Tagging trades: setups, mistakes and why the labels matter
“How should I tag my trades so the tags are actually useful later?”
A tag is what lets a book be grouped by something other than symbol and date. Setups group trades by what they had in common at entry; mistakes group them by what went wrong; notes hold everything that does not fit a label. They are three different mechanisms and they fail in different ways.
In short
A tag is only worth applying if you would ever filter by it. A vocabulary of six labels used consistently produces answers; forty labels used loosely produces a list.
Setups: grouping by what the trades had in common
A setup tag names the pattern or condition the trade was taken on. Its whole purpose is comparison: expectancy per setup, win rate per setup, average hold per setup — figures that only exist if the same label was applied the same way across enough trades.
This is why a large vocabulary defeats itself. Thirty setup names spread a hundred trades across thirty groups of three, and no group has enough in it to say anything. A small vocabulary produces groups large enough to compare.
Mistakes: a separate axis, deliberately
A mistake tag records what went wrong in the execution of a trade, independent of whether it made money. That independence is the point: a trade can be taken against plan and still win, and a book where mistakes are only recorded on losers is a book that has confused the two.
Mistakes are also the tag most likely to be applied late, during a review, because they are often only visible afterwards. That is a legitimate use — unlike a setup tag, which describes what was known at entry and is not recoverable a month later.
Notes and lessons: what does not fit a label
Free text carries the things no controlled vocabulary can. It cannot be grouped by, which is exactly why it should not be carrying anything that could have been a tag — a note reading "broke the plan again" is a mistake tag written where nothing can count it.
The practical division: if you would ever want to filter for it or count it, it is a tag. If you would only ever want to read it, it is a note.
How tagging goes wrong
- Vocabulary drift — the same idea entered as "breakout", "break out" and "BO" becomes three groups nothing can compare. A picker with a fixed list prevents it; a free-text field does not.
- Tagging only the memorable trades, which biases every per-tag statistic toward the trades someone felt strongly about.
- Tags that describe the outcome rather than the decision. "Good trade" is not a group; it is a result already recorded in the P&L column.
- Adding a label for a distinction you would not act on. It costs entry time on every trade and answers nothing.
Questions
- How many setup tags is reasonable?
- Few enough that each group ends up with enough trades in it to compare — which depends entirely on how many trades a book holds. The test is mechanical rather than a rule of thumb: if a tag's group is too small to say anything about, it is not yet earning its place.
- Can a trade have more than one tag?
- Usually yes, and mistakes in particular are multi-valued — a single trade can be both oversized and taken outside its session. Filtering then matches any of the selected values, which is why the counts per tag can exceed the number of trades.
- Should I tag emotional state?
- Some traders do, and it works the same way as any other tag: it is only analysable if the vocabulary is small and applied consistently. Recorded as free text it becomes unreadable in aggregate, which is why a fixed list normalised on read is the usual approach.
Calculate it
In the app
Keep reading
Ledger of Alpha is the record these measurements are computed from. See the full product
Keep the record it all comes from
These measurements need a complete book of trades behind them. A free account gives you somewhere to keep one.