World vs Model World vs Model · World Cup 2026
World vs Model logo

World Cup 2026 — Can a model beat the market?

Two transparent models take on the crowd across all 240 markets — one knows zero football (just price structure), one is informed (Elo ratings). We keep a public, out-of-sample scorecard. How this works →  ·  Glossary & references →
Kicks off today
A research experiment by Marcus Liew ↗  ·  updated 2026-06-11 22:53 UTC 240 markets · 48 teams scored vs the market
How the tournament unfolds today!
Jun 11Group stagetournament opens
Jun 28Round of 32knockouts begin
Jul 9Quarter-finalslast 8
Jul 14Semi-finalslast 4
Jul 19Finalchampion crowned
🔮 Outcome map🏆 Bracket score ⚽ Board💰 Books

The scoreboard — Track record, scored out of sample as results land

THE MARKET
the line to beat
live Polymarket, de-vigged
ZERO-KNOWLEDGE
price structure
knows no football
INFORMED · ELO
ratings + simulation
independent of the market
claims registered
476
timestamped, falsifiable
resolved so far
0
skill is scored, not claimed
status
PRE-KICKOFF
scorecard arms Jun 11

As it unfolds — how the model moved since kickoff, and the biggest shocks

🔮 Lights up once the first games are played. The informed model re-forecasts after every result, so this will show how it changed its mind (title odds rising and falling) and the biggest surprises — with whether the model or the market saw them coming. World Cups always spring a few.

The zero-knowledge model's biggest disagreements

MODEL HIGHER
🇧🇪 Belgium #8
Advance
market 94.9%model 98.7%
+3.8% gap
MODEL LOWER
🇨🇻 Cape Verde #44
Advance
market 30.5%model 28.0%
-2.5% gap
RISKLESS MISPRICING
🇶🇦 Qatar
nested inconsistency
priced higher to 0.011 (reach_F) than to 0.003 (reach_SF)
logically impossible — one of these prices is wrong

Outcome map — the informed model's projection

What the informed Elo model expects, from 20k simulations of the verified bracket — these are the model's probabilities, not the market's. A 20k-run simulation is a distribution, not a single prediction, so we lead with the spread of outcomes and keep the single most-likely bracket for last.

Title race — P(win the cup) across 20k simulations

📊 Most-likely final: 🇦🇷 Argentina vs 🇪🇸 Spain — yet that exact pairing lands in only 4.5% of the runs. No single outcome is likely; the spread is the forecast. These bars are the whole distribution — the bracket lower down is just its one most-likely path.
🇪🇸ESP16.4%
🇦🇷ARG12.1%
🇫🇷FRA9.9%
🏴󠁧󠁢󠁥󠁮󠁧󠁿ENG6.1%
🇧🇷BRA5.1%
🇵🇹POR4.5%
🇨🇴COL4.3%
🇳🇱NED3.9%
🇪🇨ECU3.4%
🇭🇷CRO3.0%
🇲🇽MEX3.0%
🇩🇪GER2.8%

How far each team goes — the full distribution of where the model has each team bow out (champion % at right)

Out in groupsRound of 32Round of 16Quarter-finalSemi-finalRunner-upChampion
🇪🇸ESP16%
🇦🇷ARG12%
🇫🇷FRA10%
🏴󠁧󠁢󠁥󠁮󠁧󠁿ENG6%
🇧🇷BRA5%
🇵🇹POR4%
🇨🇴COL4%
🇳🇱NED4%
🇪🇨ECU3%
🇭🇷CRO3%
🇲🇽MEX3%
🇩🇪GER3%
🇯🇵JPN3%
🇳🇴NOR3%
🇹🇷TUR2%
🇨🇭SUI2%
🇺🇾URU2%
🇧🇪BEL2%
🇨🇦CAN2%
🇸🇳SEN2%
🇵🇾PAR1%
🇲🇦MAR1%
🇦🇹AUT1%
🇺🇸USA1%
🏴󠁧󠁢󠁳󠁣󠁴󠁿SCO1%
🇮🇷IRN1%
🇦🇺AUS1%
🇰🇷KOR1%
🇨🇿CZE0%
🇩🇿ALG0%
🇵🇦PAN0%
🇸🇪SWE0%
🇪🇬EGY0%
🇺🇿UZB0%
🇨🇮CIV0%
🇯🇴JOR0%
🇨🇩COD0%
🇹🇳TUN0%
🇧🇦BIH0%
🇳🇿NZL0%
🇨🇻CPV0%
🇸🇦KSA0%
🇭🇹HAI0%
🇿🇦RSA0%
🇨🇼CUW0%
🇮🇶IRQ0%
🇶🇦QAT0%
🇬🇭GHA0%

The field narrows — who fills each round's slots; one colour per contender, so each is a stream you can follow

ESPARGFRAENGBRAPORCOLNEDECUCROfield
Round of 3232 slotsfield
Round of 1616 slotsfield
Quarter-finals8 slotsESPfield
Semi-finals4 slotsESPARGFRAfield
Final2 slotsESPARGFRAfield
Champion1 slotESPARGFRAfield

Projected group stage (advance %)

top-2 qualify, 3rd may sneak through as a best-third; order is the model's expected finish.

