Ninety-eight problems: why Armenia's Nations League window went wrong
Armenia played four Nations League games between 25 September and 5 October and took four points from them. They beat Latvia, lost at home to Montenegro, lost in Cyprus and drew in Montenegro. With two rounds left they are third, and the realistic target now is second place and a play-off.
Most of the explanations since then have been about circumstances. Fourteen players unavailable. Four games in eleven days. A young squad short on experience. Opponents who were, we were told, better than us. Some of that is true. A lot of it falls apart once you look at the numbers.
We went through all four games using UEFA's tracking data, Sofascore's event data, Elo ratings, squad lists, Transfermarkt values, this season's APL numbers and the coach's press conferences from 23 September to 5 October. Nobody is questioning the players' effort here. They ran, several of them played through injuries, and Yeghishe Melikyan is right to praise them for it. This piece is about the decisions around them: how the games were approached, how Armenia attacked, who was picked, and what was said in public.
Four games, four points
Armenia beat Latvia 2–0 in Yerevan, came from behind twice against Montenegro and still lost 3–2, lost 2–0 in Nicosia and drew 0–0 in Podgorica.
| # | Team | P | W | D | L | Goals | GD | Pts | Form |
|---|---|---|---|---|---|---|---|---|---|
| 1 |  Montenegro | 4 | 3 | 1 | 0 | 7–4 | +3 | 10 | WWWD |
| 2 |  Cyprus | 4 | 2 | 1 | 1 | 5–3 | +2 | 7 | LDWW |
| 3 |  Armenia | 4 | 1 | 1 | 2 | 4–5 | -1 | 4 | WLLD |
| 4 |  Latvia | 4 | 0 | 1 | 3 | 2–6 | -4 | 1 | LDLL |
vCyprusvLatviavMontenegrovArmeniaMontenegro are out of reach. Armenia can still get to ten points, but Montenegro have the better head-to-head record (a win and a draw). Second place is still open. Cyprus come to Yerevan on 12 November three points ahead of Armenia, and whoever finishes second goes into a play-off for League B in March.
The last line on each card is Elo's pre-match win chance. Armenia were favourites in both home games and not far off even in Nicosia. The only game they were expected to lose was the last one, and that turned out to be their best performance.
Who was the favourite?
The day before Montenegro came to Yerevan, Melikyan was asked whether the game could decide the group. He said Armenia wanted to win but that it was "also important not to lose to Montenegro". Then he added:
Nair Tiknizyan, sitting next to him, was asked who the favourites were. "I think Montenegro are the favourites in our group," he said. The same weekend, Montenegro's goalkeeper Balša Popović told the Montenegrin press: "I think we have more quality and are the better team. We are going to Yerevan to win."
So one team travelled expecting to win, and the other was ready to settle for a point. Elo had it differently.
Elo rates national teams on results, adjusted for the opponent and the venue. With home advantage it gave Armenia a 53% chance of beating Montenegro in Yerevan, which is about as close to a coin toss as it gets. Montenegro had also lost both of their previous games in Yerevan, 3–2 in 2016 and 1–0 in 2022.
To be fair, Montenegro are the strongest team in the group on paper. Transfermarkt values their squad at €88.1m against Armenia's €45.3m, and three of their forwards play in Serie A: Nikola Krstović, Milutin Osmajić and Vasilije Adžić. But the argument we kept hearing was about experience, and that's where it doesn't hold.
Half of Montenegro's starting XI in Yerevan had 10 caps or fewer, and the team's average age was 25.9. For Armenia's XI that line sat at 23–24 caps, with an average age of 26.8. In Podgorica it was the same picture: 10 for Montenegro, 18 for Armenia, and the two teams were the same age to within about a month.
In between those games Montenegro sent their senior team to Riga, where half the starting XI had 38 caps or more and the average age was 29.4. The younger side took four points off Armenia over the two games.
Armenia started the home game with five at the back against Montenegro's two strikers. The first half was even, 0.51 xG to 0.50. The second half wasn't.
After the game Melikyan said Armenia "had the upper hand in the second half and put them under good pressure". The data shows the reverse. After the break Montenegro had 1.01 xG to Armenia's 0.38, and 32 entries into the final third to Armenia's 22. Over the whole match Armenia had 55% of the ball and 11 touches in Montenegro's box. Montenegro had 20 in Armenia's.
