We can't find the internet
Attempting to reconnect
Something went wrong!
Attempting to reconnect
Our Scorito pool
This is how we filled in the World Cup form — and how you can see what to look for yourself. No gut feeling, but probability theory.
What do you look at?
Strength: Elo ratings
Every nation has an Elo rating (eloratings.net) summarising all internationals ever played. Spain stood at 2157 on 11 June, Curaçao at 1434. The gap determines the win probability. Hosts (USA, Mexico, Canada) get +100 home advantage.
Score: Dixon-Coles
Elo says who wins, not how. The Dixon-Coles model translates the win probability into expected goals per team (λ) and gives every scoreline (0–0, 2–1, …) a probability. Crucial for Scorito: we pick the most likely score within the best result column — never a score that contradicts your result pick.
Value: expected points (EV)
Per match: EV = 30 × P(result) + 15 × P(exact score). That's why sometimes a 'boring' 1–0 instead of a gut-feel 3–1: the chance of an exact hit counts. The group standing (25 pts/position) follows automatically from the predicted scores. Champion (250 pts) comes from 25,000 Monte Carlo simulations of the whole tournament.
36.9% kampioenskans (Monte Carlo 11 juni) — verreweg de hoogste.
36.9%All 72 entries
Per match: result, score and expected Scorito points (EV). Below: the group standing that follows from our scores.
Our top scorer picks
The trick is in the scoring: in the group stage a defender's goal earns 32 points, a midfielder's 16 and a forward's only 8. So you don't look for strikers — you look for scoring 'midfielders' from high-scoring nations.
Kai Havertz
Listed at Scorito as a midfielder → 16 points per goal, double a striker's. Actually plays as a forward and takes penalties. Germany scores a lot in group E (Curaçao Elo 1434).
Florian Wirtz
Same leverage: midfielder label, but in reality the creative goal-getter behind the striker. High share of Germany's goals, and Germany has a 98% chance of advancing from the group.
Jamal Musiala
The third German with the midfielder label. Three picks from one country looks risky, but here it is the strategy: every German goal from this trio counts double, and group E is the second-weakest of the tournament.
Cristiano Ronaldo
Forward (8 pts/goal), but pure volume: penalty taker, free kicks, and Portugal faces DR Congo (Elo 1652) and Uzbekistan (1714). The most reliable goal machine of the easier groups.
Lionel Messi
Argentina has the highest tournament expectation after Spain (35% final) and Messi takes the penalties and free kicks. Group J holds Algeria and Jordan (Elo 1680).
Erling Haaland
Highest shot quality in the field (FIFA shooting 92 in our database). Norway faces Iraq (Elo 1607) — exactly the kind of match where Haaland scores three, even if Norway exits afterwards.
Locked before the group stage, in line with external simulations (pouletips.nl, 10,000 runs) and our own model. Knockout stage: each round you pick 4 new top scorers — the simulator then reruns with the actual qualifiers.
Is this made with AI?
Scores & odds: statistics
All match predictions, win probabilities, EVs, group standings and the champion pick come from classic probability models — Elo ratings, the Dixon-Coles Poisson model and Monte Carlo simulation. No language model involved: every outcome is mathematically traceable.
Top scorers & strategy: AI analysis
The top scorer picks ARE AI work: Claude Fable 5 (Anthropic) combined the Scorito scoring rules (midfielder leverage), external simulations (pouletips.nl, 10,000 runs) and our own model into the final advice. That is AI reasoning, not a formula output.
The software: built with AI
The entire platform was built with Claude Fable 5 via Claude Code — the AI wrote the code, the statistical model computes the odds. The system learns during the tournament through Elo updates after every result, fully rule-based.
Honest about uncertainty
Football is rolling weighted dice: even the best entry won't hit 100%. So we show probabilities, not certainties — and on the World Cup page you can see live how right (or wrong) we are.