A Blank Table, A Heavy Lean
The UEFA Nations League returns with a League B opener that pits FYR Macedonia against Switzerland, and the most interesting tension is not the table. Both teams have played zero games and hold zero points, so the current standings are a blank slate. The supplied standings place Switzerland above FYR Macedonia, but that is about all the table can tell us before kickoff on 26 September. What matters is the gap between the model's confidence and the evidence underneath it.
Nutmegly's AI read leans away, and it leans hard: Switzerland are given a 66.2% win probability, with FYR Macedonia at 12.9% and a draw at 20.9%. The same model puts over 2.5 goals at 52.3%, only slightly above a coin flip. That is a clear headline, but the data behind it is not a complete season record. The stored samples span seasons, and the model itself flags thin recent samples, sparse historical xG and player-rating data, and no direct head-to-head matches. This is a preview built on a partial picture.
What the Stored Results Actually Say
FYR Macedonia's stored recent sample contains four matches: a 0-4 away loss to Türkiye on 1 June 2026, a 0-0 away draw with Bosnia & Herzegovina on 29 May 2026, a 0-0 away draw with Belgium on 10 October 2025, and a 1-1 home draw with Belgium on 6 June 2025. That is three draws and one defeat, 0.75 points per game, one goal scored and five conceded across the sample. The dates matter: this is not current-season form, and the sample is not five matches. It is a small, mixed record that includes two goalless draws and one heavy defeat.
Switzerland's stored recent sample is stronger. The five most recent results shown are a 1-3 away loss to Argentina on 12 July 2026, a 0-0 home draw with Colombia on 7 July 2026, a 2-0 home win over Algeria on 3 July 2026, a 2-1 home win over Canada on 24 June 2026, and a 4-1 home win over Bosnia & Herzegovina on 18 June 2026. That is three wins, one draw and one defeat from the displayed results, and the model's stored sample counts eight matches at 2.00 points per game. The wins came at home; the one away result in the displayed list was the Argentina defeat. That caveat matters for Switzerland's away trip.
The home-away split offers the clearest counterweight. FYR Macedonia's home average in the stored sample is 1.00 goal per game, while Switzerland's away average goals against is 0.60. Those are small samples, and one is just a single home match, but they point in a different direction from the overall lean. If Macedonia can keep the game tight at home, the model's away probability may look less overwhelming than the headline number suggests.
The AI Read and Its Counterweights
The system's prediction is plain: Nutmegly's AI read favors Switzerland, with a 66.2% win probability, a 20.9% draw probability and a 12.9% FYR Macedonia win probability. The over-2.5-goals estimate is 52.3%. Those are model estimates, not observed results or guarantees. A leading outcome is not a certainty, and a 66.2% favorite still loses roughly one time in three by the model's own numbers.
The case for the away lean is built on points efficiency and recent results. Switzerland's stored sample shows 2.00 points per game, compared with FYR Macedonia's 0.75. Switzerland have three wins in the five displayed results, including a 4-1 and a 2-0 at home. FYR Macedonia have not won in their four stored results and have scored just one goal in that span. On the surface, that is a mismatch in form and output.
The counterevidence is substantial. In the UEFA Nations League sample, both teams average 1.00 point per game, so the competition-specific record does not separate them. There is no head-to-head sample at all, which removes one of the usual checks on a lopsided prediction. The model notes that recent samples are under five matches for FYR Macedonia, historical xG and technical-stat samples are thin, and player-rating samples are sparse. The evidence score is 0.503. That does not make the away lean wrong, but it does mean the confidence should be read as a model output under limited data, not as a settled verdict.
What Would Change the Picture
The match could unfold differently if FYR Macedonia's home defensive sample holds. Their stored home result is a 1-1 draw with Belgium, and their overall sample includes two 0-0 draws. If that pattern repeats, the game could stay low-scoring, which would put pressure on the 52.3% over-2.5 estimate and give the draw a better chance than 20.9%. If Switzerland instead carry their home scoring form into an away setting, the 66.2% away win probability would look more justified. The data does not tell us which version will appear.
Lineup and injury information is limited. The data contains no confirmed lineups and no reported absences from the injury interface. That does not mean either team is at full strength, and it does not prove that no official announcement has been made. It simply means we cannot assess availability from this payload. Any late change to the starting eleven would sit outside the model's current inputs, and the model already notes that lineups are not fully confirmed.
The rest-window signal is also neutral. The last recorded matches are 117 days before the as-of date for FYR Macedonia and 76 days for Switzerland. The data explicitly says gaps longer than 21 days are not treated as a rest advantage, and a missing record does not prove a team has not played. So neither side should be described as fresher or rustier on this evidence.
The Watchpoint
Watch the first goal, and watch whether FYR Macedonia can keep the match level into the second half. If Switzerland score first, their 2.00 points-per-game stored sample and their recent home wins suggest the away lean would have more support. If FYR Macedonia score first, their stored sample offers little evidence of protecting a lead: three draws and one defeat, with only one goal scored across four matches. The most revealing scenario is a goalless hour. That would test the model's away lean against Macedonia's two 0-0 draws and Switzerland's 0-0 with Colombia, and it would make the 20.9% draw probability feel far more live than the headline 66.2% away number.