The table says one thing, the model says another
On Friday at Skonto, Latvia and Montenegro meet in the UEFA Nations League, League C. The current table frames the match sharply: Montenegro are first with six points from two games, five goals scored and three conceded; Latvia are fourth with one point from two, no goals scored and two conceded. That is the headline tension. One side has started with two wins, the other has started with a draw and a defeat.
Nutmegly's AI read, generated on 24 September, leans draw. It puts Latvia's win probability at 21.1%, the draw at 58.6%, and Montenegro's win at 20.3%. The same model gives over 2.5 goals a 2.31% probability. That is an estimate, not a result, and the date matters: the prediction was produced before the latest round of matches on 25 and 28 September.
Same opponents, very different returns
Latvia's newest stored result is a 0-0 home draw with Cyprus on 28 September, followed by a 0-2 away defeat to Armenia on 25 September. Their older stored sample includes a 0-5 home loss to England from October 2025. Those results span seasons, so they are a historical sample rather than a complete current-season record.
Montenegro's newest results are a 3-2 away win over Armenia on 28 September and a 2-1 home win over Cyprus on 25 September. Older entries include defeats to Croatia and the Czech Republic. Against the same two current opponents, Montenegro took six points; Latvia took one. Montenegro also scored five goals across their two current-table matches, while Latvia's current-table line shows none.
Why the draw read has a case
The draw lean is not random. Latvia's one home match in the current table was a 0-0 draw, and their home goals-for and goals-against are both zero. Montenegro's one away match was a 3-2 win, so they conceded twice while taking maximum points. The model's very low over-2.5 probability, 2.31%, suggests it expects a tight, low-scoring game rather than a repeat of Montenegro's five-goal start.
The evidence base also argues for caution. There is no head-to-head sample, no odds summary, no confirmed lineups, and the advanced-stat and player-rating samples are below the thresholds that would give those factors heavier weight. The injury feed reports no absent players, but that does not prove full availability, and lineups are not confirmed.
The counterevidence the draw must survive
The strongest supplied counterevidence is Montenegro's current-table position and their results against the same opponents. They lead the group with six points, have won both matches, and have already beaten Cyprus and Armenia. Latvia have one point, no goals, and a 0-2 defeat to Armenia. If the model's draw probability is right, it has to explain why that form gap does not translate into a clearer away lean.
The model's own counterpoints acknowledge the issue: the three-way probabilities are not widely separated, recent samples are under five matches, and the prediction was generated before the latest results. That does not invalidate the draw read, but it does mean the 58.6% draw figure should be treated as a model estimate built on a thin, partly stale evidence base.
What to watch at Skonto
The specific watchpoint is the opening goal. If Montenegro score first, their current-table record of five goals in two matches gives them a platform that Latvia have not shown they can chase, and the away-win probability may look low in hindsight. If Latvia score first, the draw lean's low-scoring assumption is immediately under pressure, especially because Montenegro's one away match produced three goals for them and two against.
If neither side scores early, the 0-0 home draw Latvia already have on file and the model's 2.31% over-2.5 probability become the most relevant reference points. The match may not follow either script, but the first goal is the cleanest way to test whether the table or the model is reading this fixture better.