The Long Wrong, The Smart Exit, and the 25-Minute Truth

Abstract

For the sixty days running into Eurovision 2026, three independent betting systems, Polymarket, Kalshi, and the aggregated European bookmaker consensus tracked by Eurovoix, all priced Finland as the runaway favourite at 36 to 40% implied probability. Bulgaria, the eventual winner, sat at 2 to 3% across every system for most of that window. The three crowds converged on the same wrong answer, with Polymarket systematically more bullish on Finland than the bookmaker books by 6 to 8 percentage points throughout the cycle and 10 to 11 points in event week.

This study reconstructs the full cycle from public API data at minute and second resolution. It finds three things. First, the wisdom-of-crowds failure on Finland was industry-wide and not platform-specific, but with measurable cross-platform disagreement that itself is signal. Second, the only country market showing meaningful informed-flow activity on the Saturday of the broadcast was Finland: six documented USD volume shocks beginning at 15:36 UTC, more than seven hours before the on-stage reveal that would crush the favourite. Informed money was exiting the wrong answer, not picking the long-shot winner. Third, when the broadcast began, the prediction markets repriced Bulgaria's win in a precise sequence: Kalshi Jury at 22:25:15 UTC, Polymarket Jury at 22:26:04 UTC, both Winners crossed 50% within three seconds of each other at 22:34 UTC, and Televote followed at 22:55 UTC. The Winner-market certainty arrived approximately 25 minutes before the stage announcement.

The piece is the inaugural publication of Tater Research, the cross-platform analytical layer Tater is building across the prediction-market and sportsbook universe.

Key findings

• For 60+ days, three independent betting systems converged on the wrong answer for Finland. Polymarket peak 40.45% (10 April 2026), bookmaker consensus peak ~31%, Kalshi tracking Polymarket. Finland finished outside the top 10.
• Polymarket was systematically more bullish on Finland than the bookmaker consensus by 6 to 8 percentage points across the cycle, widening to 10 to 11 points in event week. The three systems agreed on every other tracked country within 2 percentage points.
• The Saturday rehearsal-window informed-flow signature appeared on Finland, not Bulgaria. Six documented USD volume shocks on Kalshi Finland Winner throughout 16 May 2026 (15:36, 19:14, 20:51, 21:17, 21:18, 22:14 UTC). Zero comparable pre-broadcast shocks on Israel, Australia, Romania, or Bulgaria.
• Bulgaria's own price signature was post-reveal absorption, not anticipation. Her first USD volume shock came at 22:38 UTC, four minutes after the Winner market crossed 50%.
• During the live broadcast, prediction markets repriced Bulgaria's win in a precise documented sequence: Jury market first (22:25 to 22:26 UTC), Winner market eight minutes later (22:34 UTC), Televote market a further 21 minutes after that (22:55 UTC). Cross-platform agreement on the Winner market crossing was within three seconds.
• Market certainty arrived ~25 minutes before the official stage announcement on the Winner market, and ~34 minutes before on the Jury market.

1. Introduction

On the night of 16 May 2026, Bulgaria's Dara won the Eurovision Song Contest with the largest winning margin in the contest's history (173 points) and the rare feat of taking both the jury vote and the public televote outright. Eurovision is a singular cultural event: high public attention, objective resolution mechanism (audited jury and televote scores), and binary outcome (one winner among 25 contestants). It is also one of the most liquid music-prediction markets in the world. The Polymarket Winner event alone moved more than 200 million USDC across its sub-markets across the cycle.

Eurovision is interesting to Tater Research for three reasons.

First, the objective-resolution structure makes it an ideal stress test of how prediction markets process publicly available information. Bookmakers, regulated prediction markets, and crypto-native prediction markets all priced the same event, all using the same publicly available signals. Where they agreed and disagreed is data.

Second, the contest's well-defined broadcast structure (performances, jury reveal sequence, televote reveal sequence, winner announcement) provides a sequence of discrete information-release events against which market repricing can be measured at minute and second resolution.

Third, the result was a surprise. The runaway pre-show favourite (Finland at 36 to 40% across all systems) finished outside the top ten. The eventual winner (Bulgaria at 2 to 3% across all systems for most of the cycle) was beneath the reporting threshold of the bookmaker consensus for nine of the seventeen dated snapshots covering the run-up. A surprise outcome is the cleanest test of how markets handle out-of-distribution events.

