Decision fatigue: a scalper whose worst trades came after click number 100

[Shadow Trader]

Take a composite example that matches what a lot of scalpers report. A trader reviews a month of results and finds a puzzling split. The morning trades are clean. The afternoon trades are sloppy, with more late entries, more hesitation exits and a few oversized mistakes. The obvious guess is that the afternoon market is simply worse for the strategy, so the decision is made to stop trading afternoons.

The result barely changes. The afternoon still looks bad on the days when only a few trades happen after lunch. Plotting results by trade count rather than by clock time shows something different: quality drops after roughly the hundredth click of the day, whether that click happens at 11:00 or at 15:00. A heavy morning pushes the decline earlier. A light morning pushes it later.

That is the twist. The market was never the problem. The problem was a finite supply of decision quality being spent, and the clock was just a proxy for it.

What changes after that is how the day is structured. The fix is not trading less overall but capping the number of decisions per block, and counting every click toward it, including partial closes, stop adjustments and entries that got cancelled. Cancelled orders are decisions too, and they get ignored in most personal rules.

Another detail is that decisions are not equally expensive. A decision made from a pre-planned rule costs little. A decision made by judgment costs a lot. Moving more of the day into pre-decided rules, such as fixed targets and automatic stop placement, stretches out the good hours without reducing the number of trades.

None of this is a law of nature, and the threshold clearly differs between people and setups. The only way to find it is to sort results by trade sequence number instead of by time of day, which most platforms' history exports allow with a little spreadsheet work.

Has anyone sorted results by trade count within a session, and does a drop-off point show up in the data?