Confidence vs. Overconfidence, Measured in Position Size

“I was really confident on this one” shows up in trade notes right before both the best trades in a journal and the worst ones. That's not a contradiction — it's a sign that “confidence” is being used to describe two different things, and only one of them has any business changing how much size goes on.

Two things that both get called confidence

The first is confidence in a setup: a pattern that has shown up in the data before, with a known win rate and a known average reward-to-risk, being recognized again. That kind of confidence is earned from a sample size and is, in principle, justified — it's a read on the setup, not on the trader.

The second is confidence in a decision: a feeling of certainty about this specific trade, right now, that has nothing to do with a sample size because it's a sample of one. It often follows a win, or a string of them, and it feels identical from the inside to the first kind. The only way to tell them apart is to look at what actually changed — the setup, or the trader's mood.

Where it shows up: position size

Confidence in a setup, correctly applied, changes size in a bounded and repeatable way — a higher-probability setup earning a slightly larger allocation within a pre-defined range, the same way it would have earned that allocation the last ten times it appeared. Overconfidence changes size in an unbounded and one-off way: a position two or three times the normal risk, justified after the fact by how “obvious” the trade looked, that would not have gotten the same size if it showed up on an ordinary day.

The test isn't whether size increased. It's whether the increase can be explained by something written down before the trade — a rule that gives A-grade setups 1.5% risk instead of 1% — or only by how the trade felt once it was already being considered. A journal that records size next to the stated reason for that size is the only way to catch the difference, because after a winning trade both explanations sound equally reasonable.

The pattern that gives it away

Pull the position sizes from any account that has gone through a hot streak, and the oversized trades cluster in a specific place: shortly after a win, or a run of wins, not evenly distributed across the sample. That clustering is the signature of overconfidence rather than setup quality, because a genuine edge in a setup doesn't become more true because the last trade happened to work.

The same pattern runs in reverse after a loss, and it's worth checking for both directions. A trader who sizes up after wins and sizes down after losses — on the same setup, unchanged — is letting the outcome of the last trade set the size of the next one, which is the opposite of what position sizing is supposed to do.

What to check for in your own numbers

• Does position size correlate with the outcome of the previous trade? If size after a win is reliably larger than size after a loss on the same setup, size is being driven by mood, not edge.
• Is there a written rule for when size increases?If the only explanation for a larger position is how the trade felt, there isn't a rule — there's a rationalization.
• How wide is the actual range of position sizes used? A setup-driven sizing scheme produces a narrow, bounded range. A mood-driven one produces outliers — a handful of trades several times the size of the rest.
• Do the oversized trades cluster after wins?If they do, that's the specific signature of overconfidence rather than a considered exception.

Why this is hard to catch from memory

In the moment, a trade sized up after three wins in a row doesn't feel like a departure from the plan — it feels like recognizing an edge that's working. That's exactly what makes it invisible without a record: the feeling of justified confidence and the pattern of unjustified size increase are indistinguishable from inside a single decision, and only become separable once size is lined up against the sequence of outcomes that came before it. getALPHA's process review checks position size against your own history rather than against how a trade felt, so a size increase that tracks the last outcome instead of the setup gets flagged as what it is, before it becomes the pattern that defines a drawdown.