Weather has a genuine attraction: money is paid out daily and participants are relatively unsophisticated. I ran a real-money line here and checked two arbitrage paths. The conclusions compress to one sentence: the pool is real, the arbitrage is not, and "the participants are unsophisticated" is not a reason to bet.
| Item | Detail |
|---|---|
| Typical board | which band a city's daily high temperature falls into |
| Shape | a ladder: a dozen mutually exclusive, exhaustive bands |
| Resolution | a named weather station's official reading |
| To establish | which station, which period, and whether revisions count |
That last row is specific to this category: meteorological data gets revised after the fact. Confirm before betting whether settlement uses the first published value or a revised one — this cannot be inferred, only read from the board's terms.
This one is affirmative and independently checkable: batched transfers at a fixed daily time, enumerable on-chain. You need not take anyone's word for it — go and count.
But "money exists" and "you can get it" differ. The pool is split by quote quality, so more participants means less each. And the scoring formula specifically pays for tightness: closer to mid scores higher, with quadratic decay and zero beyond the board's maximum spread.
| Quoting at | Reward | Bad-fill $ |
|---|---|---|
| join best | ×1 | ×1 |
| −2¢ | ×0.26 | ×0.29 |
So "quote deeper, stay safe, still collect" is structurally impossible: the safety is real, and it costs three-quarters of the reward.
The ladder rule holds: a dozen mutually exclusive, exhaustive bands must sum to $1. So "they add to 0.95, buy the set" looks free.
I checked through two unrelated data paths. Both netted to zero.
| Why it cannot be collected | Worse on ladders because |
|---|---|
| Cold bands unbuyable | more bands means more very cold ones — and you need them all |
| Fees on every leg | a dozen bands, a dozen taker fees |
| Minimum 5 shares × N | the minimum viable trade scales with band count |
| Unpaired is naked | a dozen legs — one missing and the set does not exist |
Full account in ladder arbitrage.
"Forecasts are public, free and accurate, so compare them with the quotes for an edge" — the first three premises hold and the conclusion does not follow.
Measured: the 12-to-24-hour band is fully priced, with nothing left. (I reached no separate conclusion on other horizons, so none is stated.) See public forecasts vs weather prices.
I ran a liquidity-provision line and archived it at about −$210. The cause was neither execution nor luck:
When I took it apart I was looking at its past performance. Leaderboards do not tell you whether an approach still works. July's star was no longer valid in August.
The verification step I skipped was one question: "is it still making money now?" That is what the $210 bought. Full account in weather LP.
| Claim | Reality |
|---|---|
| "participants are unsophisticated, so there is opportunity" | does not follow — a few sophisticated participants suffice to set the price |
| "there is a pool, so it pays" | split by score, and you must net off the bad fills |
| "the ladder must sum to $1" | true, but the spread sits in the band you cannot buy |
| "the forecasts are accurate" | true, and everyone can see them |
Three of four are "premise true, conclusion false" — which is this category's characteristic shape: every one of its attractions is genuine, and none converts directly into money.
The money in weather is real (the pool pays daily, verifiable on-chain), but I checked three routes to it and rejected all three. The line worth taking away: "the participants here are unsophisticated" is not a reason to bet — pricing is never set by the average.