Position sizing: the only rule that matters

The brutal arithmetic of drawdowns

A 50% drawdown requires a 100% gain to recover. A 75% drawdown requires a 300% gain. This is the single most underappreciated fact in retail investing.

Most traders blow up not from bad ideas, but from sizing every trade as if it's their best one. Two losing trades sized at 20% of capital each turns a winning year into a flat one. Three of them turns it negative. Four turns it into a year you don't talk about.

The Kelly criterion (and why the textbook version will ruin you)

The Kelly criterion is the mathematically optimal bet size for a repeated wager with a known edge. The full formula is:

f = (bp - q) / b

Where p is your win probability, q is 1-p, and b is your win/loss ratio. For a trade with 55% win rate and a 1.5:1 win/loss ratio, Kelly says size = (1.5 × 0.55 - 0.45) / 1.5 = 25% of capital per trade.

If those numbers were known with certainty, that would be the right answer. They are not known with certainty. Your true edge is some unknown distribution around your estimate. The damage from over-sizing is asymmetric — over-betting reduces your geometric growth rate much faster than under-betting does.

The half-Kelly rule (what professionals actually use)

The industry standard is half-Kelly: take whatever the formula spits out and use half. So in the example above, 12.5% per trade.

This isn't a fudge. It's a precise mathematical adjustment for parameter uncertainty. The result: you give up some expected growth in exchange for a much higher probability that you survive the inevitable string of losses that any positive-edge strategy will produce.

A simulation: if your true edge gives Kelly = 25%, betting full Kelly has a roughly 1 in 3 chance of a 50% drawdown over 100 trades. Betting half-Kelly has roughly a 1 in 15 chance. Same edge. Vastly different ride.

The simpler rule for the rest of us: fixed fractional, 1-2%

If you don't know your edge with statistical confidence, you don't get to use Kelly. Period.

What you can do is use a fixed fractional rule. Risk a fixed small percentage of capital per trade, where "risk" means the loss you'd take if your stop is hit. The industry-standard band is 1-2% per trade.

At 1% risk per trade, you can take 20 losing trades in a row and still have 82% of your capital. At 5% risk per trade (where many beginners size), 20 losers takes you to 36%. At 10% (where many beginners size their "high conviction" trades), 20 losers takes you to 12%.

Position sizing vs trade sizing — they're different

Risk and dollar exposure are not the same.

• Trade size is the dollar value of the position you take on.
• Risk size is the dollar amount you lose if your stop hits.

If you put $10,000 into a stock with a 5% stop, your trade size is $10,000 but your risk size is $500. On a $50,000 account, that's a 1% risk trade.

The mistake retail traders make: they think of position sizing as "how much should I put in." It's actually "how much am I willing to lose." Once you reframe it that way, the math becomes obvious.

What concentration is acceptable

Even with great risk management per trade, position concentration is a separate risk. Three different "1% risk" trades on three different semiconductor stocks is not three independent 1% bets. It's one 3% bet on the semiconductor cycle with extra steps.

The professional rule: cap correlated exposure at 3-5x your single-trade risk. So if you size at 1% per trade, no more than 3-5% of capital in any one sector, theme, or factor.

What the math tells us about win rate

Here's the counterintuitive part. With proper sizing, a 40% win rate strategy can beat a 60% win rate strategy if the average winner is bigger than the average loser by a sufficient margin.

The full formula for long-run expected growth is:

E(growth) = p × log(1 + bf) + q × log(1 - f)

Where f is your fixed fractional bet size. Plug in different combinations and you'll find that a "boring" 40% win, 3:1 reward/risk strategy outperforms an "exciting" 70% win, 1:1 strategy at almost every fraction.

This is why every legendary trader gets boring later in life. The ego-bait of "high win rate" is a tax. The grind of "small loss, small loss, small loss, big win" is an asset.

The one rule

If you remember nothing else from this article: the size of your worst loss should be small enough that you do not change your behavior after taking it.

If a single losing trade makes you stop trading for a week to recover, you were sized wrong before the trade hit. Period.