There was one good spell. In the minutes after Spertsyan's solo equaliser, Armenia had four shots between the 67th and 73rd minutes, the best of them a 0.15 xG chance for Narek Hovhannisyan. It felt more like a wave of emotion than control of the game, and it didn't last. Armenia didn't have another shot after the 73rd minute, while Montenegro had three, the last of them Latković's winner in the 89th. Outside that ten-minute spell, Montenegro were the better team after the break.
The duels
The obvious explanation for that night is that Armenia lost the physical battle. We expected the numbers to show it, and they don't.
Armenia won 56% of ground duels and 55% of aerial duels in that game. Over the four games they had the best ground-duel record in the group (53.1%), and in Podgorica it was 59%. As a team, they held their own.
The team figure hides how uneven it was, though.
Eduard Spertsyan won 11 of his 16 duels and Artem Bandikyan won 7 of 8, which covers most of Armenia's margin. In central midfield, where opposition attacks are supposed to be broken up, Ugochukwu Iwu had two duels in 55 minutes, lost both and didn't make a tackle. Up front, Artur Miranyan and Zhirayr Shaghoyan were 0 from 6 between them.
Karlen Hovhannisyan replaced Iwu and, in 35 minutes, made three tackles, committed two fouls and won five of his eight duels. If he looked like the first Armenian midfielder willing to get stuck in, the numbers say that's because he was.
The foul count supports the complaint that Armenia were too soft. Montenegro committed 16 fouls to Armenia's 10, which is what it looks like when one team is willing to stop attacks early and the other isn't. Still, the bigger issue was what Armenia did after winning the ball. Montenegro got 61 entries into the final third out of 45% possession. Armenia got 46 out of 55%.
Plenty of the ball, nowhere to go
Over the four games Armenia had the most cautious attack in the group. You can see it in every measure of chance quality we have, starting with where the shots came from.
Most of Armenia's shots came from outside the box. Most of Montenegro's came from inside it. Only 47% of Armenia's non-penalty shots were taken inside the area, compared with 62% for Latvia, 62% for Cyprus and 72% for Montenegro. The typical Armenia shot was taken from 21.7 metres, and the average one was worth 0.061 xG, the lowest in the group.
If you only count shots with a decent chance of going in, 0.15 xG or better, Armenia had four in four games. Latvia, bottom of the group with one point, had six. Cyprus had eight and Montenegro sixteen.
People often say Armenia passed backwards too much. That isn't quite what the data shows. Measured by distance, Armenia moved the ball forward more than any other team in the group (163 metres of progression per 100 passes) and hit 55 long balls a game. The ball did go forward. It just rarely ended up in the box: 2.9 touches in the opposition area per 100 passes, the lowest in the group. That matches what the games looked like. There were lots of sideways and long passes, and very little between the lines or in behind.
It also meant the centre-forward hardly saw the ball.
Miranyan started all four games and received 11 passes per 90 minutes. Krstović received 23 and Latvia's Roberts Uldriķis 21. In Podgorica Miranyan got nine passes in 74 minutes.
Over the window Armenia averaged 0.74 xG a game, against Latvia's 0.66. Armenia started the window second in the group on Elo and on squad value, and finished it with the attack of a bottom team.
Nicosia: 38 minutes against ten men
Cyprus scored twice in the first half from very little. Leonidas Konomis scored after Arsen Beglaryan dropped a corner, and Andronikos Kakoullis finished a counter that started with a Karlen Hovhannisyan turnover. The two shots were worth 0.28 xG combined.
Across the window Armenia conceded five goals: an own goal, plus four shots worth 0.65 xG in total. In those two defeats the goalkeepers let in 2.65 more goals than the shots they faced would normally produce. That's a real problem, but a separate one.
The second half in Nicosia is the part worth watching again. Kostas Pileas was sent off in the 52nd minute, which gave Armenia almost 40 minutes against ten men to find two goals.
After half-time Armenia had 84% of the ball and entered the final third 52 times. They crossed 23 times and found a teammate three times. They had eleven shots worth 0.59 xG in total, eight of them from distance. Their best chance after the red card was a corner rebound for Iwu, worth 0.21 xG. Their best chance from open play was worth 0.09. Cyprus had one shot in the whole half.
