Do not retag last year’s losses as "not my setup"
Tag hygiene, why auto-tags are drafts, and how to retire a strategy without corrupting the record.
There is a specific temptation at the end of a financial year: quietly move the losing trades out of your favourite setup so the win rate for that setup stays respectable.
It works. That is the problem.
Tag close to the fill
The reason you took a trade is most accurate within minutes of taking it. Tag at entry, or immediately after the exit. Tagging months later is not classification, it is narrative repair.
The best-looking tag has nine trades behind it. The untagged pile is usually the interesting one. Illustrative journal data, not a live book and not a performance claim.
Treat auto-tags as suggestions
A model that guesses tags from symbol, hold time and entry hour is a useful intern. Confirm or reject each one. An automatic label is not a strategy recommendation and should never be presented as one.
Retire strategies with a date
When you stop trading a setup, leave the old tag on the old trades and start a new version for new ones. Mixing two versions under one name produces a statistic that describes neither.
- Ten tags or fewer, written down as a fixed list
- Tagged at entry or immediately after the exit
- AI suggestions reviewed, never auto-accepted
- Retired setups versioned with a date, not renamed
- Untagged trades left untagged rather than guessed
Key takeaways
- Tag at the time of the trade, not at year end.
- Confirm AI-suggested tags; they are drafts.
- Version retired setups instead of relabelling history.
Not advice, not a call. NiyataBook is a personal trading journal, not a SEBI-registered investment adviser or research analyst. Everything above looks backwards at trades that already closed. What happened in a book does not promise what happens next, and F&O can take a year’s savings in a week.