Group A
1🇲🇽 Mexico99%
2🇰🇷 South Korea84%
3🇨🇿 Czechia80%
4🇿🇦 South Africa13%
Group B
1🇨🇭 Switzerland100%
2🇨🇦 Canada99%
3🇧🇦 Bosnia-Herzegovina55%
4🇶🇦 Qatar8%
Group C
1🇧🇷 Brazil99%
2🇲🇦 Morocco89%
3🏴󠁧󠁢󠁳󠁣󠁴󠁿 Scotland79%
4🇭🇹 Haiti8%
Group D
1🇹🇷 Türkiye88%
2🇵🇾 Paraguay73%
3🇺🇸 USA62%
4🇦🇺 Australia55%
Group E
1🇪🇨 Ecuador99%
2🇩🇪 Germany99%
3🇨🇮 Ivory Coast67%
4🇨🇼 Curaçao4%
Group F
1🇳🇱 Netherlands98%
2🇯🇵 Japan94%
3🇸🇪 Sweden51%
4🇹🇳 Tunisia24%
Group G
1🇧🇪 Belgium96%
2🇮🇷 Iran83%
3🇪🇬 Egypt67%
4🇳🇿 New Zealand28%
Group H
1🇪🇸 Spain>99.9%
2🇺🇾 Uruguay97%
3🇨🇻 Cape Verde18%
4🇸🇦 Saudi Arabia16%
Group I
1🇫🇷 France99%
2🇳🇴 Norway90%
3🇸🇳 Senegal81%
4🇮🇶 Iraq7%
Group J
1🇦🇷 Argentina100%
2🇦🇹 Austria77%
3🇩🇿 Algeria51%
4🇯🇴 Jordan30%
Group K
1🇵🇹 Portugal98%
2🇨🇴 Colombia97%
3🇺🇿 Uzbekistan44%
4🇨🇩 DR Congo22%
Group L
1🏴󠁧󠁢󠁥󠁮󠁧󠁿 England100%
2🇭🇷 Croatia98%
3🇵🇦 Panama68%
4🇬🇭 Ghana4%

The single most-likely bracket — one path among many; node shading = % to reach that round, real FIFA R32 slots, model placement

R32R16QFSFFinalChampionFinalSFQFR16R32
🇪🇨ECU
🇺🇸USA
🇫🇷FRA
🇸🇪SWE
🇰🇷KOR
🇨🇦CAN
🇳🇱NED
🇲🇦MAR
🇨🇴COL
🇭🇷CRO
🇪🇸ESP
🇦🇹AUT
🇹🇷TUR
🇩🇿ALG
🇧🇪BEL
🇨🇿CZE
🇪🇨ECU
🇫🇷FRA
🇨🇦CAN
🇳🇱NED
🇨🇴COL
🇪🇸ESP
🇹🇷TUR
🇧🇪BEL
🇫🇷FRA
🇳🇱NED
🇪🇸ESP
🇹🇷TUR
🇫🇷FRA
🇪🇸ESP
🇪🇸ESP
🇪🇸
Spain
🏆 champion · 16%
🇦🇷ARG
🏴󠁧󠁢󠁥󠁮󠁧󠁿ENG
🇦🇷ARG
🇧🇷BRA
🏴󠁧󠁢󠁥󠁮󠁧󠁿ENG
🇦🇷ARG
🇵🇹POR
🇧🇷BRA
🇩🇪GER
🇲🇽MEX
🏴󠁧󠁢󠁥󠁮󠁧󠁿ENG
🇦🇷ARG
🇵🇾PAR
🇨🇭SUI
🇵🇹POR
🇧🇷BRA
🇯🇵JPN
🇩🇪GER
🇳🇴NOR
🇲🇽MEX
🏴󠁧󠁢󠁳󠁣󠁴󠁿SCO
🏴󠁧󠁢󠁥󠁮󠁧󠁿ENG
🇸🇳SEN
🇦🇷ARG
🇺🇾URU
🇵🇾PAR
🇮🇷IRN
🇨🇭SUI
🇪🇬EGY
🇵🇹POR
🇵🇦PAN

This is the modal path — it plays out in only a minority of runs. The title race and how-far-each-team-goes charts above are the fuller, more honest read of the simulation.

↔ swipe the bracket sideways to follow the path to the final

Bracket score — scoring the knockout call, market vs model

The bracket above is a projection; this keeps it honest. Each side fills the bracket from its own probabilities (the top teams per round); a correctly placed team scores that round's weight — Last-32 ×1 · QF ×4 · SF ×8 · Final ×16 · Champion ×32. Forecasts are timestamped pre-tournament; arms at the Round of 32 (Jun 28) — frozen now, scored as it plays.

