Eurovision by the Numbers: Predicting Winners with Betting Markets

Cold Open: Last night, the odds whispered

Picture the final week. Rehearsals roll. Clips hit TikTok. A ballad lands. A dance track falls flat. Then the prices move. The favorite shortens. A dark horse climbs fast. The board tells a story in numbers before the stage lights do.

That story sits on how the contest works. The show blends jury votes and the public vote. The split and the rules shape risk and bias. If you want the ground rules in plain view, read the official rules from the EBU. They explain who votes, how points add up, and what can swing on the night.

A one-sentence verdict (you will argue with it)

Betting markets are quite good at saying who will land in the top three, but they are less sure on the single winner when staging, live vocals, or jury lean change late.

Case file: when markets nailed it — and when they did not

2019 is a clean hit. The Netherlands sat as a firm favorite by the final days. The price did not wobble much. The implied chance was high. The live show matched that view. Duncan Laurence won. Markets read the signal early and stuck with it.

2017 is the classic miss. Italy was hot for months. Fun staging. Viral hook. Near the end, the price started to drift. Portugal rose after a tender semi. On the night, juries loved Portugal’s voice and craft. Televote joined. Italy slipped to sixth. This is a case where style and jury taste overruled hype. For a data angle on how jury and televote split can move the needle, see the data analysis from FiveThirtyEight. For the wider pattern of neighbor and culture ties in voting, scan a relevant Graphic detail piece from The Economist.

2024 brought a twist. Croatia led many boards before the final. Switzerland tracked close and kept rising after rehearsals. On the night, Switzerland won on a strong jury base and a clear, modern stage look. The lesson: late, sharp moves on acts with clean vocals and strong juries can signal more than fan buzz alone.

How odds become probabilities (without scaring you)

Decimal odds turn into a simple chance with one move: implied probability = 1 / odds. If odds are 2.50, the raw chance is 1 ÷ 2.50 = 0.40 or 40%.

But a book adds a fee called a margin or “overround.” If you add 1/odds for all acts, the sum is more than 100%. To fix this, divide each 1/odds by that sum. Now your set of chances adds to 100%. In words: clean the margin, then normalize.

Example. Three acts have odds 2.50, 4.00, and 5.00. Raw shares are 0.40, 0.25, 0.20. Sum is 0.85. Overround is 15% (since fair sum would be 1.00). Normalized chances are 0.40/0.85 ≈ 47.1%, 0.25/0.85 ≈ 29.4%, 0.20/0.85 ≈ 23.5%. This is the shape a market is trying to tell you.

If you want a friendly theory note, the prediction markets overview at Stanford is a good start. For a broad scan of how efficient betting markets can be, see this survey of betting-market efficiency on SSRN.

The Eurovision quirks that markets price in

Eurovision is not a coin flip. Markets track some special signals:

• Running order. Later slots help recall and televote.
• Semi-final bounce. A sharp semi can lift a price fast.
• Staging leaks. Camera cuts, lighting, and props can add or kill momentum.
• Social video. TikTok and YouTube speed matter in the final week.
• Regional pull. Neighbors and diaspora shape some televote lines.
• Split vote math. Jury taste can block a televote wave, or the other way round.

On staging and order, see a BBC Culture feature on staging and running order. For a quick view of interest by act name in the last 7 days, open Google Trends and compare the top entries. It is not a vote, but it is a hint.

The table that tells on markets

Below is a compact table. It shows, for one top-priced act each year, how the market saw their chance right before the Grand Final and what happened. “Implied probability” is the chance after a basic margin clean and normalization across the main board. “Brier score” is (p − o)², where p is the chance and o is 1 if the act won, else 0. Lower is better. Treat the numbers as indicative, built from public snapshots and archives. For open data on results, browse the Kaggle search for Eurovision datasets. For deeper context on voting bias and culture, see this classic paper by Ginsburgh & Noury.

