Ledger · prediction markets

What the market priced vs what happened

Prediction-market prices are easy to quote after the fact and impossible to verify after the fact. So we record them before the event resolves — the question, the price each venue carried, the source URL and the timestamp — and then let resolution fill in what actually happened. Nothing here is written in hindsight.

2 entries recorded · 1 resolved · 1 awaiting resolution · ledger last updated 2026-07-20

One resolved entry proves nothing on its own

An entry below has resolved, and the ledger now shows the price each venue carried beforehand next to what actually happened. That demonstrates the record works — the prices were written down in advance, with sources, and can be checked. It does not show that prediction markets are accurate, and we are not going to claim it does. Accuracy is calibration: whether things priced at 60% happen about 60% of the time, measured across hundreds of events. With 1 resolved entry there is no error bar and no score to report. The value of this page is the series it accumulates, not any single row in it.

The ledger

Which party controls the US House after the 2026 midterm elections?

Awaiting resolution

The next long-dated market Monitoring already tracks on /us-midterms-2026. Recorded on 2026-07-20, roughly three and a half months before election day on 3 November 2026 — so the price is captured while the question is genuinely open, not near the close.

Democratic Party

Polymarket84.5%

Recorded 2026-07-20 00:47 UTC · source

Read directly from Polymarket's public Gamma API (event which-party-will-win-the-house-in-2026, market outcomePrices) — the same feed behind /api/us-midterms-2026.

Republican Party

Polymarket15.5%

Recorded 2026-07-20 00:47 UTC · source

Read directly from Polymarket's public Gamma API (event which-party-will-win-the-house-in-2026, market outcomePrices) — the same feed behind /api/us-midterms-2026.

Venue spread · Not comparable — only one venue price is recorded for each side of this question, so there is nothing to compare.

What actually happened

Not yet known. This event has not resolved, so the ledger records no outcome — Monitoring does not fill this in with a guess. Scheduled to resolve 2026-11-03 00:00 UTC.

Caveats on this record

  • Only one venue is recorded. Kalshi's public elections API listed no open 2026 House-control market when this entry was written on 2026-07-20, so there is no second price to compare — and none was invented. This entry therefore has no venue spread.
  • The two figures are separate binary contracts inside one Polymarket event group, not two independent readings. They currently sum to exactly 1.00; that is a property of how the group is quoted, not corroboration.
  • This price is captured months before resolution and will move — possibly a great deal. It is a record of what the market carried on 2026-07-20, not a forecast by Monitoring and not a claim about the election.

Who wins the 2026 FIFA World Cup — Spain or Argentina?

Resolved

The final, Spain v Argentina at MetLife Stadium, 19 July 2026. The single largest market in either venue's history: Kalshi's contract on this match passed $1.27B in trading volume before kickoff, against >$25B traded across all World Cup contracts.

Spain

Polymarket59.1%

Recorded 2026-07-19 00:45 UTC · source

Read directly from Polymarket's public Gamma API (World Cup Winner event, outcomePrices).

Kalshi61%

Recorded 2026-07-17 · source

Publisher-reported: Fortune, 2026-07-17 — "As of mid-day Friday, bets on Kalshi showed the Spanish team with 61% odds". Not a same-instant reading.

Argentina

Polymarket40.9%

Recorded 2026-07-19 00:45 UTC · source

Read directly from Polymarket's public Gamma API (World Cup Winner event, outcomePrices).

Venue spread · The venues disagree by 1.9 percentage points at their widest. Two independent pools of real money landing this close is the interesting part; a wide gap would be the signal instead. The spread says nothing about which venue was closer to the outcome — a spread is a disagreement, not a score.

What actually happened

Spain · 1–0 · after extra time (AET) · recorded 2026-07-20

Ferran Torres scored in the 106th minute, assisted by Nico Williams; Argentina played extra time a player down after Enzo Fernández was sent off (second yellow) in second-half stoppage time.

What each venue priced beforehand

  • Polymarket·Spain59.1%·this side happened
  • Polymarket·Argentina40.9%·this side did not happen
  • Kalshi·Spain61%·this side happened

That is the whole comparison: the number each venue carried, and which side occurred. It is not a score. A price is not a prediction, and neither venue is graded by a single event — that only becomes possible across many of them.

Caveats on this record

  • No Kalshi price for Argentina is recorded: the 61% Spain figure is publisher-reported and no matching Argentina figure was published alongside it. The gap is left visible rather than filled by arithmetic — in particular, Argentina was NOT derived as 1 − 0.61.
  • The two venue readings are ~2 days apart (Kalshi 17 July, Polymarket 19 July). Treat the venue spread as indicative, not as a simultaneous quote.
  • Polymarket's Argentina price is recorded as 40.95%, not rounded to 40%. The recorded figure is the one the venue carried.
  • Both recorded prices are for the OUTRIGHT question (who lifts the trophy, extra time and penalties included). Kalshi's public API also carries a separate regulation-time 3-way match market, which prices differently and is not the same question.
  • Resolution was verified against two independent primary sources (FIFA's match centre record and ESPN's match report) before it was written here. The recorded prices above were not altered when the result was added.
  • One resolved entry is an anecdote, not evidence of accuracy. A single event has no error bar, and a 59% favourite winning once is equally consistent with a well-calibrated market and a badly-calibrated one. Only the series can answer the question.

