Most trading advice says the same thing about exits.
Let winners run.
Move your stop to breakeven.
Trail behind price.
I assumed that advice would survive the test.
It didn't.
I ran four exit variants across twelve validated cells.
The trailing-plus-breakeven variant won zero.
On twelve out of twelve cells, a different variant came in higher.
The variant that won outright on seven of twelve cells was the boring one:
a fixed bracket at entry, with a stop two units of ATR below and a target six units above. Set it and walk away.
That default is one of the survivors from last week.
It is a procedure that earned its place by killing the alternatives.
A cell is a (strategy × symbol × timeframe) triple.
Twelve of them cleared the gauntlet I described last week.
I wanted to know whether, within those twelve, the popular exit logic improves them or hurts them.
I built a small gauntlet that runs every cell against four exit configurations and reports a profit factor for each. (Profit factor is the ratio of total gross profit to total gross loss — above 1.0 means the system earns more than it loses; higher is better.)
Two minutes per pass, forty-eight backtests per run.
The four variants:
Default. Stop at 2× ATR below entry. Target at 6× ATR above. No trailing. No breakeven. Bracket placed at fill, walk away.
Trail (with breakeven). Same stop and target, plus: when price reaches +1R unrealized, move the stop to entry; when it reaches +2R, trail the stop at one R behind the close. This is the popular advice in one variant.
Tight target. Stop at 2× ATR. Target at 4× ATR. Same stop, target pulled in.
Wide target. Stop at 2× ATR. Target at 8× ATR. Same stop, target pushed out.
I ran the matrix on the actual cells in the fleet.
Not synthetic data.
Not toy assets.
The same cells whose live forward observation I track every day.
The matrix was unambiguous.
Trail won zero cells outright. On twelve out of twelve cells, a different variant came in higher than the trail-with-breakeven setup. On most cells, the gap was significant: trail hurt the profit factor versus default on nine of twelve cells, with the worst drops landing between half a point and one and a half points (a turtle breakout strategy on silver 4-hour: 2.07 → 0.88; a VWAP mean-reversion strategy on silver 30-minute: 2.02 → 0.88).
Trail did help on three cells, but never enough. On a turtle breakout strategy on platinum hourly, trail lifted profit factor from 2.07 to 2.39, but tight target on the same cell hit 2.53. On a turtle breakout strategy on copper hourly, trail lifted 1.39 to 1.42 while tight target hit 1.52. On a rotation strategy cell, trail lifted 1.39 to 1.50 while wide target hit 1.53. Every cell where trail moved the number forward had a non-trail variant move it further.
Default won seven cells outright. Wide target won three (silver and gold on the 4-hour timeframe, where trends are slower and ride further). Tight target won two (platinum mean-reversion on 1-hour).
I was prepared for the test to show me nothing.
A clean null result would have told me to keep the cells with whatever exit was already locked. Instead the test showed a clean pattern. Most of these cells do not move in a straight line. They move toward target, revisit entry, then complete the move. The breakeven rule mistakes normal behavior for failure. Trailing tightens the noose at the wrong time.
A full-ATR-distance stop survives the oscillation. The winner gets paid.
The advice to trail stops comes from trend-following intuition.
If your edge is price moves further than expected in one direction, trailing makes sense. You are trying to ride a one-way move, and the trail catches you if the move reverses.
If your edge is price moves to a level and reverses, trailing is wrong.
You are betting on the reversal, which means you expect price to come back.
If you tighten your stop when price first goes your way, you exit before the reversal pays.
A trailing stop on a mean-reversion cell turns a winning trade into a break-even or small-loss trade by exiting on the oscillation. Over hundreds of trades, that compounds into a destroyed edge.
The edge existed on paper.
The exit rule killed it.
The exit rule is part of the edge.
It is not a downstream detail.
It is a gate the candidate has to pass before I let it near deployment.
The gauntlet now includes an exit-variant pass.
Every cell promoted into the audit-validated set has to clear the same matrix.
The default may not win for every strategy (trend cells almost certainly want different exits), but the rule must be measured per cell, not inherited from convention.
When a popular framework contradicts the data, the data wins.
I do not get to keep trailing stops because every YouTube channel says to use them.
I get to keep what survives the test.
The graveyard is the asset.
Twelve cells where a non-trailing variant won.
Three where the trail moved the number forward and still lost to a simpler exit.
The procedure is in the matrix.
If you have your own backtests, run the same test.
Pick three to five of your strongest cells.
Run them against four exit configurations: fixed bracket, trail-plus-breakeven, tighter target, and wider target.
Compare profit factor. The pattern will be specific to your edges.
If the pattern matches mine (trail kills, default survives), your edges are probably mean-reversion or range-bound. If trail wins your matrix, your edges are trend-following, and you should keep doing what you are doing.
The point is not that trailing stops are wrong. The point is that the answer is empirical, not received. Most trading advice is received.
It does not claim the default exit is right for every system.
It does not claim trailing stops are useless.
It does not claim zero outright wins for trail generalizes beyond this fleet.
It claims one thing.
On twelve specific cells that cleared the full gauntlet, the popular trailing-plus-breakeven logic won zero of them outright, and the boring fixed bracket won the most. That is a finding, not a doctrine.
The PrimeFold Press publishes the filter.
Each issue shows one gate the candidates have to clear.
This week's gate was exits.
The validation procedure is the product.
The survival rate is the receipt.
— Collin
The PrimeFold Press
One question to carry into Issue #3:
What is the exit rule you have never tested?