This study reconstructs the entire arc at the resolution the public APIs permit. The dataset, methodology, and all derived computations are publicly available for independent verification.

2. The Long Wrong: how three systems missed Finland together

For the sixty days running into the final, the betting systems were aligned on the wrong answer.

Polymarket Finland-yes peaked at 0.4045 implied probability on 10 April 2026, sustaining 0.36 to 0.40 through March and April. Kalshi Finland Winner tracked the same level on roughly 1/29th the dollar volume. The Eurovoix bookmaker consensus, aggregated from 15 international books including bet365, William Hill, Unibet, Betfair, Paddy Power, Ladbrokes, Sky Bet, Coral, and Betway with overround removed, peaked at approximately 0.31.

Three-line chart showing Polymarket, Kalshi, and Eurovoix consensus implied probability for Finland to win Eurovision 2026, with daily resolution over the 76-day pre-show window. X-axis: dates. Y-axis: implied probability 0% to 50%. Source line at bottom citing all three APIs.

The disagreement between the three systems is itself the more interesting story. Polymarket sat consistently 6 to 8 percentage points above the bookmaker consensus on Finland throughout the cycle, widening to 10 to 11 points during event week. The same three systems agreed on every other tracked country (Bulgaria, Israel, Romania, Australia, Italy) within 2 percentage points.

This is not a calibration story about one platform being uniquely wrong. It is a story about three different audiences with three different risk preferences pricing the same event with one systematic, country-specific divergence. Polymarket's crypto-native retail crowd was more seduced by Finland's hype than the professional bookmakers were. The bookmakers, who set their consensus through margin-removed mid-prices on real-money books with risk management overhead, were less aggressive on Finland but still wrong by an enormous margin.

Two-line chart showing Polymarket Finland-yes vs Eurovoix consensus Finland implied probability, with the gap (in percentage points) shaded as a third series. X-axis: dates 1 March to 16 May. Y-axis (left): implied probability. Y-axis (right): gap in percentage points.

Bulgaria meanwhile was barely visible. She sat at 0.02 to 0.03 implied probability across Polymarket and Kalshi for most of the cycle. On the bookmaker side, she appeared in only 8 of the 17 dated Eurovoix consensus snapshots covering the window, because she was beneath the reporting threshold. The Saturday afternoon Eurovoix bulletin had her at 7.3%, third behind Finland (39.4%) and Australia (23.5%). Still not the favourite, just no longer invisible.

The wisdom of crowds did not work here. Not because one platform was uniquely bad, but because all three were drinking from the same well: pre-show signals (rehearsal coverage, fan polls, song popularity metrics, the conventional Eurovision wisdom) all pointed at Finland. Bulgaria's return to the contest after a three-year absence, and Dara's performance arc through the contest week, were not visible enough in those signals to move the line.

3. The Smart Exit: where informed flow appeared on Saturday

The most surprising finding in this dataset is also the most defensible.

On 16 May 2026, the day of the final, we screened every country market on Kalshi Winner for elevated volume activity. The rule for what counted as a meaningful flow event: a one-minute bucket of USD-notional trading volume exceeding $5,000 AND at least three times the trailing 10-minute median. Independent of price direction.

Across the six country markets we examined (Bulgaria, Israel, Romania, Finland, Australia, Italy), only one country tripped the filter during pre-broadcast hours: Finland.

Six documented USD volume shocks on Kalshi Finland Winner across 16 May 2026:

The 15:36 UTC shock is the spine of the finding. It is more than seven hours before the on-screen reveal that would crush Finland's chances, and well inside the window when the Saturday jury rehearsal show was concluding in Vienna. Anyone present at or proximate to the rehearsal would have known by that point what the contest's pattern would be. The Kalshi data shows roughly 24,000 USD of notional volume flowing through Finland Winner in that single minute, 55 times the trailing baseline.

Trade-level inspection of the 15:36 minute and the others shows the flow direction: in every Finland shock, the dominant taker side was the YES book getting hit (sellers of Finland-yes), and the price moved down. This is consistent with an actor reducing exposure to Finland, not adding to it.

Two-panel chart. Top panel: Kalshi Finland-yes price line over the day in 1-minute resolution. Bottom panel: USD notional volume per minute as bars. Vertical markers at each of the six shock timestamps. Annotations: 15:36 ("Vienna jury-rehearsal window"), 22:14 ("Pre-jury reveal").

