The table points one way, the away split another
Armenia arrive at Skonto third in the table with five points from five matches. Latvia are fourth with one point from five. Armenia also won the only head-to-head stored in the sample, 2-0 on 25 September. On that evidence, the obvious pre-match read is an Armenia win. Nutmegly's AI read instead makes the draw its leading outcome at 45.66%, with Armenia at 34.8% and Latvia at 19.54%. The draw is not a majority, but it is the model's lean, and the reason is hiding in the home and away splits.
Latvia's home record and Armenia's away record are almost mirror images. Both have played two matches in those splits, and both have one draw, one defeat and one point. The difference is goals: Latvia have scored once at home and conceded twice; Armenia have scored none away and conceded twice. That is the draw case in miniature. Armenia's overall table position is better, but their away matches have produced no goals for them, and Latvia's home matches have produced only one.
This is not a case of Latvia being secretly good. It is a case of the match profile being narrow. Latvia have one point and have not won; Armenia have five points and have won once. But the specific venue split strips away some of Armenia's advantage. At Skonto, the question is whether Armenia can do something they have not done in their two away matches in the table: score.
Recent form: Latvia lose, Armenia draw
Latvia's stored form is bleak. In their last five listed matches they have one draw and four defeats: 1-2 away to Montenegro on 12 November, 1-2 away to Cyprus on 5 October, 1-2 at home to Montenegro on 2 October, 0-0 at home to Cyprus on 28 September, and 0-2 away to Armenia on 25 September. Each of the four defeats involved conceding two goals. Latvia scored three and conceded eight across those five, which matches their current table totals of three for and eight against.
Armenia's last five are more mixed but not convincing: 1-1 at home to Cyprus on 12 November, 0-0 away to Montenegro on 5 October, 0-2 away to Cyprus on 2 October, 2-3 at home to Montenegro on 28 September, and 2-0 at home to Latvia on 25 September. That is one win, two draws and two losses. In the stored sample, Armenia have not won since beating Latvia. Their two most recent matches were draws, and their two away matches in that run produced no goals for them.
The wider stored sample is also part of the model's evidence. Latvia's points-per-match average is 0.17; Armenia's is 0.71. Those are sample averages, not current-season totals, and they span more matches than the five listed above. They explain why the model still gives Armenia a 34.8% win chance despite the away blank. The draw lean is not a claim that Latvia have been the better team.
The draw case, and the strongest countercase
The draw case rests on three connected facts. Armenia have zero goals in two away matches. Latvia have one goal in two home matches. And Nutmegly's model puts over 2.5 goals at only 8.36%, which points to a tight, low-scoring game rather than an open one. Those facts do not guarantee a draw, but they explain why the model's leading outcome is 45.66% draw rather than an Armenia win.
The strongest counterevidence is the head-to-head and the table gap. Armenia beat Latvia 2-0 in September, and they have four more points. Their overall goal difference is minus one; Latvia's is minus five. Armenia also have a 34.8% win probability, which is not a trivial number. If the September result is more predictive than the away split, the draw lean fails. The fair response is that the September match was at Armenia's home, while this one is at Skonto. That does not erase the result, but it limits how directly it applies.
There are also data limits worth stating once. No confirmed lineups or injury reports were available, so we cannot assess who is available or how either side will line up; that absence is not evidence that either squad is fully healthy or settled. There was no odds market to calibrate the probabilities, and the historical xG and technical sample is thin. The model's read is therefore an estimate with real uncertainty, not a guarantee, and a single early goal could change the match state that the low-scoring projection assumes.
The watchpoint: Armenia's away blank
The match turns on a simple conditional: can Armenia score away from home? They have not done it in two away matches in the table. If they score at Skonto, the cleanest piece of evidence behind the draw lean weakens immediately, and the model's 34.8% Armenia win becomes the more relevant branch. If they do not, the game fits the low-scoring shape the model expects, and Latvia's one home goal becomes the other side of the same question.
That is the question to watch, not another probability recap. Armenia's away blank is the storyline because it cuts against the table. Latvia's winless run is the counterweight because it explains why the model will not make them favourites. The draw is the model's leading outcome, but the match will tell us quickly whether the away split holds. The first Armenia goal, or another long blank, is the clearest evidence either way.