Market0pts
Informed · Elo0pts

🏆 To lift the trophy — Market: 🇪🇸ESP · Informed · Elo: 🇪🇸ESP (they can agree on the favourite yet score very differently on the path there)

RoundWtWhere the brackets disagree (model picks vs market picks)
Last 32×1agree 30/32 🇵🇦PAN🇦🇺AUS vs 🇩🇿ALGBOS
Quarter-finals×4agree 7/8 🇨🇴COL vs 🇩🇪GER
Semi-finals×8agree 3/4 🇦🇷ARG vs 🇧🇷BRA
Final×16agree 1/2 🇦🇷ARG vs 🇫🇷FRA
Champion×32agree 1/1 identical picks

Only the informed model and the market appear here: the zero-knowledge model just re-shapes the market's own prices, so its bracket is identical to the market's by construction — it competes on calibration (Brier), not on picks. And one bracket is a single high-variance draw, so the round-by-round Brier scores are the meaningful verdict; this points race is the legible one.

The ensemble — a third forecast: Bayesian model averaging

Rather than commit to one model, the ensemble averages both — weighted by each model's per-round Brier track record. The weight at each rung self-corrects: whichever model is better-calibrated at that rung earns a bigger share. Pre-tournament both weight 0.5; tied 50/50 per rung — will start drifting once the first round resolves.

RoundModel confidence (Zero-knowledge ←→ Informed · Elo)Leader
Last 32
50%
50%
tied
Quarters
50%
50%
tied
Semis
50%
50%
tied
Final
50%
50%
tied
Champion
50%
50%
tied

Why average? Combined forecasts beat their components on average — a replicated finding across forecasting (IPCC averages climate models, BoE averages inflation models, ensembles routinely win Kaggle). The catch with one tournament: weights move modestly and the ensemble itself is now a third falsifiable claim.

The informed model — Elo ratings vs the market an independent second opinion

The informed contender: an independent forecast (not derived from the market). The engine simulates the verified bracket on real per-team World Football Elo ratings (what is Elo?, as of 2026-06-01). The group stage trusts Elo as-is; the knockout gets a disclosed ×0.6 shrink (single-elimination is coin-flippier than match-Elo). The three 2026 co-hosts get a +60 Elo home bump, and every run draws each team's rating from a ±70 Elo uncertainty band — so the favorite sits at a plausible ~16% and we don't print false-precision 0%/100% for minnows and giants. All disclosed priors, not tuned to the market. It can genuinely disagree with the crowd — but a public-ratings model is usually less sharp than a liquid market, so a big gap is more likely the model being cruder than the market being wrong. Its two books (Buy & Hold / Active) sit alongside the zero-knowledge ones below; the scorecard adjudicates.

MODEL LOWER
🇵🇹 Portugal
Win the Cup
market 10.5%model 4.5%
-6.1% gap
MODEL LOWER
🇫🇷 France
Win the Cup
market 15.6%model 9.9%
-5.7% gap
MODEL LOWER
🏴󠁧󠁢󠁥󠁮󠁧󠁿 England
Win the Cup
market 10.3%model 6.1%
-4.1% gap
The underlying data