The control case is the same screen applied to Bulgaria, Israel, Australia, and Romania across the same window. Zero pre-broadcast shocks tripped the threshold on any of them. Bulgaria's first USD shock of the day did not come until 22:38 UTC, four minutes after her Winner market had crossed 50%. By that point the jury reveal had largely happened and Bulgaria's lead was visible in the broadcast. The flow into Bulgaria-yes at 22:38 UTC is best read as the market absorbing the new public information, not anticipating it.

Five-line small-multiples chart showing 1-minute USD volume for Bulgaria, Israel, Romania, Australia, Finland on Kalshi Winner across 12:00 to 23:00 UTC on 16 May 2026. Finland panel highlighted to show the six shocks. Other four panels essentially flat through pre-broadcast hours.

This is not the romantic story we expected to find when we started this analysis. The intuitive prior was that any informed flow during the rehearsal window would be loading up on Bulgaria, the surprise winner. The data says otherwise. The informed actor was not picking the long-shot winner. The informed actor was selling the popular loser. That is more aligned with how betting markets actually work: smart money exits the wrong answer before the crowd does. Smart money rarely picks the 2% upset because the 2% upset is, by definition, hard to identify.

We cannot identify the actor or actors behind the Finland flow from public data. Kalshi does not expose taker identity at the wallet level. What the data shows is that the activity was specific to Finland, concentrated in a window that aligns with the Vienna jury-rehearsal cycle, sustained through six discrete time-stamped events across the day, and entirely absent from every comparable country market we examined.

4. The 8-Minute Truth: cross-market lead-lag during the broadcast

When the broadcast finally revealed Bulgaria's win, the prediction markets repriced in a precise sequence that we can document to the second.

The Eurovision Grand Final broadcast began at 19:00 UTC on 16 May 2026. By 21:25 UTC all 25 performances were complete. Public voting closed at 21:59 UTC. The jury reveal sequence began at approximately 22:05 UTC.

Across the four key markets relevant to this analysis:

Three observations.

The Jury market repriced first. Kalshi led Polymarket by 49 seconds on the Jury market crossing. The Winner market crossings on both platforms agreed within 3 seconds. This is consistent with Kalshi's smaller order book repricing faster on the same incoming information rather than with one market having genuinely earlier information. Cross-platform agreement on the Winner crossing to within 3 seconds is strong evidence of efficient information flow across the two largest prediction-market venues.

The cross-market lead-lag was 8 minutes between Jury and Winner. The Jury market priced in Bulgaria's victory eight minutes before the Winner market did. This corresponds to the information processing time required for the market to translate the jury-vote standings (where Bulgaria had taken a commanding lead, eventually finishing first on jury with 204 points) into an updated belief about the overall contest outcome.

The Televote market lagged the most. The Bulgaria Televote market did not cross 50% implied probability until approximately 22:55 UTC, when Bulgaria's televote score was actually being revealed on stage. This is consistent with the Televote market repricing only after the on-screen reveal of the televote total made the outcome unambiguous.

Three-line chart showing Bulgaria-yes implied probability across Polymarket Jury, Polymarket Winner, and Polymarket Televote markets at 1-minute resolution. Optional fourth and fifth lines for Kalshi equivalents. Vertical markers: 22:05 (jury reveal begins), 22:26 (Jury market crosses 50%), 22:34 (Winner market crosses 50%), 22:55 (Televote market crosses 50%), 23:00 (stage announcement).

Threshold sensitivity. The 8-minute Jury-to-Winner lead is robust at the 0.50 and 0.60 crossing thresholds. At 0.40 the relationship breaks down because pre-reveal noise dominates the threshold (the Winner market actually crosses 0.40 before the Jury market because of order-book noise). At thresholds above 0.70, the Jury market's lead grows beyond 10 minutes because the Jury market reaches certainty faster than the Winner market does. The 50% threshold is the appropriate analytical anchor: it represents the moment a market transitions from "leaning that way" to "leaning the other way."

Small table chart showing the Jury-to-Winner lead time at thresholds 0.40, 0.50, 0.60, 0.70, 0.80, 0.90. Annotations: which thresholds are robust, which are noise-dominated.

Anyone watching only the Winner market saw Bulgaria's win priced in twenty-five to twenty-six minutes before the stage announcement. Anyone watching the Jury market saw it thirty-three to thirty-five minutes early. The lead is not currency-arbitrage-grade alpha (the markets are too small), but it is a real information advantage measurable in minutes for anyone with cross-market visibility.

