Eurovision Asia Odds: Why the Favourite Lost

The favourite finished fourth. Malaysia closed at 60.5% to win Eurovision Asia 2026 and

came home 152 points and three places behind the Philippines, who closed at 31.0% and won on 211. That is the

headline, and on its own it is the least interesting thing that happened to these markets on Saturday night.

The useful question is not whether the book was wrong — a 60.5% favourite loses roughly two times in

five, and nobody should be surprised by it. The useful question is which of the six Polymarket books on this

contest actually carried information, and which were noise dressed up as a price. Scored against the real

scoreboard, the answers are not the ones you would guess: one of these books performed worse than

knowing nothing at all, and it was not the winner market.

The closing book against the result

These are the Polymarket midpoints as this site recorded them at 17:04 UTC on 13 November, the last reading

before the Grand Final, published that evening in

the closing-prices post. Nothing here is

reconstructed after the fact:

Two rows do the damage. Thailand at 1.55% finished second, on the second-highest public

score of the night. Malaysia at 60.5% finished fourth, and was not top of either half of the

vote. Between them they say something specific about where this market's information actually came from, which

we come to below.

Scoring the three outright-style books properly

“The favourite lost” is not a measure of anything. The standard way to score a probabilistic

forecast that named one of eleven outcomes is to ask how much probability it put on the thing that actually

happened, and compare that against knowing nothing — which here means 1/11, or 9.09%, on every country.

In log-loss terms (lower is better), knowing nothing scores 2.398.

So the winner book, the one that “got it wrong”, was in fact substantially better than

chance. It put the eventual champion second of eleven and gave her nearly three and a half times the

no-information probability. A market that names the winner as its second pick at 31% has done real work.

The public-vote book was the best of the three by a clear margin, and it was the one

that moved latest: the Philippines took that book back on 13 November, 34.5% to 46.5% in a day, a

seventeen-point swing in the gap. CARMELLE then won the public half by ten points, 118 to Thailand’s 108.

Whatever that money was reading, it was reading it correctly.

The jury book was worse than knowing nothing

And then there is the jury winner market. It closed with Malaysia at

72.0% — its high for the week — and the Philippines at 2.8%. The

Philippines topped the juries on 93 points. Malaysia was third, on 82.

A trader who ignored this book entirely and assigned 9.09% to all eleven countries would have scored better

than the market did. That is the finding worth keeping from this contest: not that a favourite lost, but that

one specific book carried negative information, and it is the book that should structurally have been

the best informed of the six.

Here is why that is damning rather than merely unlucky. The eleven national juries did not vote during the

live show. They voted at the Jury Final on Friday 13 November, watching the second dress

rehearsal and ranking the songs there. This site made that point in the closing post before the final: by the

time the 17:04 UTC reading was taken, the jury half of the Eurovision Asia 2026 result already existed on

paper. The jury book was not pricing an uncertain future event. It was pricing an unpublished past one,

and it priced it at 2.8% against 72.0% the wrong way round.

The lesson generalises to every future edition, and it is the opposite of the intuitive one. A market on a

resolved-but-unpublished outcome is not better informed because the event has happened — it is

worse informed, because no new public evidence can arrive to correct it. The price just sits there

compounding whatever the crowd already believed. Rehearsal clips and betting momentum had convinced that crowd

Malaysia was a jury act. Nothing after Friday afternoon could have told them otherwise.

The place books, and the one that went missing

The three placing books did better than the split books, with one exception. The

Top 3 market resolved on the Philippines, Thailand and Vietnam. It had the Philippines

and Vietnam comfortably, and Thailand at 19.0% — having marked the host nation down from 35.0% in a

single day, a move this site published as unexplained because no incident, staging problem or negative

rehearsal report accounted for it. That move cost the book its third leg.

The third-place book got Vietnam right. The second-place

book — the one this site declined to quote for four consecutive days — landed on Thailand, from

12.0% at the close. It was the right answer from a book that had Nepal quoted at 9.85% for second and 0.15% for

third, a ratio of about 65 to 1 on under $100 of volume. Nepal finished ninth. Four days of refusing to put

that book in front of readers looks, in hindsight, like the correct editorial call rather than excessive

caution.

That is the general shape of it: the books with real two-sided liquidity were broadly sound, and

the thin ones were decorative. It is the same conclusion this site reached about Kalshi’s

eleven-contract board, where seven contracts had no bid at all and a last price with nothing underneath it is

an ask being lifted, not a market view.