The evidence — market vs model, one view at a time

👆 tap a tab to switch view
Market = live Polymarket ↗ prices, de-vigged so each round sums to its slots (32 advance · 8 QF · 4 SF · 2 final · 1 win). Model = pick one below. Edge = model − market.
Model:👆 tap to switch model
TeamPolymarket — de-vigged to round slots (what the WORLD prices)Win the CupEdge
model−world
AdvanceReach QFReach SFReach FinalWin Cupmarket
world
model
zero-knowledge
🇪🇸 Spain #2★199.3%53.7%38.2%22.6%16.5%16.5%18.7%+2.2%
🇫🇷 France #3★297.3%52.3%35.7%21.4%15.6%15.6%17.6%+2.0%
🇵🇹 Portugal #697.1%45.7%29.4%16.3%10.5%10.5%11.2%+0.6%
🏴󠁧󠁢󠁥󠁮󠁧󠁿 England #4★196.4%49.2%27.3%17.9%10.3%10.3%10.8%+0.6%
🇦🇷 Argentina #1★397.2%46.6%26.5%15.5%8.6%8.6%8.9%+0.3%
🇧🇷 Brazil #5★597.5%45.2%27.3%15.5%8.4%8.4%8.6%+0.2%
🇩🇪 Germany #9★497.3%35.5%19.3%9.9%5.0%5.0%4.7%-0.3%
🇳🇱 Netherlands #793.1%33.3%19.3%8.3%4.2%4.2%3.9%-0.3%
🇳🇴 Norway #2486.6%27.1%14.7%4.4%2.3%2.3%1.9%-0.4%
🇧🇪 Belgium #896.1%33.7%12.2%6.0%2.0%2.0%1.6%-0.3%
🇨🇴 Colombia #1290.6%25.3%13.0%4.6%1.7%1.7%1.4%-0.3%
🇯🇵 Japan #1880.5%22.2%9.2%3.1%1.7%1.7%1.4%-0.3%
🇲🇦 Morocco #1187.6%22.2%9.7%4.2%1.5%1.5%1.2%-0.3%
🇨🇭 Switzerland #1695.1%21.3%8.8%3.6%1.2%1.2%0.9%-0.3%
🇲🇽 Mexico #1599.5%27.5%13.9%6.1%1.2%1.2%0.9%-0.3%
🇺🇸 USA #1484.5%21.3%8.8%2.3%1.1%1.1%0.8%-0.3%
🇹🇷 Türkiye #3680.5%22.2%9.2%3.0%1.0%1.0%0.8%-0.3%
🇺🇾 Uruguay #13★289.6%20.8%6.7%3.4%0.9%0.9%0.7%-0.2%
🇪🇨 Ecuador #2288.6%17.3%7.5%2.3%0.8%0.8%0.6%-0.2%
🇭🇷 Croatia #1083.5%17.7%7.9%2.9%0.8%0.8%0.6%-0.2%
🇸🇳 Senegal #1771.4%14.2%5.3%1.9%0.6%0.6%0.4%-0.2%
🇦🇹 Austria #2381.0%12.4%4.1%1.7%0.4%0.4%0.3%-0.1%
🇨🇦 Canada #2886.6%11.5%2.6%0.9%0.3%0.3%0.2%-0.1%
🇨🇮 Ivory Coast #2978.0%12.0%4.3%1.3%0.3%0.3%0.2%-0.1%
🇸🇪 Sweden #2561.2%5.8%3.8%1.4%0.3%0.3%0.2%-0.1%
🇰🇷 South Korea #2076.4%11.1%3.1%1.3%0.2%0.2%0.1%-0.1%
🏴󠁧󠁢󠁳󠁣󠁴󠁿 Scotland #3470.4%8.0%2.8%1.1%0.2%0.2%0.1%-0.1%
🇪🇬 Egypt #2672.4%8.4%1.8%0.8%0.2%0.2%0.1%-0.1%
🇮🇷 Iran #1962.2%4.8%1.7%0.7%0.1%0.1%0.1%-0.1%
🇬🇭 Ghana #4250.6%5.5%2.0%1.1%0.1%0.1%0.1%-0.1%
🇩🇿 Algeria #3166.3%8.0%1.8%1.2%0.1%0.1%0.1%-0.1%
🇧🇦 Bosnia-Herzegovina #45····0.1%0.1%0.1%-0.1%
🇵🇾 Paraguay #3564.3%8.9%3.4%1.2%0.1%0.1%0.1%-0.1%
🇨🇩 DR Congo #3844.0%5.4%0.8%0.9%0.1%0.1%0.1%-0.1%
🇨🇿 Czechia #3369.3%8.0%2.9%1.3%0.1%0.1%0.1%-0.1%
🇦🇺 Australia #2142.5%4.0%2.8%0.9%0.1%0.1%0.1%-0.1%
🇳🇿 New Zealand #4832.4%2.7%1.0%1.3%<0.1%0.0%0.0%-0.0%
🇭🇹 Haiti #4713.7%1.4%0.2%0.3%<0.1%0.0%0.0%-0.0%
🇯🇴 Jordan #4120.2%1.8%1.1%0.4%<0.1%0.0%0.0%-0.0%
🇨🇼 Curaçao #467.6%1.4%1.1%0.3%<0.1%0.0%0.0%-0.0%
🇹🇳 Tunisia #3234.9%1.9%1.2%0.8%<0.1%0.0%0.0%-0.0%
🇺🇿 Uzbekistan #3931.9%2.1%0.6%1.1%<0.1%0.0%0.0%-0.0%
🇵🇦 Panama #2734.4%2.0%0.8%0.3%<0.1%0.0%0.0%-0.0%
🇮🇶 Iraq #4013.7%3.2%1.0%0.5%<0.1%0.0%0.0%-0.0%
🇿🇦 South Africa #4320.8%2.3%1.1%0.9%<0.1%0.0%0.0%-0.0%
🇨🇻 Cape Verde #4430.9%2.7%0.3%0.3%<0.1%0.0%0.0%-0.0%
🇶🇦 Qatar #3020.8%2.7%0.3%0.9%<0.1%0.0%0.0%-0.0%
🇸🇦 Saudi Arabia #3738.0%2.1%1.4%1.1%<0.1%0.0%0.0%-0.0%
↔ swipe the table sideways · the QF/SF/final columns are hidden on small screens (tap a team for its full route)
Secondary · what-if