5. The Honest Reveal: what Bulgaria's price actually did

A piece like this risks overselling what prediction markets did on the night.

The 25-minute lead between Winner-market certainty and stage announcement is not "prediction markets are clairvoyant." It is "prediction markets are efficient at processing the broadcast in real time." The information was on screen for traders to see (the jury totals coming in country by country, the running scoreboard, the impossibility-by-arithmetic of any non-Bulgarian winner once the jury totals were locked). The traders priced it in. That is information-processing speed, not predictive power.

Bulgaria's own volume signature confirms this. Her first genuine USD volume shock did not come during the rehearsal window or in the run-up. It came at 22:38 UTC, four minutes after her Winner market crossed 50%. The flow into Bulgaria-yes at that point was traders absorbing the new public information that had just been revealed, not anticipating it.

The honest framing of the night is therefore:

• Prediction markets failed at predicting Bulgaria for sixty days. Three independent betting systems collectively assigned her a 2 to 3% probability for most of the cycle. She won by the largest margin in Eurovision history.
• Prediction markets succeeded at processing the broadcast in real time once the public information began arriving. The lead-lag between the Jury, Winner, and Televote markets follows the information sequence on screen with documented precision.
• One actor (or one small group of actors) appears to have been exiting Finland hours before the broadcast made the collapse public. That signature is real and time-stamped.

Both failure and success are useful information. Prediction markets are excellent at processing public information into price. They are less good at picking surprise winners. They sometimes reveal informed flow that the public scoreboard does not. All of these are useful properties to know about, and none of them require overclaiming.

6. Discussion: implications for the prediction-market and sportsbook industries

Three implications worth highlighting.

One. Cross-platform analysis is the missing layer. Every claim in this study required integrating data across at least two venues, and most required three (Polymarket, Kalshi, and the Eurovoix bookmaker consensus). The Polymarket-vs-bookmaker Finland disagreement, the cross-platform agreement on the Winner crossing to within three seconds, the comparison of pre-broadcast volume signatures across countries. None of these are visible from inside a single venue. The structural opacity between venues is itself the gap that needs closing.

Two. The wisdom-of-crowds framing has limits. The Eurovision case is a clean example of three independent systems converging on the wrong answer for sixty days. This is uncomfortable for the broader prediction-market industry narrative that markets are systematically more accurate than polls, experts, or pundits. The honest finding is more nuanced: markets are good at processing public information in real time, and they sometimes reveal informed flow that polls do not. They are not systematically good at identifying low-probability events that turn out to win.

Three. Informed-flow signatures may be a useful real-time indicator across event categories. The Finland 15:36 UTC shock, replicated across six discrete time-stamped events on a single market with zero comparable activity on any of five control country markets, is the kind of pattern that should be possible to detect in real time. If similar signatures appear in election markets, in macro markets, in awards-show markets, the same screening rule (1-minute USD bucket above threshold AND above 3x trailing baseline AND in one market specifically vs no others) might surface informed flow as it appears. This is one of the directions Tater Research will pursue in subsequent publications.

7. Methodology

Data sources:

• Polymarket Gamma API (gamma-api.polymarket.com): event and sub-market metadata
• Polymarket CLOB API (clob.polymarket.com): historical prices via /prices-history at fidelity 1 minute, paged at 15-day windows for 90-day historical coverage
• Polymarket data API (data-api.polymarket.com): trade-level data, paged via offsets 0/1000/2000/3000 within the API's 3,000 offset cap
• Kalshi Trade API v2 (api.elections.kalshi.com/trade-api/v2): candlesticks and trades
• Eurovoix bookmaker consensus (eurovisionworld.com/odds mirror via Eurovoix): dated daily bulletins covering 1 March 2026 to 16 May 2026

Time normalization: All timestamps are UTC. The broadcast event timeline (jury reveal start, jury reveal end, televote reveal, winner announcement) was triangulated from the SBS Australia live blog, BBC, Eurovoix, Wiwibloggs, and the official EBU press materials. Six per-country televote reveal timestamps are at high confidence (sourced directly from SBS). The remaining nineteen per-country televote timestamps are piecewise-linearly interpolated between the six high-confidence anchors and are footnoted as such throughout.

