Trade duration histogram: the picture that exposes bad habits

[Shadow Trader]

Plot how long each trade lasted, in buckets of ten seconds or one minute. Most strategies have an intended holding time, and the histogram shows how closely reality matches it.

A common surprise is a second hump out on the right. The plan said three-minute scalps, but a cluster of trades lasted 25 minutes. Those are usually losers that were held because the stop was moved or ignored. Splitting results by duration often shows the long trades dragging down the whole average.

A second pattern: a spike at the very short end. Trades closed within seconds, often from panic exits or fat-finger entries.

Colouring the bars by winners and losers adds another layer. If winners cluster in the 1 to 4 minute range and losers stretch beyond ten, a time stop would probably help.

It's a five-minute chart that usually produces a good hour of reflection.

Another useful cut: split the histogram by session. A strategy may behave as intended in London but turn into a slow bleed during the Asian session, simply because price moves too slowly for the targets. Duration plots by session make the mismatch obvious.

And for automated systems, the histogram doubles as a bug detector. A trade that lasted six hours in a system designed for two-minute scalps is either a feature gone wrong or an order that never closed.

What does the duration histogram look like, and is there a suspicious second hump?

[FXS HongKong Scalper]

The second hump is a classic. When we did this on our desk, the long trades were almost all losers where the stop had been moved further away. The planned three-minute trades had a positive average; the 20-minute ones lost almost everything the short trades made.

A time stop fixed most of it. If a scalp hasn't reached its target or stop within a fixed time, we close it. Our histogram now has one clear hump, and the right tail is mostly news-day trades.