drawdown is the number to fear, not the headline return. In these runs that drawdown was about 35% on the always-in 4-hour rule, about 48% and about 46% on the profit-only exit rule, beside open losses of about 36%. Part 4 asks a different question: even when a window looks excellent, did you only look at the dates you already liked?
Key takeaways
- A stop loss limits the damage of being wrong. It is not a claim that you will be right.
- “Exit only with profit” can trap a trade for months and make the win rate look fake.
- If a test has no stop, the drawdown, not the headline return, is the number to fear.
FAQ
What drawdown did the no-stop 4-hour rule show?
It made a modest gain, but the drawdown was about 35%. The same family of rules with a stop, a trailing exit, and 2% risk had a smaller drawdown: about 18% in 1 year and 24% in 2 years.
What happened when a rule would only exit in profit?
Over 1 year it lost about 37%, with a 48% drawdown. It closed only one trade, then held a long from late November 2025 that was still open and not in profit 355 days later. It ignored 21 opposite signals. The worst open loss was about 36%.
Why is a 100% win rate a trap here?
The rule only closed trades that were already profits, so the closed-trade win rate was 100%. It refused to count the large open loss. The report looked perfect while the account was stuck.
Were these tests using leverage?
The 36–48% drawdowns were in a conservative all-in, one-times test. The same path would have been far more dangerous with leverage. A stop does not promise a profit. It limits how long one bad hold can last.
This is education only and not financial advice. Trading involves risk, and you can lose money.