The Model's Surprising Lean
Italy and Belgium open League A with a blank table: both have played zero games, zero points, and sit second and third respectively. The headline tension is that Nutmegly's AI read still leans home. It gives Italy a 43.6% win probability, a 27.5% draw, and Belgium 28.9%. That is a lean, not a majority, and it arrives against a stored form picture that looks more favorable to Belgium. The match matters because it is the opening round, there is no head-to-head sample to check the read, and confirmed lineups are not yet in the data. The model's estimate is a forecast, not an observed result.
What the Stored Results Actually Say
Italy's stored sample is three results, newest first: a 1-1 draw away to Bosnia & Herzegovina on 31 March 2026, a 1-4 home loss to Norway on 16 November 2025, and a 0-3 away loss to Norway on 6 June 2025. That works out to 0.33 points per game across the stored sample. Belgium's stored sample is eight results and 1.88 points per game. The five most recent listed are a 1-2 away loss to Spain on 10 July 2026, a 4-1 away win over the USA on 7 July, a 3-2 home win over Senegal on 1 July, a 5-1 away win over New Zealand on 27 June, and a 0-0 home draw with Iran on 21 June. Those dates span seasons, so this is not current-season form. The samples are also unequal: three for Italy, eight for Belgium. The gap since the last stored match is 178 days for Italy and 76 for Belgium, but the data warns that gaps longer than 21 days should not be treated as a rest advantage, and missing records do not prove a team has not played.
The Case for the Home Lean
The model's home lean is not built on Italy's recent points efficiency, which is the weakest supplied signal. The home/away matchup is level: Italy's home average is 1.00 goal, and Belgium's away average is 1.00 goal against. A same-competition signal puts both at 1.00 points per game, though the quality counts list no competition-specific matches for either side, so that is not a separator. The table gives Italy rank 2 and Belgium rank 3 with zero points each. The clearest home-direction signal is the result distribution itself: 43.6% home, 27.5% draw, 28.9% away. The model also puts over 2.5 goals at 47.81%, meaning it slightly favors a lower-scoring game than a high-scoring one. In a match with no head-to-head sample, home status and the distribution are doing much of the work.
The Counterevidence
Belgium's stored form is the strongest argument against the home lean. Its 1.88 points per game over eight stored results is far above Italy's 0.33 over three. The listed Belgian wins include 4-1, 3-2 and 5-1 scorelines, and the one listed loss is to Spain. Italy's listed run includes two losses to Norway, one of them 1-4 at home. But the model's own counterpoints are important: the recent sample for Italy is under five matches, historical xG and technical-stat samples are thin, player-rating samples are thin, and there is no head-to-head data. The quality counts show one advanced-stat sample for Italy and seven for Belgium, and two player-rating matches for Italy against eight for Belgium. That asymmetry cuts both ways: Belgium has more evidence, but Italy has less evidence to explain the lean. The data also carries no confirmed lineups and no injury reports; that is an absence of information, not proof of a fully available squad.
What to Watch
The watchpoint is whether Belgium's stored scoring output translates into chances against Italy's home defense, and whether Italy can exceed the 1.00 home goal its stored sample shows. If Belgium's recent attacking returns carry over, the 28.9% away-win probability may look low; if the game stays tight, the 27.5% draw and the 47.81% over-2.5 probability point to a narrow, lower-scoring contest. The most useful new information would be confirmed lineups, which are not in the data. Until then, the honest read is a home lean with a thin evidence base. The question to watch: can Italy's home sample produce more than one goal, or does Belgium's stored form turn the opener into the away performance the model is not favoring?