Armenia have played this kind of game many times: the opponent sits deep and waits. There was no sign of a plan to break them down, so the team kept crossing.
Then the attackers ran out. After Miranyan came off, Armenia had no centre-forward on the pitch. "When we had no attacking players left," Melikyan said afterwards, "I sent Georgiy Harutyunyan up front." Harutyunyan is a centre-back. Armenia had called up 29 players. Shaghoyan, a winger, had been left out of the matchday squad that night for "rotation". Arayik Eloyan, a centre-forward and the squad's in-form APL striker with four league goals, had been moved down to the U21s four days earlier. Gevorg Tarakhchyan, a winger with four APL goals, hadn't been called up at all.
So there was a proper number nine available in Eloyan, and he wasn't there. But the bigger point is that you don't need a specialist striker to break down ten men. You need players who can carry the ball, combine in tight spaces and shoot, and Armenia had at least one on the pitch. Karen Nalbandyan came on after 63 minutes and, in under half an hour, had three shots, all from 20 metres or more, and won four of his five duels. He is the highest-rated APL regular Melikyan called (7.53 on Sofascore, with four goals and four assists for Alashkert this season). Three days later in Podgorica, in a game where Armenia barely threatened and a midfielder who shoots from distance could have made a difference, he stayed on the bench for all 90 minutes. Armenia used four of their five substitutions.
Podgorica
The best performance of the window came in the last game.
Montenegro had 64% of the ball and 92 entries into Armenia's final third, and produced 0.52 xG. In their other three games they managed 3.13, 2.41 and 1.51. Armenia won 59% of their ground duels, their best of the window, and ran 112.6 km by UEFA's tracking, more than Montenegro and more than in any of their other games. Beglaryan, three days after his mistake in Nicosia, made five saves.
A few things had changed. Armenia played a back four. Before the window Melikyan had said that was the plan ("we planned to play with four defenders") until the injuries pushed him back to five. Spertsyan started on the bench with a knock. And with first place realistically gone, there was less pressure.
That last point is our reading, not something the data proves, but it fits. The one game in which Armenia weren't trying not to lose was the one where they looked most comfortable. Going forward it was still thin (0.21 xG), and they did little with the last few minutes after Lončar's red card. But they held the best team in the group to almost nothing, with the same injury list as before.
Young and inexperienced?
"Young" and "inexperienced" came up again and again. Before Podgorica: "We have a very young team now, so it will be difficult." After it: "It is a young national team." The caps chart above shows why that doesn't work as an explanation. Against the team Armenia worried about most, Armenia were neither the less experienced side nor the younger one.
The coach wasn't consistent about it either. Before the window he explained why he had brought back players he had previously moved on from:
So the window started with Ognjen Čančarević (37) and Karen Muradyan (33) back in the squad, and a debut for Argishti Petrosyan at 33. The same reasoning picked the goalkeeper in Nicosia. Asked why he hadn't started Aleksandr Mishiev or Gor Matinyan, Melikyan said: "It is still too early for them. They need to gain experience … so we decided to rely on experience. But it did not work."
The experienced keeper made the mistake that decided that game. That doesn't prove a younger one would have done better. It does show that experience was being used as a reason without anyone checking whether it was actually helping.
Ready for what?
If experience isn't the test, form should be. Melikyan was asked about this before the window started. A journalist mentioned two APL players who had been scoring regularly, Levon Vardanyan of Sardarapat and Gevorg Tarakhchyan of Pyunik. On Vardanyan, Melikyan said it was "too early to call him up". On Tarakhchyan:
Armenia then lost fourteen players and ran out of attackers in Nicosia, and Tarakhchyan still wasn't called. Here is how much club football the attacking players who did play had behind them.
Artur Miranyan played 66 minutes of club football between mid-July and the start of the window. Melikyan acknowledged that Miranyan "had problems at his club in the summer and didn't play much", and started him in all four games anyway. Narek Grigoryan had played 45 club minutes; Melikyan said he couldn't "remember when he last played". Edgar Sevikyan, in the coach's words, "lacks regular playing time".