If you'd traded it — a paper book to keep score, not advice

How the paper books work

A secondary, "what-if" view — purely to put a number on the disagreements above. No real money: a $1,000 paper book, conviction-weighted and dollar-neutral. Edges are shown net of a ~1c half-spread — a gap that doesn't clear the cost to trade it is sized to zero (that half-spread is the cost buffer). Buy & Hold enters once at day 0 and holds to resolution; Active Trading is re-evaluated daily and rebalances whenever a market settles or a fresh edge clears that buffer — any day, not only matchdays (a riskless inconsistency is the clearest example). Methodology →

paper bankroll $1,000 👇 each book below shows its own capital at risk & max ↑/↓
realized $+0.00unrealized $+11.26total PnL $+11.26capital at risk $1,000book max ↑ +$249book max ↓ $-1,000

Buy & Hold — entered once at day 0, held to each market's resolution, marked to today's prices. Click a row for the why; click a header to sort.

SideTeamRoundResolvesEntryNowPnLMax ↓Max ↑
LONG🇦🇷 ArgentinaAdvanceJun 2795.5%96.0%$+0.30$-51.67+$2.43
The market prices 🇦🇷 Argentina to Advance at 95.5%. Our zero-knowledge model (favorite–longshot shape correction on the de-vigged ladder) puts it at 99.9% — a +3.9% gap. Because the market systematically underprices favorites — we buy the gap. No football knowledge is used: the edge is the price structure.
stake (most you can lose) $51.67  ·  max upside if it resolves your way +$2.43  ·  breakeven at a true probability of 95.5%  ·  view market on Polymarket ↗ (may be restricted in your region)
LONG🇨🇭 SwitzerlandAdvanceJun 2793.6%93.9%$+0.15$-46.80+$3.20
The market prices 🇨🇭 Switzerland to Advance at 93.6%. Our zero-knowledge model (favorite–longshot shape correction on the de-vigged ladder) puts it at 98.5% — a +4.6% gap. Because the market systematically underprices favorites — we buy the gap. No football knowledge is used: the edge is the price structure.
stake (most you can lose) $46.80  ·  max upside if it resolves your way +$3.20  ·  breakeven at a true probability of 93.6%  ·  view market on Polymarket ↗ (may be restricted in your region)
LONG🇧🇪 BelgiumAdvanceJun 2795.9%94.9%$-0.47$-45.55+$1.95
The market prices 🇧🇪 Belgium to Advance at 95.9%. Our zero-knowledge model (favorite–longshot shape correction on the de-vigged ladder) puts it at 99.7% — a +4.8% gap. Because the market systematically underprices favorites — we buy the gap. No football knowledge is used: the edge is the price structure.
stake (most you can lose) $45.55  ·  max upside if it resolves your way +$1.95  ·  breakeven at a true probability of 95.9%  ·  view market on Polymarket ↗ (may be restricted in your region)
LONG🇵🇹 PortugalAdvanceJun 2796.2%95.9%$-0.11$-41.73+$1.67
The market prices 🇵🇹 Portugal to Advance at 96.2%. Our zero-knowledge model (favorite–longshot shape correction on the de-vigged ladder) puts it at 99.9% — a +4.0% gap. Because the market systematically underprices favorites — we buy the gap. No football knowledge is used: the edge is the price structure.
stake (most you can lose) $41.73  ·  max upside if it resolves your way +$1.67  ·  breakeven at a true probability of 96.2%  ·  view market on Polymarket ↗ (may be restricted in your region)
SHORT🇵🇦 PanamaAdvanceJun 2728.5%34.0%$-3.16$-41.11+$16.39
The market prices 🇵🇦 Panama to Advance at 28.5%. Our zero-knowledge model (favorite–longshot shape correction on the de-vigged ladder) puts it at 30.6% — a -3.4% gap. Because the market systematically overprices longshots — we fade the gap. No football knowledge is used: the edge is the price structure.
stake (most you can lose) $41.11  ·  max upside if it resolves your way +$16.39  ·  breakeven at a true probability of 28.5%  ·  view market on Polymarket ↗ (may be restricted in your region)
SHORT🇺🇿 UzbekistanAdvanceJun 2731.5%31.5%$-0.00$-40.48+$18.62
The market prices 🇺🇿 Uzbekistan to Advance at 31.5%. Our zero-knowledge model (favorite–longshot shape correction on the de-vigged ladder) puts it at 28.1% — a -3.5% gap. Because the market systematically overprices longshots — we fade the gap. No football knowledge is used: the edge is the price structure.
stake (most you can lose) $40.48  ·  max upside if it resolves your way +$18.62  ·  breakeven at a true probability of 31.5%  ·  view market on Polymarket ↗ (may be restricted in your region)
SHORT🇨🇻 Cape VerdeAdvanceJun 2732.0%30.5%$+0.89$-40.32+$18.98