How to read this

What a 59% price means

A prediction market sells a contract that pays $1 if something happens and $0 if it doesn't. When that contract trades at 59 cents, the market is saying: at this price, buyers and sellers are both content. Convert it and you get an implied probability of 59%— read as “if this exact situation played out many times, this outcome happens roughly 59 of every 100.” The other side is priced at roughly 41%.

The most common misreading is treating 59% as a prediction. It isn't. A 59% favourite is expected to lose about two times in five, and when it does, the market was not “wrong” — its own stated 41% simply arrived. This is exactly why one event can never settle the question.

Why two venues agreeing is informative

Polymarket and Kalshi are separate venues with separate users, separate liquidity and separate rules. When they price the same question within a point or two of each other, that agreement is weak but real evidence the number reflects shared information rather than one venue's crowd or one venue's thin order book. When they diverge, the gap is the story: it usually means low liquidity, subtly different contract wording — a market on the trophy is not the same market as a 90-minute result — or genuine disagreement. That is why this ledger records the spread instead of averaging it into a single tidy number.

Why one event proves nothing and a ledger proves something

Take any single resolved event and you can tell whichever story you like: the market “called it” or the market “got it wrong.” Neither claim survives contact with statistics, because a single observation has no error bar. The only meaningful test is calibration — collect every occasion the market said roughly 60%, and check whether those things happened roughly 60% of the time. Do that across hundreds of events and you have an answer. Do it across one and you have an anecdote.

That is the entire design of this page. Each entry is recorded before resolution, with sources, so it can be audited later; the ledger accumulates; and the calibration question only gets asked once there are enough resolved entries to ask it honestly. If you want the underlying mechanics — how prices move, what moves them, and what the number is not — the full guide is how prediction markets work.

Why the prices are recorded in advance

A price nobody wrote down cannot be recovered. Once an event resolves, the market that priced it is gone — the order book closes, the page changes, and every later account of “what the market thought” is reconstruction. Writing the venue, the number, the source URL and the timestamp down beforehand is the only version of this that can be checked by someone who doesn't take our word for it.

Related

Common questions

What does a 59% price actually mean?

It means traders are collectively willing to take both sides of the question at 59 cents on the dollar. Read it as: if this exact situation played out many times, the market thinks this outcome happens roughly 59 times in 100 — and the other outcome happens roughly 41 times in 100. It is not a prediction that the outcome will happen. A 59% favourite losing is not the market being wrong; it is the market's own stated 41% arriving.

Why does it matter that two venues agree?

Two independent pools of real money reaching nearly the same number is weak evidence that the price reflects information rather than one venue's crowd or one venue's liquidity. When venues disagree widely, that gap is itself the signal — it usually means thin liquidity, different contract wording, or genuinely different opinion. We record the spread rather than averaging it away.

Can one resolved event tell you whether prediction markets are accurate?

No. A single event has no error bar. A market that says 60% and is wrong once has told you nothing; the same market being wrong 40% of the time across hundreds of 60% calls is exactly what accuracy looks like. Only a ledger of many recorded predictions, scored against outcomes, can say anything — which is why this page records prices before resolution and accumulates them, rather than writing up results after the fact.

Why record prices before the event instead of writing about it afterwards?

Because a price that was not recorded before the event cannot be reconstructed afterwards. Post-hoc accounts of "what the market thought" are unverifiable and tend to flatter whichever narrative the result produced. Recording the price, the venue, the source URL and the as-of timestamp in advance is the only way the ledger can be checked by anyone later.

Does Monitoring set these prices or take positions?

No. Monitoring records what public real-money prediction markets are already pricing, with a named source and an as-of date for every figure. This page is information and entertainment only: not betting, financial or legal advice, not a recommendation to wager, and not affiliated with any exchange or bookmaker.

Monitoring Pro · early access

The free API stays free. Pro adds higher rate limits, full history, and webhooks — for teams building on the feed.

Leave your email to hear when it opens. One email, no marketing.

Every price on this page is a recorded observation of a public real-money prediction market, attributed to a named source with an as-of date. They are not forecasts by Monitoring, not an endorsement of any outcome, platform or product, and not a claim about what will happen. This page is information and entertainment only: it is not betting, financial or legal advice, not a recommendation to wager, and not affiliated with any exchange or bookmaker. Where an event has not resolved, no outcome is recorded and none is implied.