Finals day turnover: the books nearly doubled

Total turnover across the six books went from about $147.4K at the close on Friday to about

$269.7K by 17:04 UTC on Saturday — up about $122.3K, or 83%, in 24 hours, after a

previous record single-day rise of $43.4K. The winner book alone took $82.7K of that, more than doubling.

Most of that is not forecasting; it is settlement. Once the result is known, holders of losing contracts sell

at a fraction of a cent and holders of winning ones buy up to parity, and that churn prints as volume. The

proportions are still worth noting: the second-place book, the thinnest and least trustworthy all week, more

than doubled too, which is what a book looks like when it finally has something unambiguous to resolve against.

The unsettled half: no official scoreboard

One thing is still open as this is published, and it is the direct consequence of a risk this site flagged

before the final. The organisers announced CARMELLE as the winner but published no official

scoreboard — no per-country jury and public breakdown. The eleven-country split everyone is

using, including us, comes from the fan press; we explain in the

full result and scoreboard piece why it is

nevertheless trustworthy, and it comes down to the jury column summing to exactly 638 and the public column to

exactly 696, which are the structural maxima the EBU itself published on 9 November.

But a prediction market cannot settle on a plausibility argument. It settles on the official ranking as

published, and for the two split books there is nothing official to settle against. You can read the hesitation

straight off the screen: three hours after the show the public-vote book was quoting the Philippines at about

84.6% rather than 100%, on an outcome the available scoreboard calls by ten points. Roughly

$51K of positions across the jury and public books are waiting on a document that may never appear.

That is a disclosure failure, not an integrity failure. Nobody is alleging the count was wrong. But a contest

that published a full scoreboard for only one of its eleven national selections all season has now carried that

habit into its Grand Final, and it is the single most fixable thing on this list before 2027.

Where the two halves actually diverged

One more pattern is worth pulling out of the scoreboard, because it is the thing the markets were trying to

price all week and never resolved. Four countries scored materially better with the public than with the juries:

Laos (78 public against 61 jury), Bhutan (71 against 55), Thailand (108 against 62) and the Philippines (118

against 93). Three went the other way: Vietnam was almost exactly level (83 against 81), while Mongolia (57

against 37) and South Korea (61 against 52) were jury-leaning entries.

Malaysia is the interesting case, because it was neither. Aina Abdul took 82 jury points and 70 public

points — third and sixth respectively. The market had spent a fortnight treating her as a jury act,

pushing the jury book to 72.0% while letting her public line sag to 35.0%, and the scoreboard says the juries

did like her: 82 points is a strong jury score. It simply was not a winning one, because two other

countries scored higher with the juries as well. Being right about the direction of an entry and wrong about

its magnitude is the most expensive kind of half-correct there is, and it is what the 60.5% was built on.

Bhutan deserves a line of its own as the night’s biggest overperformance against price. Nyendra was

the longest shot on every board, described on this site as the contest’s San Marino slot, and she took 71

public points — fourth-best of the eleven, one more than Malaysia. No market anywhere priced that as a

live possibility, and no market needed to: it is the tail doing what tails do.

What to carry into Eurovision Asia 2027

Five things worth writing down while they are fresh:

1. Score books, not favourites. The winner market was better than chance even though its

   favourite finished fourth. A market that names the champion second at 31% is working.

2. Distrust a book on an outcome that has already happened in private. The jury market was

   the worst of the six and had the most information available to it. That is not a coincidence.

3. The public half is bigger than the jury half. Twelve public votes against eleven juries,

   696 points to 638, and the public book was the sharpest of the three. Weight it accordingly.

4. A host nation with no home televote is not a dead entry. Thailand took 108 public points

   from everyone except Thailand, and was priced at 1.55%.

5. Ignore any book where the quotes are one-sided. Nepal at 65 to 1 between second and third

   on under $100 of volume told you nothing, in either direction.

The full eleven-country scoreboard, every country’s page and the complete dated record of how these

markets moved through contest week are on the Eurovision Asia hub. For the

European contest, the 2027 winner market and the

Burgas participant list are already live, and the Eurovision Asia 2026 winner

now has a performance slot there.

Prices in this article are Polymarket midpoints recorded by this site at 17:04 UTC on 13 November 2026 and

re-read at 17:04 UTC on 14 November 2026 via the Gamma API. The scoreboard is from

Eurovoix

and ESCplus, 14 November 2026. Prediction markets are not fixed-odds bookmakers and no UK bookmaker priced this

contest; see our bookmaker guide for what is and is not available.