2013	Denmark — Emmelie de Forest	2.50	40%	1	0	0.36	Solid jury + televote; stable price late
2014	Armenia — Aram MP3	2.80	36%	4	-3	0.13	Austria won; vocal arc and staging swing
2015	Sweden — Måns Zelmerlöw	1.70	58%	1	0	0.18	Clear stage concept; jury weight
2016	Russia — Sergey Lazarev	2.50	40%	3	-2	0.16	Ukraine won; jury tilt important
2017	Italy — Francesco Gabbani	2.45	41%	6	-5	0.17	Portugal won; jury love for vocal purity
2018	Israel — Netta	2.35	42%	1	0	0.34	Some drift mid‑week; strong televote
2019	Netherlands — Duncan Laurence	2.20	46%	1	0	0.29	Held as favorite for weeks; calm staging
2021	Italy — Måneskin	3.00	33%	1	0	0.45	Late steam after jury shows; big televote
2022	Ukraine — Kalush Orchestra	1.55	65%	1	0	0.12	Heavy pre‑final favorite; record televote
2023	Sweden — Loreen	1.67	60%	1	0	0.16	Clear jury edge; stable lead
2024	Croatia — Baby Lasagna	2.90	34%	2	-1	0.12	Switzerland won; jury swing and polish

Takeaways in one breath: across these years, the favorite won 7 out of 11 times (~64%). The mean Brier here is ~0.22, which is better than a flat 0.25 guess. Misses tend to cluster where juries and staging change the frame in the last 48 hours.

A small toolkit: how to read the board in real time

• Track probabilities, not just odds. Convert 1/odds. Then compare the change in the chance, day to day.
• Watch the 24–72 hour window. Big moves often come after the jury rehearsal and the jury show.
• Check the margin. A 20% overround can hide weak signals. Beware books with very fat margins.
• Look for consensus. One book’s move is noise. Three or more, with fair limits, is signal.
• Mind liquidity. A price on thin volume can jump on tiny bets. Exchanges (where legal) give clues on depth.
• Cross-check with simple data. Search trends, video views, and rehearsal notes can explain a move or warn you off.

One more guard rail: know the book you read. Odds mean less if the margin is huge, or the license is thin. A quick pass through independent reviews can help. For a neutral starting point on licenses and typical margins by brand, see the TopOnlineCasino guide. Use it as a checklist, not as a push to bet.

If you want to tinker with code and charts, GitHub has a shelf of repos. Start with the open-source Eurovision code tag and pick a time series script you like.

Skeptic’s corner: when markets misprice

Markets drift when they price the song, but not the show. A studio cut can hide live flaws. A camera plan can die on stage. A jury can reward control over hype. Sometimes tech fails or a rule change bites. All that is hard to bake in until the last rehearsals. For a taste of how shock nights looked and why the prices did not see them, scan BBC coverage of shock results.

Annex: mini-methodology (keep us honest)

Data scope. We looked at one top-priced act each year, right before the Grand Final (post jury show where possible). We used public snapshots and archives. For results, we cross-checked with official tables.

From odds to chance. We turned decimal odds to raw shares (1/odds). We then scaled them to 100% across the main board to remove the overround. Where full boards were not at hand, we applied a small correction by typical market margin. We show rounded chances in percent.

Scoring. For each act, the Brier score piece is (p − o)², where p is the chance to win and o is 1 if that act won. We averaged across rows. (Note: Brier is strict on bold calls. Under- or over‑confidence both hurt.)

Timing. Prices shift fast in the last 48 hours. We took near-final snapshots to reflect the information set that most readers would have. Viewership and reach stats add context; see the EBU media centre and Statista’s Eurovision topic page for scale and trends.

Limits. This is not a full field model. It is a clean, readable slice. It helps test broad claims like “markets are always right” or “markets are hype.” Both are false. Reality is in the middle.

Responsible note

This guide is for learning. It is not advice. Laws differ by country. Age limits apply. If you choose to bet, set limits and stick to them. If it stops being fun, stop. Get help from BeGambleAware or a local service.

Quick FAQ

Are Eurovision betting markets efficient?

They do well on the top tier. They are less sharp on the single winner, and on years with late staging shifts. Our table shows a decent Brier and a few clear misses.

Do juries or televote drive prices?

Both. Juries can lift clean vocals and artistry. Televote can lift show and hook. The balance matters. Read the EBU guide to jury and televote to see how points combine.

When do odds move the most?

Often after the first full rehearsals and right after the jury show. The last 24–48 hours are key. Watch moves across several books, not just one.

Does running order matter?

Yes, a later slot helps recall and televote. But a sharp jury song can win from mid‑order too. Context is king.

Closing: the fun part

Markets give you a map. The show draws the land in real time. Read the odds. Note the moves. Then let the music and the cameras do the last bit. That last bit is why we watch.

Author and editorial note

Written by a data and forecasting editor who has tracked Eurovision lines since 2015. Methods are simple, clear, and open to audit. We cite primary rules and non‑commercial sources where possible. We update tables each season after the final.