Position Sizing: The Math of Not Blowing Up

In the spring of 2019, I lost 38% of a portfolio in six weeks. Not from bad stock picks. Not from a market crash. From one decision I made before I ever bought a single share: I put too much into one name.

The company was real — strong management, a defensible product, growing revenue. The thesis was sound. The size was not. I had 31% of my investable capital in a single position. When the stock dropped 45% on a sector rotation, there was nothing to cushion it. The rest of the portfolio was fine. It didn't matter.

That was the last time I sized a position by feel.

What the Math Says

The Kelly Criterion is the most important formula most investors have heard of and almost none actually use correctly.

Kelly's formula: f = (bp − q) / b*, where b is the net odds (how much you win per unit risked), p is the probability of winning, and q is the probability of losing (1 − p).

Plug in a trade with a 60% win rate and 1:1 payoff: f* = (1 × 0.60 − 0.40) / 1 = 20%. The math says put 20% of your capital into any trade where you're right 60% of the time and your gains equal your losses.

Almost no serious investor does this. And there's a very good reason.

Why Full Kelly Will Ruin You

The Kelly Criterion assumes you know your edge with precision. You don't. Nobody does.

Your win rate isn't 60% — it's somewhere between 50% and 70%, and you're estimating that range from a limited sample of trades. Your payoff ratio isn't exactly 1:1 — market impact, slippage, and the fat tail of catastrophic loss distort the real distribution.

When you run Kelly on noisy estimates, the formula misbehaves badly. If you overestimate your edge by even 10%, full Kelly will produce drawdowns that are psychologically unbearable — and that's the real risk, not the math itself. You'll panic-sell at the bottom, which is the worst possible behavior.

Ed Thorp, who invented card counting and then applied Kelly to markets, ran half-Kelly throughout his career. He wrote that the difference between full Kelly and half-Kelly is the difference between acceptable drawdowns and career-ending ones.

I don't even run half-Kelly. I cap at 8%.

My Framework: Four Tiers

After 2019, I built a tiered position-sizing framework. Every position I consider gets classified before I size it.

Tier 1 — Core positions (max 8% each): These are businesses I understand deeply, with durable competitive advantages, long track records, and free cash flow. I could own them for ten years. The 8% cap means I can be completely wrong on two of them simultaneously and still survive it.

Tier 2 — Speculative positions (max 5% each): Higher growth potential, earlier stage, more uncertain outcome. Biotech catalyst plays, pre-profitability software companies, turnaround situations. I size these at 5% not because I'm less confident in the thesis, but because the variance distribution is wider. A 5% position that goes to zero costs me 5%. A 5% position that goes to 3x adds 10% to the portfolio.

Tier 3 — Highest conviction, one exception (max 15%): I allow myself one position at up to 15% — but only when three conditions are met simultaneously: I've held the name for more than two years, I have direct experience with the product or service, and the current valuation still offers margin of safety. Right now, that exception is a large-cap technology holding I've owned since 2021. It represents approximately 13% of my investable capital.

Tier 4 — Options overlays (notional 2-4% each): Covered calls on core positions, cash-secured puts on names I want to own at lower prices. These don't count toward position limits unless exercised.

The Calculator I Actually Use

My position sizing process takes about ninety seconds. I open a spreadsheet with three inputs: total portfolio value, my classification for the trade (Tier 1/2/3), and current price.

The output is maximum shares. Not a target allocation — a ceiling. If the trade meets my entry criteria at any size up to the ceiling, I enter at whatever the natural entry level is. If my entry level would require more than the ceiling allows, I either wait for a better price or don't take the trade.

The rule sounds mechanical because it is. That's the point. When I'm in front of a trade I love, I don't trust myself to be objective about size. The framework removes the decision. I built the rules when I was calm; I follow them when I'm not.

What Changed After 2019

The 2019 loss taught me something that sounds obvious but wasn't: position sizing is the primary risk control, not stop-losses. A stop-loss is a reaction. Position sizing is a decision architecture.

If I hold 31% of my portfolio in one name, a stop-loss at 15% down still means a 4.65% portfolio loss on one position. If I hold 8%, that same 15% stop costs me 1.2% of the portfolio — an inconvenience, not a crisis.

The math of recovery is asymmetric in ways that aren't intuitive. A 25% drawdown requires a 33% gain to recover. A 50% drawdown requires 100%. Staying alive in the market long enough to compound is more valuable than optimizing for any individual trade.

The Rule

1. Classify every position before you size it: Tier 1, 2, or 3.
2. Tier 1 max: 8%. Tier 2 max: 5%. Tier 3 (one exception): 15%.
3. Never adjust these limits because you "really like" a trade. You always really like it.
4. Review allocations quarterly. Drift happens. A 7% position that runs to 12% is now oversized.
5. If you can't stomach a full 50% loss on a position without it breaking the portfolio — the position is too large.

I haven't had a single-position catastrophe since 2019. Not because my stock picks got better — they didn't meaningfully. Because I stopped letting any one idea be big enough to matter that much.

The best trade you'll ever make is surviving long enough to make the next one.

— Ruslan Averin