How to Tag Plan and Improvised Trades in Your Journal
Tag every futures trade as plan or improvised so the split is measurable. A precise rule, a clean ledger, and a per-category expectancy you can trust.
By Imperial Analytics
A futures journal that mixes planned entries with improvised ones reports a single expectancy number for two different traders sharing one account. The number is a blend, not a signal. The fix is mechanical, not motivational. Tag every entry as plan or improvised at the moment it is taken, then read the two streams separately.
What does it mean to tag a trade as plan or improvised?
A plan trade is one that matches a written setup rule the trader committed to before the session opened. An improvised trade is anything else taken during the session: a chase, an opportunistic add, a discretionary fade. The tag is binary and applied at entry, not after the result is known.
The split lives or dies on the timing. A trade tagged after the close is a trade tagged by the result. A green improvised trade gets relabeled as a "plan B" entry. A red plan trade gets demoted to "I should have known." The journal then reports two streams that look reasonable, and the actual behavior stays hidden.
The discipline is to assign the label inside the trade entry workflow itself. One checkbox. One field in the broker note. One keystroke in the journal. If the rule for the setup was checked before the entry, the trade is plan. If not, the trade is improvised. Result has no vote.
This is not about labeling improvised trades as bad. Some traders run a profitable discretionary book on top of a profitable systematic book. The point is to know which is which, in dollars, over a sample large enough to mean something.
Why the split matters more than total P&L
Total P&L collapses two different behaviors into one number. The plan stream is what the trader thinks the strategy is doing. The improvised stream is what the trader is actually doing during the session. The gap between those two numbers is the cost of acting outside the written rules.
A trader who reports a profitable month from a single expectancy figure has no way to answer the question that matters: how much of the profit came from the setup they sat down to trade, and how much came from the trades they did not plan to take. The expectancy number does not know.
A worked example, drawn illustratively to show the shape:
Data note
The numbers below are an illustrative split, not a measured Imperial Analytics customer result. Real splits vary widely by trader and instrument.
A 90-day window with 120 trades on MES closes at +$1,800 net. The journal reports an expectancy of +$15 per trade and a win rate of 54 percent. Split the trades by the tag taken at entry:
- Plan trades: 80 entries, win rate 61 percent, expectancy +$48 per trade, total +$3,840.
- Improvised trades: 40 entries, win rate 40 percent, expectancy −$51 per trade, total −$2,040.
The plan strategy is a real edge. The improvised stream is a slow drain. The blended expectancy of +$15 is the average of a working setup and a leak. Without the tag, the trader has no reason to act, because the total is green. With the tag, the action is obvious: cut the improvised stream, or write a rule that turns it into a planned one.
The reverse case is just as common. A losing month can hide a profitable plan stream that is being papered over by a few large improvised losses. The conclusion changes from "the strategy is broken" to "the strategy works, the discipline failed twice."
How to write a rule precise enough to tag with
A taggable rule names the setup, the instrument, the time window, the entry trigger, and the invalidation point. If two readers of the rule would tag the same trade differently, the rule is not yet precise enough. Sharpen it until the tag is mechanical.
Vague rules invite the trader to rationalize. "Trade the open range break" is not a rule. It does not specify the instrument, the size of the range, the confirmation, or the time the range is allowed to form. Every entry the trader takes near the open can be labeled plan under that wording, including the chase fifteen minutes after the actual break.
A taggable rule reads more like this: "On MES, between 9
and 10 ET, enter on a one-minute close that breaks the high or low of the opening five-minute range. Stop is the opposite side of that range. Skip if the range is less than four points." Now the tag is mechanical. If the entry was inside the window, on the named instrument, on a confirmed break of the named range, with the named stop in place, the trade isplan. If any condition was missed, the trade is improvised.
The same precision applies to the conditions that make a trade improvised. The trader who routinely adds to runners outside the written rule is taking a second strategy, not extending the first one. That add belongs in the improvised bucket until a written rule for runner adds is added to the playbook.
One useful test: read the rule to another trader without showing them any charts. If they can apply it to a live tape and produce the same entries the trader does, the rule is precise enough to tag with. If they ask "what counts as confirmation?" the rule is still discretionary.
How to compute and read the plan-versus-improvised split
For each tag, compute win rate, average winner, average loser, and expectancy as separate streams. Read the streams against the matching sample size. Do not compare a plan-trade expectancy from twenty trades against an improvised-trade expectancy from eighty, because the noise bands are different.