Tarakhchyan, by comparison, had played 406 APL minutes and scored four goals, with the best xG rate of any Armenian on 300 or more APL minutes (0.77 per 90). Vardanyan had started all eight of Sardarapat's games, with two goals and five assists. His teammate Levon Petrosyan had four goals.
The APL players Melikyan did call didn't get much of a chance either. Karen Nalbandyan has four goals and four assists for Alashkert this season, and his 7.53 Sofascore rating is the highest of any APL regular in the squad. He played 28 minutes in four games, all of them in Nicosia. Gor Manvelyan, with three goals and three assists for Noah, got 135 minutes, one of them at home to Montenegro. Both offer exactly what the attack was short of: players who take on the ball between the lines and shoot. Both spent most of the window on the bench while Miranyan, with 66 club minutes behind him, played 261.
This doesn't mean Melikyan ignored the APL. Among outfielders with at least 300 minutes, the players he called generally rate slightly higher on Sofascore (6.85) than the ones he didn't (6.70). The point is narrower than that. In the positions where Armenia were short, in-form APL players were left out in favour of players who weren't getting games for their clubs.
Defence shows the same pattern.
Three uncalled APL defenders rate higher than Argishti Petrosyan, the 34-year-old who started at home against Montenegro: Sergey Kochkanyan of Sardarapat (7.04), Hakob Hakobyan of Alashkert (6.95) and Shirak's 21-year-old Suren Tsarukyan (6.93), who was with the U21s during the window. Shirak's defenders (Tsarukyan, Hovhannes Pahlevanyan, Tigran Sumbulyan and Hamlet Mnatsakanyan) have played almost every minute of their club's season, which is exactly the playing time Melikyan said some of his picks were missing. A Sofascore rating is a rough measure, but it's the same rough measure for every player in the league.
The usual objection is that the APL isn't international football. Fair enough, but look at who these opponents actually are.
Latvia's whole squad is valued at €14.8m. Noah's is valued at €14.1m. Four of Latvia's 27 players are at Riga FC, the Latvian champions Ararat-Armenia knocked out of Champions League qualifying in July, and another four are at RFS. Four players from those two clubs started against Armenia. Montenegro's reserve goalkeeper plays in Latvia too.
Cyprus is a different case. Their league is a lot stronger than the APL, and we're not claiming every opponent was APL level. What we are saying is that these are national teams at Armenia's own level, ranked 85th, 118th and 149th in the world by Elo, with Armenia 116th. Against teams like that, an APL player in good form is a sensible option. This squad ran out of options and still didn't turn to them.
The calendar
"Four games in eleven days" was the explanation we heard most often. Every team in the group had exactly the same calendar on the same dates, and all of them travelled.
Armenia didn't rotate any less than the others. They used 23 players, Montenegro 24, and Cyprus and Latvia 21 each. Armenia made ten changes to their starting XI over the window, the same as Latvia. Cyprus made eleven and Montenegro fifteen. The group leaders rotated the most and still took four points off Armenia with a second-string XI.
The difference is that nearly all of Armenia's changes were forced: injuries to Bueno, Čančarević and Bandikyan, suspensions for Serobyan and Sevikyan, and Spertsyan's knock. When Melikyan did have a free choice, he mostly kept the same players on. The clearest case is the Latvia game.
Armenia were 2–0 up from the 75th minute against Latvia, with three games to play in the next eight days. These players stayed on for the full 90:
Iwu, just back from a long injury. Three days later Melikyan took him off after 55 minutes against Montenegro: "I felt he was very tired … It will be difficult for him to play all four matches."
Bandikyan, who, Melikyan said that same night, had "a knee problem — another player in that situation would already have gone for surgery". He was injured in Nicosia.
Spertsyan, who according to Melikyan after Podgorica "has now played four matches while dealing with pain".
Tiknizyan, Piloyan and Harutyunyan also played the whole game, and finished the window on 360, 359 and 352 minutes out of 360. After the Latvia game Melikyan said: "If we had only one game left, I could use them to the very limit … But we have three more matches, and we have to think ahead." We'd argue the last fifteen minutes against Latvia were exactly the time to do that. Any APL player who's playing every week can see out a 2–0 against Latvia.
"I no longer want to talk about problems"
That was how Melikyan opened his interview of 24 September. So we counted how often he did.