The market prices 🇨🇻 Cape Verde to Advance at 32.0%. Our zero-knowledge model (favorite–longshot shape correction on the de-vigged ladder) puts it at 27.0% — a -3.5% gap. Because the market systematically overprices longshots — we fade the gap. No football knowledge is used: the edge is the price structure.
stake (most you can lose) $40.32  ·  max upside if it resolves your way +$18.98  ·  breakeven at a true probability of 32.0%  ·  view market on Polymarket ↗ (may be restricted in your region)
SHORT🇳🇿 New ZealandAdvanceJun 2732.5%32.0%$+0.30$-40.16+$19.34
The market prices 🇳🇿 New Zealand to Advance at 32.5%. Our zero-knowledge model (favorite–longshot shape correction on the de-vigged ladder) puts it at 28.6% — a -3.4% gap. Because the market systematically overprices longshots — we fade the gap. No football knowledge is used: the edge is the price structure.
stake (most you can lose) $40.16  ·  max upside if it resolves your way +$19.34  ·  breakeven at a true probability of 32.5%  ·  view market on Polymarket ↗ (may be restricted in your region)
SHORT🇸🇦 Saudi ArabiaAdvanceJun 2734.5%37.5%$-1.80$-39.23+$20.67
The market prices 🇸🇦 Saudi Arabia to Advance at 34.5%. Our zero-knowledge model (favorite–longshot shape correction on the de-vigged ladder) puts it at 34.3% — a -3.2% gap. Because the market systematically overprices longshots — we fade the gap. No football knowledge is used: the edge is the price structure.
stake (most you can lose) $39.23  ·  max upside if it resolves your way +$20.67  ·  breakeven at a true probability of 34.5%  ·  view market on Polymarket ↗ (may be restricted in your region)
SHORT🇶🇦 QatarAdvanceJun 2721.0%20.5%$+0.25$-39.11+$10.39
The market prices 🇶🇦 Qatar to Advance at 21.0%. Our zero-knowledge model (favorite–longshot shape correction on the de-vigged ladder) puts it at 17.1% — a -3.4% gap. Because the market systematically overprices longshots — we fade the gap. No football knowledge is used: the edge is the price structure.
stake (most you can lose) $39.11  ·  max upside if it resolves your way +$10.39  ·  breakeven at a true probability of 21.0%  ·  view market on Polymarket ↗ (may be restricted in your region)
SHORT🇯🇴 JordanAdvanceJun 2719.0%20.0%$-0.46$-37.50+$8.80
The market prices 🇯🇴 Jordan to Advance at 19.0%. Our zero-knowledge model (favorite–longshot shape correction on the de-vigged ladder) puts it at 16.6% — a -3.4% gap. Because the market systematically overprices longshots — we fade the gap. No football knowledge is used: the edge is the price structure.
stake (most you can lose) $37.50  ·  max upside if it resolves your way +$8.80  ·  breakeven at a true probability of 19.0%  ·  view market on Polymarket ↗ (may be restricted in your region)
LONG🇲🇽 MexicoAdvanceJun 2790.5%98.3%$+3.23$-37.47+$3.93
The market prices 🇲🇽 Mexico to Advance at 90.5%. Our zero-knowledge model (favorite–longshot shape correction on the de-vigged ladder) puts it at 99.9% — a +1.6% gap. Because the market systematically underprices favorites — we buy the gap. No football knowledge is used: the edge is the price structure.
stake (most you can lose) $37.47  ·  max upside if it resolves your way +$3.93  ·  breakeven at a true probability of 90.5%  ·  view market on Polymarket ↗ (may be restricted in your region)
LONG🇩🇪 GermanyAdvanceJun 2796.5%96.2%$-0.12$-37.23+$1.37
The market prices 🇩🇪 Germany to Advance at 96.5%. Our zero-knowledge model (favorite–longshot shape correction on the de-vigged ladder) puts it at 99.9% — a +3.8% gap. Because the market systematically underprices favorites — we buy the gap. No football knowledge is used: the edge is the price structure.
stake (most you can lose) $37.23  ·  max upside if it resolves your way +$1.37  ·  breakeven at a true probability of 96.5%  ·  view market on Polymarket ↗ (may be restricted in your region)
SHORT🇿🇦 South AfricaAdvanceJun 2738.0%20.5%$+10.46$-37.08+$22.72
The market prices 🇿🇦 South Africa to Advance at 38.0%. Our zero-knowledge model (favorite–longshot shape correction on the de-vigged ladder) puts it at 17.1% — a -3.4% gap. Because the market systematically overprices longshots — we fade the gap. No football knowledge is used: the edge is the price structure.
stake (most you can lose) $37.08  ·  max upside if it resolves your way +$22.72  ·  breakeven at a true probability of 38.0%  ·  view market on Polymarket ↗ (may be restricted in your region)
SHORT🇹🇳 TunisiaAdvanceJun 2738.5%34.5%$+2.39$-36.72+$22.98
