Ask a trader with a $5,000 account what they’re working on and you’ll usually hear something about entries. Ask what their position size was on their last twenty trades and most cannot tell you without checking. That asymmetry is worth examining, because size variance destroys more small accounts than bad entries do.

The reason is arithmetic rather than psychology, and it holds regardless of how good your setups are.

It’s worth being precise about what “size” means here, because the word gets used loosely. What matters is not notional value or leverage. It’s the amount you lose if the trade goes to your stop - risk per trade, in currency. Two positions can differ by 10x in notional and carry identical risk if the stops are placed accordingly. Traders who track leverage instead of risk often believe they’re consistent while their actual exposure swings wildly, which is the most common way this problem hides in plain sight.

Two traders, identical win rates, different outcomes

Both traders run a $10,000 account, take 20 trades in a month, and win exactly half at 1R.

The first risks 1% on everything. Ten wins at $100, ten losses at $100. The month ends flat, minus fees. Unremarkable, and completely survivable.

The second risks 1% on eighteen trades and 5% on two - the two they felt strongest about. If both convictions lose, that’s -$1,000 from those alone, against a flat book on the rest. The month ends down 10%, and recovering it takes ten winning 1R trades at the normal size. Two decisions consumed the output of the next fortnight.

Worth noting that the second trader also had the better month available to them. Had both convictions won, they’d be up 10% and entirely convinced the approach works. That version happens roughly as often, and it’s the one that teaches the wrong lesson.

Now the uncomfortable part. Nothing in this example required the second trader to be wrong more often. Their win rate matched. Their edge matched. The entire difference came from how much they bet when they felt certain, and feeling certain is not a signal that has been validated against anything.

Conviction correlates with the wrong things

Traders size up when a setup looks clean, when they’re recovering from a loss, or when the last three trades worked. None of those predict the outcome of the next trade, and two of them are known failure modes.

Sizing up after a loss is the mechanism behind most account-ending sessions - the position gets larger precisely when judgement is worst. Sizing up after wins is the same error with better feelings attached, and it’s why a win streak is a risk event rather than a reward. Both produce the same signature in the data: your largest positions cluster around your worst decisions.

You can check this in an afternoon. Sort your trades by size, take the top decile, and compare their average R-multiple against everything else. If the biggest positions underperform the small ones, conviction is costing you money, and it will keep costing you until the sizing rule stops depending on how you feel.

Drawdown is asymmetric, and that’s the whole argument

Losses and recoveries are not symmetric, which is why variance is more dangerous than it looks:

The gap widens fast. Every point of extra variance pushes you toward the part of the curve where recovery stops being realistic. A consistent trader losing 10% has an ordinary month ahead. A variance-heavy trader who takes the same edge into a 40% hole needs a 67% gain, which usually leads to more size, which is how the hole gets deeper.

Consistency is not a virtue here, it is capital preservation. Staying in the shallow part of that curve is what keeps you around long enough for an edge to express itself over the few hundred trades it needs.

The case for sizing up, and where it actually belongs

None of this argues for never increasing size. Accounts that never scale never compound, and a trader who risks 1% forever on a growing balance is already sizing up in absolute terms.

The distinction is what triggers the change. Scaling that follows the account works: risk stays at a fixed percentage, so the position grows as the balance does, automatically and without a decision. Scaling that follows a feeling doesn’t, because the feeling arrives most strongly at the two moments already identified as dangerous.

If you want a discretionary layer on top, tie it to something that can be checked. A setup type with 100+ trades behind it and a better average result than your baseline is a defensible reason to weight it slightly heavier. “This one looks really clean” is not, because you’ve never counted how often that feeling was right.

The test is simple: could you have written the sizing rule down last month? Rules that only exist at the moment of the trade are moods.

Variance costs you twice

Beyond the drawdown maths, uneven sizing quietly wrecks your ability to learn anything from the record.

If every trade risks the same amount, your P&L curve reflects the quality of your decisions. When sizing varies five to one, the curve mostly reflects sizing, and the decision quality underneath becomes invisible. Two months with identical setups and identical hit rates can produce completely different results, so you have no way to tell whether a change you made helped.

That’s the second cost, and it compounds. It means every question in this article - which hour, which setup, whether the new rule worked - becomes unanswerable until size stops moving. Consistent sizing is what makes the rest of your data readable.

Measure the variance, not the intention

Everyone believes they size consistently. The record usually disagrees, and the gap between belief and record is where the money goes.

Two numbers make it concrete. First, the spread: your largest position of the month divided by your median position. Under 2 is tight. Over 5 means your risk per trade is effectively unmanaged, whatever your written plan says. Second, look at your consistency score in the app, which tracks how evenly you’re operating rather than how much you made.

Then check the timing. Pull your five largest positions and look at what came immediately before each one. A loss? A win streak? A gap of several quiet days? The pattern is usually obvious once the trades are lined up, and it’s the same pattern for most people: size follows emotion, not analysis.

Try this tonight

Open Trady, go to the Review tab, and sort your last thirty trades by size. Look only at the five largest. For each, note what happened in the session before it.

Then open Insight and compare the average result of those five against your overall average. If the big ones are worse, you’ve found something you can fix this week - and unlike an entry edge, it requires no new skill, no new indicator, and no larger sample. It only requires that the number stops changing when your mood does.

Trading carries risk. This article is about analytics and risk measurement, not financial advice.