In nine press appearances, about 8,000 words of answers, Melikyan used injury, difficult, problem, absence and losses 98 times, roughly once every 81 words. He said four times that he wasn't making excuses. Tactics came up four times, pressing once and space not at all.
These are edited translations, so if anything the real count is higher. Nobody expects a coach to pretend injuries don't exist. But players hear what their coach says in public, and for two weeks the main subject was what the team didn't have. It's not surprising that the team so often played as if it had something to protect.
The future
Another word kept coming back. After the 2–3: "I'm sure this team has a good future." Before Cyprus: "These players have a big future." After Podgorica: "This team has a future. It is a young national team."
Most Armenia coaches of recent years have said something similar. Each one starts a rebuild and asks to be judged on a future that hasn't arrived yet. This is what the last five years look like.
Armenia's Elo peaked at 1588 in September 2021. After Nicosia it was 1341, the lowest since April 1997, a drop of 247 points across four permanent coaches. Melikyan's own record is 13 games: two wins, three draws and eight defeats, with 10 goals scored and 28 conceded.
International football doesn't really have a long term. It comes in windows of two or four games, and the points count straight away. The job is to pick the players who are best and in form now, for the opponents in front of you. Age and caps don't come into it.
What went right
Some things went Armenia's way, and they're worth spelling out.
Going by the quality of chances, Armenia's four games were worth about 5.5 points. They got four. Cyprus's were worth about 5.4, and they got seven. Most of that gap is goalkeeping and finishing. By xG Armenia had the best defence in the group, conceding 0.70 a game, and the two goalkeeping errors cost them more than two goals. The table is a bit harsh on them.
That also means the margins were tight, so the smaller decisions mattered more: who led the line, who played the last fifteen minutes against Latvia, and whether the home game against Montenegro was treated as one to win. There's nothing to fault in the effort. Armenia ran between 105 and 113 km a game on UEFA's tracking. They did the least high-intensity running in the group, though (5.16 km a game, against 5.35 to 5.70 for the others), and Latvia outran them in Yerevan. That's a question of tempo and approach. Nobody was hiding.
Before Cyprus
Leave the numbers aside for a moment. Forget the ratings, the market values, the old head-to-heads and the injury list, and just watch the four games again. It's hard to find a stretch where Armenia were outclassed. Latvia offered nothing. Cyprus scored twice from mistakes and then sat back. Montenegro were the best team in the group and still looked beatable in both games. What you're left with is the feeling that at almost any point Armenia could have stepped on the gas and gone for it, and kept choosing not to, as if something was holding the team back on the safe side. That, for us, was the most frustrating part of the window, and it's what most of the numbers above keep coming back to.
Armenia started this window second in the group on Elo and on squad value, with two home games in which they were favourites and the same calendar as everyone else. After four of six rounds they're third.
Injuries played a part. Most of what went wrong had nothing to do with them. A beatable Montenegro side was treated as the favourite, and in Yerevan its younger XI outplayed Armenia for most of the second half while the coach said Armenia had the upper hand. The attack created four good chances in four games and nothing against ten men. In-form APL players stayed at home while players without club minutes started, and in Nicosia the squad ran out of attackers. And most of what was said in public was about what the team was missing.
All of that can change by 12 November, when Cyprus come to Yerevan with second place and a promotion play-off at stake. Armenia need to win, and they need to win by three or more to take the head-to-head outright if the two teams finish level. Elo will have Armenia as favourites again. Before that game we'll be looking at three things: whether the squad is picked on current form or on experience again, whether Armenia set up at home to win rather than not to lose, and whether anyone at the press conference calls Cyprus the favourites.
The last two rounds, Armenia v Cyprus on 12 November and Latvia v Armenia on 15 November, are still to be played in Nations Draft, our free Nations League game.
Match tracking (distance, high-intensity running, pass networks) from UEFA. Shots, expected goals, duels and player ratings from Sofascore. Elo ratings from eloratings.net. Squad lists and caps from national-team squad tables, ages from UEFA line-ups. Market values from Transfermarkt, 6 October 2026. APL minutes, goals and assists from the APLGames Fantasy database. Press-conference quotes from sportaran.com's published transcripts. All counts and models are our own.