Four matches today, all from Groups G and H. The numbers matter, but the stories come first.
Belgium vs Egypt: two squads running out of time
Kevin De Bruyne turns 35 this month. Romelu Lukaku is 33. Thibaut Courtois is 34. Belgium finished third in 2018, ranked number one in the world for three consecutive years, and never won a trophy. A quarter-final loss to France in 2018. A group-stage exit in 2022. The trajectory has been pointing down for four years.
On the other side of the pitch: Mohamed Salah, 34 in June. Egypt's qualifying campaign was built on his goals. This is almost certainly his final World Cup too. Two ageing squads of stars who never quite got there, drawn together in the Group G opener.
The model gives Belgium 60.4%, a draw at 24.4%, Egypt at 15.2%. Prediction markets sit at roughly 62/23/15, which is close enough to call aligned. This is not a match where the model disagrees with consensus. It is a match where the human stakes are higher than any probability can capture.
One tactical detail to watch: Egypt's set-piece expected-goals share (17.3%) is the highest of any team playing today. If they are going to hurt Belgium, dead balls are the most likely route.
Saudi Arabia vs Uruguay: the 2022 comparison everyone will make
Saudi Arabia beat Argentina 2-1 in the 2022 group stage. That result is the lens through which today's match will be viewed.
The model gives Saudi Arabia 7.3%, a draw at 23.7%, Uruguay at 69.0%. Prediction markets give Saudi Arabia roughly 12%, nearly double the model's estimate. Markets may be pricing in 2022 narrative and the energy of a large Saudi diaspora in the US. The model sees the Elo gap (324 points) and stays sceptical.
The tactical reality matters more than the memory. Argentina in 2022 played a high defensive line that Saudi Arabia exploited with precisely timed runs behind the back four. Uruguay's profile is the opposite: compact, balanced, 49.3% possession, a pressing intensity (PPDA 18.0) that controls space rather than conceding it. There is no room to run in behind.
A draw is the realistic Saudi path. At 23.7%, the model actually rates that higher than markets do (21%). And the tournament so far has been running at 37.5% draws (3 of 8 scored matches), well above the model's predicted 25%. If that pattern holds, this is a candidate.
Iran vs New Zealand: the contrarian call gets tested
Before the tournament, we published Five Contrarian Calls. Iran at 81.2% to advance was the third. The reasoning: Group G has one of the cleanest top-two splits in the 48-team field. Belgium at 95.7% to advance. Iran at 81.2%. Then daylight: Egypt at 67.1%, New Zealand at 24.9%.
Today that call faces its first real test.
The model gives Iran 62.8%, a draw at 24.7%, New Zealand at 12.6%. Markets give New Zealand 21%, nearly double the model's estimate. This is the biggest model-versus-market divergence on today's card: 8.8 percentage points on Iran, 8.4 on New Zealand.
The model's confidence comes from Elo (Iran 1760 vs New Zealand 1585, a 175-point gap). But Iran's tactical profile is the most extreme in the tournament: 33% possession, 46 long balls per 90, a pressing intensity (PPDA 29) so loose it barely qualifies as pressing. They will sit deep and counter. New Zealand's data has only partial coverage, which means the model is working with less information here than almost any other match.
If New Zealand get a result, the 81% contrarian call starts to unravel.
Spain vs Cape Verde: the margin question
Spain at 87.6% is the heaviest favourite on the June 15 card. The Elo gap of 616 points is the widest in any match today. Cape Verde, population roughly 600,000, are playing their first ever World Cup match.
The interesting wrinkle is tactical. Cape Verde's profile is classified as "high-press" (PPDA 17.2, 53.5% possession in qualifying). They will not sit in a low block. Against Spain's 68.5% possession dominance, the most dominant in the tournament, that creates two possible matches: an open, entertaining contest where Cape Verde's aggression gives Spain space to exploit, or a lopsided scoreline that arrives quickly.
Markets give Spain 91%, slightly higher than the model's 87.6%. The model hedges more toward a draw (10.5% vs the market's 7%), continuing its tournament-long pattern of assigning higher draw probabilities than consensus.
The draws thesis, updated
Through eight scored matches, the tournament draw rate is 37.5% (three draws from eight). The model's average predicted draw probability across those eight matches was roughly 25%. The draws thesis, that the model assigns higher draw probabilities than consensus and that reality may run even higher, is being validated beyond the model's own elevated estimates.
Today's four matches all carry draw probabilities between 10% and 25%. Belgium-Egypt (24.4%), Iran-New Zealand (24.7%), and Saudi Arabia-Uruguay (23.7%) are all candidates. If even one draws, the tournament draw rate stays above 30%.
All probabilities are pre-match model outputs, locked before kickoff. The model publishes probabilities, not recommendations. Methodology: /docs/methodology/. Full Terms of Use.