Position Sizing When You Might Be Wrong
Sizing determines survival more than selection does. The arithmetic of ruin, why fractional approaches exist, and why the book is the real constraint.
How much to put on matters more than what you put it on. A good selection process with bad sizing loses money; a mediocre one with careful sizing survives to improve.
This is a description of how sizing works, not advice about what any individual should do.
The asymmetry of losses
Lose half and you need to double to recover. Lose 80% and you need five times. The arithmetic of recovery is brutal and it is why survival dominates.
A run of losses is not evidence that your process is broken — with a genuine small edge, long losing stretches are expected. But you only get to find out whether it was a run if you are still there. Variance and drawdown.
What the Kelly criterion says
For a bet with a known edge, there is a fraction of capital that maximises the long-run growth rate. Betting more than it reduces growth; betting enough more than it guarantees ruin.
The formula requires knowing your edge precisely. You do not. Your edge estimate comes from a small sample and is almost certainly overstated, because the ones you noticed are the ones that looked good. Luck versus skill in small samples.
This is why practitioners who use it at all typically use a fraction of it. Overestimating your edge by a factor of two while betting the full amount is the case that ruins people, and the fractional approach exists because that error is the normal one.
The constraints that bind first
In event markets, the theoretical optimum is usually irrelevant because something else binds sooner.
The book. If forty contracts are available at a reasonable price, your position is forty contracts. No amount of conviction changes this. Depth and slippage.
Capital lockup. Every position ties up capital until settlement. Ten simultaneous positions may commit most of your capital for months, regardless of how small each felt. Capital lockup.
Correlation. Positions that look independent often are not. Several markets on the same underlying event are one bet. Correlated positions.
Practical principles
Size on committed capital, not on entry price. Selling a contract at 4 cents commits 96 cents. It is a large position that feels like a small one.
Cap per-question exposure. Not per contract — per underlying event, after accounting for correlation.
Do not size up after a good run. The run is the weakest evidence you will encounter and it arrives when you are most inclined to trust it.
Do not size to make a target. Needing a particular return is not information about the market.
Leave room to be wrong repeatedly. Not once. Repeatedly, in a row, on things you were confident about.
Sizing a cross-venue position
Both legs must be sized together, at the smaller of the two books, and both must be entered and exited together. An unwound single leg is an outright position you never chose.
Because the edge on such a pair is small and the capital committed is large, the honest size is usually smaller than the book would permit. What a gap is worth.
The summary
Size as though your edge is smaller than you think, your sample is shorter than you think, and your positions are more correlated than they look. All three are usually true.
Next
Educational content only. Nothing here is financial advice, a solicitation, or an offer to trade. Prediction markets can move against you and a position can lose its entire value. Where this guide describes a difference between venues it describes a theoretical cross-market edge in observed prices, which is not a profit: fees, spreads, capital lockup, transfer time and differing settlement rules all come out of it first.
Keep reading
Risk and Costs · 3 min readCapital Lockup, the Cost Nobody Bills You ForMoney committed to an event contract earns nothing until settlement. On long-dated markets this cost dominates everything else and it appears on no statement.
Risk and Costs · 3 min readThe Limits of Any Measured EdgeA measured price difference is a measurement. The distance between that and a return is made of costs, capacity, decay and the chance the measurement is wrong.
Risk and Costs · 3 min readHow Fees Eat an EdgeA strategy with a three-cent edge and a four-cent round-trip cost is a losing strategy. Working through where every cost lands and what survives.
Kalshi · 3 min readHow a Kalshi Contract Is Put TogetherSeries, event, strike and settlement rule. Taking apart a single Kalshi market so the ticker stops looking like noise and starts reading as a sentence.