The market prices 🇹🇳 Tunisia to Advance at 38.5%. Our zero-knowledge model (favorite–longshot shape correction on the de-vigged ladder) puts it at 31.1% — a -3.4% gap. Because the market systematically overprices longshots — we fade the gap. No football knowledge is used: the edge is the price structure.
stake (most you can lose) $36.72  ·  max upside if it resolves your way +$22.98  ·  breakeven at a true probability of 38.5%  ·  view market on Polymarket ↗ (may be restricted in your region)
LONG🇧🇷 BrazilAdvanceJun 2796.5%96.4%$-0.06$-36.48+$1.32
The market prices 🇧🇷 Brazil to Advance at 96.5%. Our zero-knowledge model (favorite–longshot shape correction on the de-vigged ladder) puts it at 99.9% — a +3.5% gap. Because the market systematically underprices favorites — we buy the gap. No football knowledge is used: the edge is the price structure.
stake (most you can lose) $36.48  ·  max upside if it resolves your way +$1.32  ·  breakeven at a true probability of 96.5%  ·  view market on Polymarket ↗ (may be restricted in your region)
LONG🇳🇱 NetherlandsAdvanceJun 2790.0%92.0%$+0.80$-36.00+$4.00
The market prices 🇳🇱 Netherlands to Advance at 90.0%. Our zero-knowledge model (favorite–longshot shape correction on the de-vigged ladder) puts it at 96.2% — a +4.2% gap. Because the market systematically underprices favorites — we buy the gap. No football knowledge is used: the edge is the price structure.
stake (most you can lose) $36.00  ·  max upside if it resolves your way +$4.00  ·  breakeven at a true probability of 90.0%  ·  view market on Polymarket ↗ (may be restricted in your region)
LONG🇪🇨 EcuadorAdvanceJun 2789.0%87.5%$-0.56$-33.11+$4.09
The market prices 🇪🇨 Ecuador to Advance at 89.0%. Our zero-knowledge model (favorite–longshot shape correction on the de-vigged ladder) puts it at 90.8% — a +3.3% gap. Because the market systematically underprices favorites — we buy the gap. No football knowledge is used: the edge is the price structure.
stake (most you can lose) $33.11  ·  max upside if it resolves your way +$4.09  ·  breakeven at a true probability of 89.0%  ·  view market on Polymarket ↗ (may be restricted in your region)
SHORT🇨🇩 DR CongoAdvanceJun 2743.0%43.5%$-0.28$-32.49+$24.51
The market prices 🇨🇩 DR Congo to Advance at 43.0%. Our zero-knowledge model (favorite–longshot shape correction on the de-vigged ladder) puts it at 40.7% — a -2.8% gap. Because the market systematically overprices longshots — we fade the gap. No football knowledge is used: the edge is the price structure.
stake (most you can lose) $32.49  ·  max upside if it resolves your way +$24.51  ·  breakeven at a true probability of 43.0%  ·  view market on Polymarket ↗ (may be restricted in your region)
SHORT🇮🇶 IraqAdvanceJun 2714.5%13.5%$+0.38$-32.32+$5.48
The market prices 🇮🇶 Iraq to Advance at 14.5%. Our zero-knowledge model (favorite–longshot shape correction on the de-vigged ladder) puts it at 10.6% — a -2.9% gap. Because the market systematically overprices longshots — we fade the gap. No football knowledge is used: the edge is the price structure.
stake (most you can lose) $32.32  ·  max upside if it resolves your way +$5.48  ·  breakeven at a true probability of 14.5%  ·  view market on Polymarket ↗ (may be restricted in your region)
LONG🇫🇷 FranceAdvanceJun 2796.8%96.2%$-0.21$-31.85+$1.05
The market prices 🇫🇷 France to Advance at 96.8%. Our zero-knowledge model (favorite–longshot shape correction on the de-vigged ladder) puts it at 99.9% — a +3.8% gap. Because the market systematically underprices favorites — we buy the gap. No football knowledge is used: the edge is the price structure.
stake (most you can lose) $31.85  ·  max upside if it resolves your way +$1.05  ·  breakeven at a true probability of 96.8%  ·  view market on Polymarket ↗ (may be restricted in your region)
SHORT🇭🇹 HaitiAdvanceJun 2713.0%13.5%$-0.17$-30.02+$4.49
The market prices 🇭🇹 Haiti to Advance at 13.0%. Our zero-knowledge model (favorite–longshot shape correction on the de-vigged ladder) puts it at 10.6% — a -2.9% gap. Because the market systematically overprices longshots — we fade the gap. No football knowledge is used: the edge is the price structure.
stake (most you can lose) $30.02  ·  max upside if it resolves your way +$4.49  ·  breakeven at a true probability of 13.0%  ·  view market on Polymarket ↗ (may be restricted in your region)
LONG🇨🇴 ColombiaAdvanceJun 2787.5%89.5%$+0.66$-28.70+$4.10