Probability and price conventions: Polymarket and Kalshi prices in this study are expressed as implied probabilities (price between 0 and 1) of the Yes side. The 50% crossing is defined as the first sample strictly greater than 0.50 in the price series. For Polymarket, we use the p field from /prices-history. For Kalshi, we use the trade-level yes_price field for second-level resolution and the candle close_dollars for minute-level resolution. The two are noted where they disagree on the second-level claims (e.g. the Kalshi Jury crossing time differs by approximately 45 seconds between trade-level and candle-close measurement; we report the trade-level number as primary).

Volume normalization: Kalshi candle volume in the API is expressed in contracts, not USD notional. To compute USD notional volume, the contract count is multiplied by the candle's mean dollar price. An earlier internal version of the dataset incorrectly labeled the contract count as USD volume; the bug was identified during the pre-publication verification audit and is documented in the public repository as a transparency record. All numbers cited in this study use the corrected USD-notional values.

Volume shock definition: A volume shock event is defined as a 1-minute USD-notional volume bucket exceeding $5,000 AND exceeding three times the trailing 10-minute median. The $5,000 floor is intended to exclude statistical noise on thin pre-broadcast minutes. The 3x ratio is intended to identify genuine excess flow against a moving baseline. Sensitivity to these parameter choices is documented in the appendix.

Cross-platform alignment methodology: Cross-platform comparisons are performed at common UTC timestamps. Where the two platforms publish at different resolutions (Polymarket minute, Kalshi trade-level), the higher-resolution series is downsampled to the lower-resolution grid for direct comparison and the alignment method is reported in the relevant figure caption.

Bookmaker consensus methodology: The Eurovoix consensus represents an aggregate of 15 European bookmaker yes-prices with overround removed via the standard normalization (each book's prices divided by the sum across all countries to produce a proper probability distribution). The Eurovoix methodology is comparable but not identical to ESCToday's tracker (which was the originally intended source but was blocked by Cloudflare during data collection); the methodology divergence is documented in the repository.

8. Limitations and alternative interpretations

On the duration of the Finland mispricing. The 60-day Finland favourite consensus may reflect a single high-variance draw from a well-calibrated long-run distribution rather than systematic Bayesian error. We cannot reject this with one event. Multi-year analysis across multiple Eurovision contests and other entertainment markets would be required to make the calibration claim rigorous.

On the interpretation of the Finland volume shocks. The 15:36 UTC and subsequent Finland shocks are consistent with informed flow exiting the favourite, but several alternative interpretations exist. A single large trader rebalancing for unrelated reasons would produce a similar signature. Cross-venue arbitrage from a different exchange or platform could appear as one-sided volume on Kalshi. Market-maker rebalancing for upcoming weekend illiquidity is plausible. Random Saturday noise that happens to clear the threshold cannot be entirely ruled out. We have chosen the most parsimonious reading consistent with the trade-level evidence (the specificity to Finland, the absence on control markets, the dominant taker-side direction). Readers are free to reach different conclusions from the same data.

On the Jury-leads-Winner finding. The 8-minute lead of the Jury market over the Winner market may reflect order-book depth and reactivity asymmetries rather than informational asymmetry. The Jury market is structurally less liquid than the Winner market on both platforms, and a less liquid market can reprice faster on the same incoming information. The honest framing is "the Jury market repriced first" rather than "the Jury market knew first."

On the cross-platform agreement to within seconds. The 3-second agreement between Polymarket and Kalshi on the Winner crossing is consistent with both efficient cross-platform arbitrage AND independent reaction to the same broadcast trigger by traders on each venue. The data does not distinguish between these. Either reading is interesting; the first implies sophisticated arbitrageurs operating across both venues, the second implies efficient processing of the same external signal by independent populations.

On country selection. The country sample (Bulgaria, Israel, Romania, Finland, Australia, Italy) was chosen as the top five finishers plus the pre-show favourite. Several non-finalists had higher Polymarket Winner volume than included countries (notably Poland and Austria). Spot-checks confirm that none of the excluded countries' Winner prices ever crossed 5% implied probability during the broadcast, and none of the excluded countries' Kalshi markets showed pre-broadcast volume shocks under the threshold criteria. The selection therefore does not undermine the headline findings, but the choice itself is acknowledged.

On bookmaker source pivot. ESCToday was the originally intended source for the bookmaker consensus. Their tracker is Cloudflare-protected and was inaccessible during data collection. Eurovoix's tracker is methodologically comparable (both publish overround-removed mid-prices from comparable 15 to 26 book panels), but the swap mid-pull is documented and the implications discussed in the repository.

