Tracking by setup, not just by win/loss

The aggregate hides which patterns are actually working

Here's a trap most new traders fall into. They look at their overall winrate — say 48% — and conclude their strategy is borderline. So they tweak the strategy, change indicators, add filters. None of that touches the real problem, which is that the 48% is a blended average of several very different sub-strategies. Some of those sub-strategies are great. Some are terrible. The aggregate hides them all.

Setup-level tracking fixes this. Every trade you log has a setup tag — the named pattern you took. Pullback to 20 EMA. Breakout from range. Trend continuation after pause. Failed breakout reversal. Whatever your named patterns are. When you have a few weeks of data, you slice by setup tag and look at performance per group. The picture changes immediately.

A worked example. Sample data — 60 trades. Pullback setup: 20 trades, 12 winners, 60% winrate, average R is +0.8. Breakout setup: 25 trades, 9 winners, 36% winrate, average R is -0.2. Reversal setup: 15 trades, 8 winners, 53% winrate, average R is +0.3. Aggregate: 29 of 60, 48% winrate, average R of +0.2. Looks meh. But the breakdown shows pullbacks are clearly your edge, breakouts are bleeding you, and reversals are mediocre. The actionable move? Stop taking breakouts entirely for the next 30 trades and see what happens. Your aggregate winrate will jump dramatically.

What counts as a 'setup' for tagging purposes? Anything you can name and recognize before the trade. The test: could you describe the pattern to a friend in one sentence such that they'd recognize it on a chart? If yes, it's a setup. If you can't name it without arm-waving, you're taking discretionary trades, not setups, and you can't track them meaningfully. The tagging discipline forces you to be precise about what you're actually doing.

Watch the sample size. With only 10-15 trades on a setup, your data is noisy — a 70% winrate on 10 trades doesn't mean the setup is great, it might mean you got lucky. Wait for at least 20-30 trades per setup before drawing conclusions. Until then, treat each setup as 'still under evaluation.' The 90-day demo gate exists partly to give you sample sizes worth trusting.

Recap: tag every trade by setup. Slice your data by setup tag. Cut the setups that bleed; keep the ones that earn. The aggregate winrate lies; the breakdown tells the truth.