The market prices 🇨🇴 Colombia to Advance at 87.5%. Our zero-knowledge model (favorite–longshot shape correction on the de-vigged ladder) puts it at 93.2% — a +3.7% gap. Because the market systematically underprices favorites — we buy the gap. No football knowledge is used: the edge is the price structure.
stake (most you can lose) $28.70  ·  max upside if it resolves your way +$4.10  ·  breakeven at a true probability of 87.5%  ·  view market on Polymarket ↗ (may be restricted in your region)
SHORT🇨🇭 SwitzerlandReach QFJul 722.0%24.0%$-0.68$-26.68+$7.52
The market prices 🇨🇭 Switzerland to Reach QF at 22.0%. Our zero-knowledge model (favorite–longshot shape correction on the de-vigged ladder) puts it at 20.6% — a -3.4% gap. Because the market systematically overprices longshots — we fade the gap. No football knowledge is used: the edge is the price structure.
stake (most you can lose) $26.68  ·  max upside if it resolves your way +$7.52  ·  breakeven at a true probability of 22.0%  ·  view market on Polymarket ↗ (may be restricted in your region)
SHORT🇹🇷 TürkiyeReach QFJul 721.5%25.0%$-1.19$-26.69+$7.31
The market prices 🇹🇷 Türkiye to Reach QF at 21.5%. Our zero-knowledge model (favorite–longshot shape correction on the de-vigged ladder) puts it at 21.6% — a -3.4% gap. Because the market systematically overprices longshots — we fade the gap. No football knowledge is used: the edge is the price structure.
stake (most you can lose) $26.69  ·  max upside if it resolves your way +$7.31  ·  breakeven at a true probability of 21.5%  ·  view market on Polymarket ↗ (may be restricted in your region)
LONG🇺🇾 UruguayAdvanceJun 2786.5%88.5%$+0.60$-25.95+$4.05
The market prices 🇺🇾 Uruguay to Advance at 86.5%. Our zero-knowledge model (favorite–longshot shape correction on the de-vigged ladder) puts it at 92.0% — a +3.5% gap. Because the market systematically underprices favorites — we buy the gap. No football knowledge is used: the edge is the price structure.
stake (most you can lose) $25.95  ·  max upside if it resolves your way +$4.05  ·  breakeven at a true probability of 86.5%  ·  view market on Polymarket ↗ (may be restricted in your region)
LONG🇲🇦 MoroccoAdvanceJun 2786.0%86.5%$+0.14$-24.42+$3.98
The market prices 🇲🇦 Morocco to Advance at 86.0%. Our zero-knowledge model (favorite–longshot shape correction on the de-vigged ladder) puts it at 89.6% — a +3.1% gap. Because the market systematically underprices favorites — we buy the gap. No football knowledge is used: the edge is the price structure.
stake (most you can lose) $24.42  ·  max upside if it resolves your way +$3.98  ·  breakeven at a true probability of 86.0%  ·  view market on Polymarket ↗ (may be restricted in your region)
LONG🇳🇴 NorwayAdvanceJun 2785.5%85.5%$+0.00$-23.09+$3.92
The market prices 🇳🇴 Norway to Advance at 85.5%. Our zero-knowledge model (favorite–longshot shape correction on the de-vigged ladder) puts it at 88.4% — a +2.9% gap. Because the market systematically underprices favorites — we buy the gap. No football knowledge is used: the edge is the price structure.
stake (most you can lose) $23.09  ·  max upside if it resolves your way +$3.92  ·  breakeven at a true probability of 85.5%  ·  view market on Polymarket ↗ (may be restricted in your region)
realized $+0.00 · unrealized $+11.26 · 28/28 open$+11.26↤ click a row
About. A solo research & education project by Marcus Liew ↗ — an open test of whether transparent models can beat a liquid market, with an honest public scorecard. Built in the open, scored against the crowd; no positions, no capital, no advice. Common questions (FAQ) →  ·  Source & method on GitHub ↗
⚠️ A research & education experiment — NOT gambling, and NOT an encouragement to gamble. We hold no positions and invest no real capital; every figure here is a paper simulation of market structure, not investment returns or a tipping service. This is not financial advice and not a solicitation to trade or bet. Prediction-market platforms such as Polymarket are restricted or banned in several jurisdictions (e.g. Singapore) — know and follow your local laws; the Polymarket links are reference-only. Book max ↑/↓ is a loose theoretical envelope (outcomes are correlated). Whether the model actually beats the market is settled by the live scorecard as matches resolve.
FIFA ranking & titles are reference data (rank Jul 2025 (approx.)) — source: FIFA / inside.fifa.com ↗. Flag images via flagcdn.com. We are not affiliated with FIFA or Polymarket.