On the units bug. An earlier internal version of the dataset incorrectly labeled Kalshi contract counts as USD volume. The bug was caught during pre-publication verification, the correction is documented, the pre-fix data is preserved as a transparency record, and all numbers in this study are computed from the corrected dataset. We mention it explicitly because methodology transparency is the bar Tater Research operates under.

9. Reproducibility

Every claim in this study can be reproduced from the public dataset published on this page and the source APIs.

Public dataset: all underlying data is downloadable directly from this article. Chart-by-chart CSVs are linked under each chart. Inline-table CSVs are linked under each table. A consolidated data index appears in the Data section below.

Reproducibility instructions: a step-by-step guide for re-querying each source API to derive the cited numbers is provided on the methodology page at taterit.com/research/eurovision-2026/methodology.

Independent reproduction: during pre-publication verification, an independent agent operating with no access to our cached data reproduced the central timing claim (Bulgaria Winner crossing 50% at 22:34 UTC) from the public APIs end-to-end. The reproduction was successful to within 1 second.

10. Data availability and disclosures

Data availability. All data underlying this study is publicly available from the source APIs and is downloadable directly from this article. Per-trade wallet addresses on Polymarket are hashed (SHA-256, deterministic but irreversible) in the published dataset to prevent identification while preserving the ability to verify aggregate flow signatures.

Conflict of interest disclosure. Tater (taterit.com) is a prediction-market and sports-betting discovery layer that surfaces Polymarket and Kalshi markets in its product. Tater has a commercial interest in the markets discussed in this study. The dataset and methodology are nonetheless drawn from public APIs and the findings are independently verifiable.

Funding. Self-funded. No external research grants or sponsorships were involved in this study.

Personnel. The dataset was assembled, audited, and documented by the Tater team between 17 May and 19 May 2026. The full verification audit (independent reproduction, sensitivity testing, alternative interpretation enumeration, units-bug correction) is summarized on the methodology page.

11. References

• Bürgi, C., Deng, W., and Whelan, K. (2025). Makers and Takers: The Economics of the Kalshi Prediction Market. CESifo Working Paper No. 12122. SSRN
• Le, N. A. (2026). Decomposing Crowd Wisdom: Domain-Specific Calibration Dynamics in Prediction Markets. arXiv preprint.
• Federal Reserve Board (2026). Kalshi and the Rise of Macro Markets. Working paper.
• Wealt (2026). Kalshi: Building Exchange Infrastructure for a Probabilistic World.
• ESCToday Eurovision Betting Odds Tracker, weekly bulletins, 1 March 2026 to 16 May 2026 (data accessed via cached snapshots; tracker source pivoted to Eurovoix due to Cloudflare block during data collection)
• Eurovoix Eurovision 2026 odds tracker, daily snapshots
• SBS Australia Eurovision 2026 Grand Final live blog (16 May 2026)
• European Broadcasting Union (EBU) press materials, Grand Final 2026

12. About the authors

Ivo Dimitrov

Ivo Dimitrov is the co-founder of Tater (taterit.com), a cross-platform discovery layer for prediction markets and regulated sports betting. He brings eighteen years of operational experience across the iGaming and sportsbook industry, with senior roles at Pragmatic Play Sports, Sportradar, and operator and platform providers.

Stephen Crystal

Stephen Crystal is the co-founder of Tater and the founder and CEO of SCCG Management, a leading gaming industry advisory firm. He has extensive experience advising operators, technology vendors, regulators, and investors across the global gambling and sports betting ecosystem.

About SCCG Management

SCCG Management is a leading gaming industry advisory firm and global connector across the gambling and sports betting ecosystem. Headquartered in Las Vegas, SCCG advises operators, technology vendors, regulators, and investors across North America, Europe, Asia, and Latin America. More at sccgmanagement.com.

About Tater Research

Tater Research is the cross-platform analytical layer Tater Labs is building across the prediction-market and sportsbook universe. This is the inaugural publication. The next case studies will cover the 2026 FIFA World Cup, beginning 11 June 2026, with the analytical layer running live across both prediction markets and US sportsbooks for the first time.

Subscribe to Tater Research for future publications.

Update log

• 19 May 2026, 09:00 CET. Initial publication.

(Future corrections or extensions to be logged here with timestamps.)

Acknowledgments

This work was made possible by the public data infrastructure of Polymarket, Kalshi, and Eurovoix. The Tater team thanks the academic researchers cited above whose prior work informed the methodological framing of this study.