Sizing Down After a Losing Run
A losing run does two things at once. It reduces the account, which mechanically reduces whatever a fixed fraction of it will risk, and it changes the person operating the account, which is not mechanical at all. Sizing can handle the first automatically. The second needs a decision made in advance, because it cannot be made well in the middle of the run.
Percentage Risk Already Reduces
If the risk per trade is a fixed fraction of the account rather than a fixed money amount, size falls as the account falls without anyone doing anything. This is the main argument for percentage risk and it is a good one. The reduction is gradual and proportional, and it never requires a judgement call at the moment when judgement is at its worst.
It is also slow. A modest string of losses barely moves the fraction, so the size after several losing sessions is close to the size before them. That is appropriate if the losses were ordinary variance. It is inadequate if something has changed, and the arithmetic cannot tell the difference between the two cases.
Reducing on Purpose
The deliberate version cuts size further after a defined trigger: a number of consecutive losses, or a drawdown past a threshold set beforehand. The reduction is stepped rather than continuous, and it is reversed on a stated condition rather than when you start feeling better.
What this buys is not statistical. If the rule has an edge, cutting size after losses lowers expected return. What it buys is the ability to keep trading the rule at all. A trader who halves size and continues is still collecting the outcomes the rule produces. A trader who holds size, takes another two losses and stops entirely has left the sample, and a rule cannot work for someone who is no longer there.
Coming Back Up Is the Harder Half
Cutting is easy to specify and easy to do. Restoring is where most schemes quietly fall apart. Reduced size means slower recovery, which means the condition for returning to full size takes longer to satisfy, which invites a shortcut. The shortcut is usually taken after one good day.
The restoration condition should be defined at the same moment as the reduction, and it should be symmetric in kind rather than in feeling. If the trigger to cut was a run of losses, the trigger to restore might be a matching run of ordinary sessions, or the recovery of a stated portion of the drawdown. What it should not be is a single large winner, because a single large winner in a small sample is precisely the event that says least about whether anything has improved.
Variance or a Broken Rule
Sizing down manages the consequence of a losing run without answering what caused it. Both questions deserve attention and they rest on different evidence. A run of losses within the range that a small sample routinely produces is not information. A run where the losses arrive for a reason the rule never anticipated, on session types it was never tested against, is information.
The way to tell them apart is to look at the trades rather than the total. If the losing trades are ordinary trades that went the wrong way, the rule is doing what it does. If they share a feature the rule ignores, such as a market regime that arrived and then stayed, reducing size is treating a symptom while the underlying assumption goes unexamined.
Decide the Ladder in Advance
All of this has to exist on paper before it is needed. The state of mind that follows a bad week is not one in which sensible sizing decisions get made, and a decision will get made either way, so it may as well be made early and calmly.
A short written ladder covers it: the normal size, the trigger that cuts it, what it cuts to, and the condition that restores it. Four lines. The value is not in the sophistication of the scheme but in its existing at all, because the alternative is not a better scheme. The alternative is sizing by mood, which tends to produce the largest position immediately after the most painful loss.