How Many Stars Should You Commit? A Sizing Guide

Two players make identical calls all season. Both are equally accurate. One finishes near the top of the leaderboard; the other spends half the season rebuilding a balance. The difference is sizing, and it is the half of forecasting that almost nobody practises.

Stars are a virtual currency — you cannot deposit money to get more and you cannot withdraw them as money — but they are finite, and running out means sitting out the markets where you actually had an edge. Treating that constraint seriously is what sizing is.

The one idea that matters: edge, not confidence

The size of a position should be proportional to how far your probability sits from the market's, not to how sure you feel.

Those come apart constantly. You can be extremely confident that a heavy favourite will win — and if the market already prices them at 94 and you think 95, you have almost no edge, and the correct position is tiny or nothing. Meanwhile a market at 40 that you honestly assess at 55 is a genuinely large edge on a question you feel much less certain about.

Feeling certain is about the outcome. Edge is about the disagreement. Only the second one pays.

A workable ladder

You do not need optimal-growth mathematics to size well. You need a ladder you actually follow. Something like:

• Edge under 5 points: pass. This is inside your own estimation error. You cannot tell a real 5-point edge from a rounding error in your own head, and neither can anyone else.
• 5–10 points: small position. Call it 1–2% of your balance.
• 10–20 points: standard position. 3–5%.
• 20+ points: re-read the resolution criteria. If it survives that, this is your maximum — and your maximum should still be a single-digit percentage of your balance.

Two things about that ladder are more important than the exact numbers.

There is a cap, and it is low. Whatever your maximum is, it should be small enough that being wrong is annoying rather than structural. If a single resolution can meaningfully damage your season, the position was too big regardless of how good it looked.

The bottom rung is "pass". Most markets, most days, produce no position. That is the correct output of a disciplined process, not a failure to find anything.

Why the mathematically optimal answer is still too big

There is a classical formula — the Kelly criterion — that gives the position size maximising long-run growth given your edge. Experienced practitioners in every field that uses it apply a fraction of it, typically a quarter to a half, and here is why that applies doubly to fandom markets.

Kelly assumes your probability estimate is correct. Yours isn't. It is an estimate with error bars, and if your error runs in the wrong direction — you thought 60, the truth was 45 — the formula that felt optimal was aggressive on a position you should not have held at all.

The asymmetry is brutal: oversizing risks ruin, undersizing merely slows you down. When the costs of the two errors are that unequal, err heavily toward the recoverable one.

Sizing rules that survive contact with a real week

Flat-stake as a default. If you are not going to compute an edge every time, use the same modest position on every call. It is unglamorous, it removes emotion, and it beats intuitive sizing decisively for most players, because intuitive sizing systematically means "biggest position on the call I'm most excited about" — which is the call most contaminated by fandom.

Cap your exposure to correlated outcomes. Five separate calls on the same match are not five positions; they are one position with extra steps. If the favourite collapses, all five go together. Add up everything riding on a single event and treat that total as the real position.

Never chase. The urge to make it back with one large call after a bad run is the most expensive instinct in any scoring game. Your balance does not know it is down, and the market does not owe you a recovery. Chasing converts a bad week into a bad season.

Never scale up on a hot streak either. A run of correct calls feels like evidence of skill and is mostly evidence of variance. Ten straight wins is unremarkable if your calls average 80%. Keep the ladder.

Leave room to be active. A balance is only useful if it lets you take positions when opportunities appear. Committing most of it to a market that resolves in three weeks means missing everything in between — including the better opportunity that shows up on Thursday.

Sizing for a weekly leaderboard, specifically

Standom's board runs on a weekly cycle and ranks by net profit, and that changes the calculus in a way worth naming explicitly.

Over a long horizon, steady well-sized calls compound and win. Over a single week, they mostly won't top a leaderboard — because someone will get lucky on a large position, and a short window is dominated by variance.

You therefore have two coherent strategies, and it is worth knowing which you are running:

Play the season. Size conservatively, accumulate, and let calibration do the work. You will rarely top a single week and you will finish well ahead over a quarter.

Play the week. Concentrate on a small number of high-variance calls, accept a high chance of finishing nowhere in exchange for a real chance of finishing first.

Both are legitimate. What loses is doing the second one accidentally — sizing up because a call felt exciting, while believing you are playing the first strategy. Pick deliberately, and if you are playing the week, do it with a portion of your balance you have consciously set aside for it.

When to break your own rules

Rarely, and for one reason only: when the market is clearly wrong about the question rather than the world.

If you find a take where the price implies people are misreading the resolution criteria — the crowd is pricing "will the film be good" and the criteria ask "will it be released before a date" — that is a different kind of edge from a forecasting judgement. It does not depend on you predicting the future better than anyone else, only on you having read more carefully.

Even then: size up, don't go all-in. You could be the one misreading it.

The uncomfortable summary

Most players do not lose Stars because their predictions are bad. They lose because their good predictions were small and their bad ones were large — and the reason for that is that excitement, not analysis, was setting the size.

Fix that and your existing knowledge starts converting into leaderboard position without you learning a single new thing about cricket or cinema.