Overtrading: Cost Times Frequency
Overtrading is taking more positions than a method genuinely generates, usually by loosening criteria rather than by finding more setups. Because execution costs are fixed per trade, doubling the count doubles the bill without doubling anything on the other side.
How it works
Overtrading is not trading a lot. It is trading more than the method generates. A scalping approach producing thirty trades a day is not overtrading; a swing method producing four a week and taking twelve is.
Every trade costs 2% of a median bar’s range in round-trip costs on this site’s shared history — and 45% of the smallest bar in the series. That cost does not scale with conviction, quality or anything else.
Which makes the arithmetic simple and unforgiving. Doubling frequency doubles costs exactly. It does not double the edge, because the additional trades are by construction the ones that did not meet the original criteria.
On the smallest bars the cost is most of the available move, which is why the extra trades tend to cluster in exactly the conditions where they are least affordable.
Why it happens
The usual explanation is greed and the usual cause is boredom. Sitting at a screen for six hours with a method that produces two signals is an experience most people resolve by finding a third.
Hours present predicts trade count better than any market variable. Which means the most effective intervention is not a rule about entries — it is a rule about how long you are in front of the chart.
And the additional trades are systematically the weakest. They are the ones nearest the threshold, taken because the threshold was relaxed. So the extra frequency does not sample randomly from the method’s distribution — it samples from the bottom of it.
That asymmetry is what makes overtrading expensive twice over: more cost, and worse trades.
In practice: it is a counting problem
Write down the number of trades your method should produce per week before the week starts. Then count what you took. The gap between the two is the entire diagnosis, and it is available without any judgement about whether individual trades were good.
A running count is more effective than a rule. Seeing “trade 9 of an expected 4” on a screen changes behaviour in a way an instruction not to overtrade does not.
Thin hours are the expensive version of it. Low volume means small bars and wider spreads, so the same fixed cost consumes a larger share of a smaller opportunity.
A longer holding period pays the same cost far less often, which is the structural fix rather than the behavioural one.
Tightening stops produces the same effect by a different route. More stop-outs means more re-entries, and each re-entry is a full round trip — so a change intended to reduce risk can increase total cost substantially.
And the counterparty’s economics are worth stating plainly. A broker earns on volume, not on outcomes, which is not a conspiracy and is a reason to be sceptical of anything that encourages more activity.
One structural fix is worth more than any behavioural one: reduce the number of instruments you watch. A trader following one market takes the trades that market offers; a trader following twelve will always find something happening somewhere, and the something is rarely a setup that would have qualified on its own.
Watchlist size and trade count move together. Cutting a list from twenty names to four is a change that takes a minute, requires no self-control, and reduces the opportunity to overtrade by removing the opportunities. It also improves the quality of what is left, because attention on four markets is a different thing from attention on twenty.
What overtrading is not
It is not high frequency. A method that genuinely produces many signals is not overtrading them.
It is not a discipline problem alone. It is a design problem with a behavioural symptom.
It is not fixed by better entries. The extra trades are the problem, not their quality.
And it is not visible in a single day. It is a count across weeks.
When it fails
A range produces it almost automatically. Setups appear constantly, none of them run, each one pays the round trip, and the accumulated frustration produces more of them rather than fewer.
The second failure is attributing the losses to strategy. A method producing a small edge and taking three times its intended trades loses money, and the strategy gets blamed and replaced.
A third is fixing it with resolve. Deciding to trade less has no mechanism; leaving the screen does.
A fourth is not knowing the expected count. A method with no stated frequency cannot be overtraded by definition, which is why the number belongs in the plan.
And a fifth is treating a busy day as a productive one. Activity and progress are unrelated here, and the feeling that they are is the thing being exploited.
The original data
On this site’s shared 576-bar history the round-trip cost is 0.0098 price units — 2% of the median bar
range of 0.493, 45% of the smallest bar of 0.022, and more than 10% of the bar’s range on 15 of the 576
bars. Bar ranges span 0.17 to 1.10 between the tenth and ninetieth percentiles, a ratio of 6.5. The
figures are in research/series-measurements.json, produced by site/measure_series.py.
The 6.5-fold range spread is what makes the extra trades so costly. They cluster in quiet conditions, where a tenth-percentile bar offers 0.17 of range against a fixed cost of 0.0098 — a far larger share of the opportunity than the same cost takes from an active bar. Count your trades against your plan’s expected number for four weeks, and if the ratio is above one and a half, the cheapest available improvement is not a better method but fewer hours at the screen.
Related
Intraday trading is where the cost arithmetic bites hardest. Discipline covers the design changes that reduce trade count. And trading plan is where the expected frequency should be written down.
The month I counted rather than felt was the month this became solvable. My plan implied about eight trades; I had taken twenty-nine. Nothing about my analysis had changed - I had simply been present for six hours a day and found things to do.
— Michael Whitman
This page is educational, not financial advice. Test every idea on your own charts before risking money.