forex-scalping.com

How long trades last: the duration histogram

Most journals record the entry and exit time. Few use it. A histogram of trade durations shows how the strategy behaves.

A typical scalping system has a cluster of short trades, a few seconds to a couple of minutes, and a tail of longer ones. The tail often hides losers: trades that didn't work quickly and were held while the trader hoped.

Splitting profit by duration bucket is revealing. Trades under 30 seconds might earn money, trades between 2 and 10 minutes lose it, and the average blurs them together.

If the long trades are the problem, a time stop can fix it: exit any position that hasn't reached its target within a set window. The data tells which window.

The opposite also happens. Quick exits at small profits and long holds for bigger wins can produce positive expectancy, but the quick exits may be fear-driven.

The shape of the histogram also shows consistency. A tight cluster means a disciplined process. A smear across all durations suggests improvisation.

In the journal, which holding-time bucket produces the best profit per trade?

Most scalpers define a price stop and a price target. Fewer define a time stop: if the trade hasn't worked within N seconds, close it.

Time matters in scalping because the edge often decays quickly. A momentum entry that hasn't moved within thirty seconds probably lacked momentum. The longer it sits, the more it's exposed to a reversal, to news, to spread changes, and to the temptation to rationalize.

A review of trade durations usually reveals that winners resolve fast and losers drag on. Setting a time exit at the point where most winners have already finished cuts a lot of the slow losers early, often at a small loss or even near break-even.

Time stops also reduce exposure to events. A trade that is closed within two minutes has less chance of being caught by a news release or a rollover.

Implementation is simple in most EAs and possible manually with a visible timer. The hard part is finding the right duration, which needs data from a few hundred trades.

A good approach to finding the limit is to build a histogram of trade durations separately for winners and losers. In many scalping systems, the winners cluster near the beginning, and the losers have a long tail. The point where the winners' density drops off is a natural candidate for the time stop.

Something else worth testing is a progressive time stop. If a trade hasn't reached a fraction of its target by half the time limit, tighten the stop to break-even or exit at market. It cuts losers earlier without sacrificing the winners that are slow but real.

The time stop also protects against scheduled events. Setting the limit so that trades never straddle the top of the hour or a known release time avoids a whole category of problems.

One caution: testing time stops on bar data can mislead, since the exact duration of trades inside bars is not recorded. Tick data gives a cleaner answer.

What duration separates the typical winner from the typical loser in the trade history?