The ledger is a per-trade table with at least these columns: instrument, entry time, exit time, tag (plan or improvised), gross P&L, fees, net P&L. The tag is filled in once, at entry, and never edited. The other columns come from the broker fill record.
Per-tag aggregates are straightforward:
- Win rate by tag: count of trades with net P&L above zero, divided by total trades in that tag.
- Average winner by tag: mean net P&L of trades with positive net P&L.
- Average loser by tag: mean net P&L of trades with negative net P&L.
- Expectancy by tag:
(win rate × average winner) + ((1 − win rate) × average loser).
Sample-size discipline is non-negotiable. A per-tag claim under twenty trades is a hypothesis, not a measurement. The Imperial sample-size threshold of twenty trades exists because a binomial proportion at n = 20 carries a standard error of roughly eleven percentage points around its win rate. That band is wide enough that a "60 percent win rate" overlaps a "40 percent win rate" inside two standard errors. The fix is to keep tagging and let the sample grow.
↳ Note
A green improvised stream is information, not a license. Either write the rule that produced it, or treat it as a leak that happened to pay this month.
The two streams can also be read alongside each other, not against each other. A plan stream with expectancy +$48 per trade and an improvised stream with expectancy +$12 per trade is a different finding than the same plan expectancy with an improvised stream at −$51. In the first case, both streams are paying, and the question is whether the improvised stream is a hidden second strategy worth writing down. In the second case, the improvised stream is a tax on the plan stream, and the question is what stops the trader from taking those entries.
What to do when the split surfaces something honest
A surfaced split asks for one of four responses: write a rule that turns the improvised pattern into a plan trade, stop taking the improvised pattern, scale the plan stream where the expectancy is paying, or shrink the plan stream where it is not. Pick one, run the next twenty trades under that decision, then re-read the split.
Writing a rule. If the improvised stream is paying and the trader can name the conditions it shares, say fading the first failed break of the prior session high, that pattern can be promoted into the playbook. The promotion is a written rule, not a mental one. After the rule is added, future trades that match it are tagged plan. The improvised bucket should now shrink for that pattern.
Stopping the pattern. If the improvised stream is losing and the trader can name the trigger that sets it off, such as a missed plan setup followed by an opportunistic add, the response is to install friction at the trigger. A pre-session note, a cooldown timer after a missed setup, a hard rule that adds require the same written conditions as the original entry. The improvised bucket should shrink because the entries stop being taken.
Scaling or shrinking the plan stream. The plan-stream expectancy is a more reliable input for sizing decisions than the blended expectancy is, because it isolates the behavior the trader is sizing for. If the plan stream is paying at a stable expectancy across at least fifty trades, the trader has earned the right to increase contract size on plan trades only. If it is not, the answer is the opposite.
Each of these responses is testable. Run the next twenty trades under the chosen decision, tag every one at entry, and read the new split. The journal will report whether the decision moved the needle or only moved the trader.
Frequently asked questions
- q: Should the tag ever change after a trade is closed?
a: No. The tag is set at entry and frozen. Editing the tag after the close lets the result rewrite the behavior, which is the failure mode the tag exists to prevent. If the trader realizes mid-trade that the entry did not match the rule, the tag was already
improvisedand stays that way. - q: Where does a "missed setup, taken late" entry belong? a: Improvised. The rule named a window or a trigger; the entry happened outside it. The behavior that matters is the one the trader took, not the one the trader meant to take.
- q: What if the trader is fully discretionary and has no written setup? a: Then every trade is improvised by definition, and the goal of the tag is to surface the conditions under which the discretionary book is paying. The trader can group improvised trades by hand-written notes such as "first fade," "second test," or "mean revert," and let the per-group expectancy reveal which patterns to write a rule for first.
- q: How does this interact with the twenty-trade sample-size minimum? a: The minimum applies per tag, not per total. Twenty plan trades and twenty improvised trades is the floor before the split is a measurement rather than a guess. Until each bucket clears twenty, the trader is still gathering evidence.
- q: Does the split work for swing traders with fewer total entries? a: Yes, but the calendar lengthens. A swing trader taking three entries a week needs roughly seven weeks per tag to reach the twenty-trade minimum. The discipline is the same; the window is wider.