Max trades per day: how to set a limit from your own data

There is no number that works for everyone, and any article that hands you one is guessing. Your own log already holds the answer, and it takes about twenty minutes to pull out.

The method is simple: number every trade by its position in the session, then look at what each position earned. Trade one, trade two, trade three, and so on. If your results fall off a cliff at a certain position, that's where your day should end.

One log, sixty sessions

Here is a worked example, built to the shape these tables usually take. Sixty sessions, 214 trades, every result measured in R so that a small day and a large day can sit in the same column.

The first two trades of the day produced +45.0R across sixty sessions. Everything after them gave back 34.0R. The account finished at +11.0R, which at 200 per R is 2,196 where the first two trades alone had earned 8,992.

That is the whole argument in one line: two trades a day carried this account, and the hours after them quietly undid three quarters of the gain.

Before you trust the cut

A cliff in a table is not proof. Two things decide whether yours is real.

Enough trades in the bucket. Around thirty per bucket is where the average starts to mean something. In the table above, trade 4 has only 24 trades and a spread of results wide enough that its -0.35R could be noise. Run the arithmetic: with a typical standard deviation of 1.2R, the standard error on 24 trades is 1.2 / √24, which is 0.24R. A -0.35R average sits 1.4 standard errors from zero, and that is not a finding.

Group the thin buckets instead. Trades 3 and later together are 102 trades averaging -0.33R. The standard error there is 1.2 / √102, or 0.12R, which puts that average 2.8 standard errors below zero. That one you can act on.

A reason behind the number. The position in the session explains nothing on its own. What weakens a third trade is usually the state you have reached by the time you take it. Check your log for the usual suspects: was the risk on that trade larger than your median? Did it follow a loss? Was it taken later in the session, outside the hours your setups actually appear? Was it tagged with a setup from your plan, or untagged?

The position in the day is a label for the state you were in. Once you know which state it is, you can write a rule that names the cause rather than the symptom.

How to run it on your own log

Five steps, and a spreadsheet handles all of them.

1. Sort by date, then by entry time. Number each trade within its session: 1, 2, 3.
2. Convert every result to R using the risk you planned at entry, so position size stops distorting the comparison. If you have never done this, tracking R-multiples is the place to start.
3. Average each position and count the trades behind each average.
4. Merge buckets under thirty trades into a single "and later" row.
5. Look for the sign change, and then look for its cause in the columns you already log: risk, time of entry, setup tag, and what happened on the trade before.

Your analytics page does the grouping and the R conversion for you if your trades carry entry times. A date on its own leaves this analysis guessing, so check that your journal entries carry the time you entered, not only the day.

The rule you write afterwards

Say your cut lands after trade two. The weak version of the rule is a promise to yourself to stop. The strong version is checkable before the entry, not after it:

Trades 1 and 2 are taken from the plan. A third trade requires a setup from the listed set, risk no larger than the median, and at least fifteen minutes since the previous exit. If any of those three fail, the day is done.

That is a rule you can answer with yes or no in a few seconds, and your log will tell you next month whether it held.

Two warnings worth stating plainly. A cap sets a ceiling and never a quota, so a session that offered one clean setup is finished after that setup. And a cap that comes from a stranger's article is a number you will break, because nothing in your own record supports it. The point of the exercise is that the limit arrives with evidence attached.

When the table shows no cliff

Some logs simply do not have one. The average holds steady from trade one to trade eight, and the sign never flips. That is a real result, and it means your losses are coming from somewhere other than session length. Look instead at expectancy by setup, at your worst mistake tags, or at whether the planned reward on your winners is actually being realized.

The absence of a cliff is worth twenty minutes too. It rules out a fix you might otherwise have spent a month on.

Short version

Number your trades by position in the session, convert results to R, and average each position with at least thirty trades behind it. If the sign flips and stays flipped, you have found where your day should end. Then find the cause behind the position, write the rule so it can be checked before the entry, and let next month's log grade it. For more on what the extra trades tend to cost, see overtrading.