AI predictions for the UEFA Nations League: this round's picks

The AI's strongest pick of the round: France to win in Turkey, at odds 1.51.

📊 October: matchdays 3 and 4

The October window went the opposite way to September. The AI's free picks won 1 of 6: this time they were not on clear favourites to win but on goals and the draw, and the games broke the other way. The High Stake held up at 2 of 3, and the Draw Martingale had its best week.

At 1 unit per pick, October finished at −4.10 units. Across the four matchdays the AI is 6 of 12 and −2.46 units. We publish it the same way as when it wins: every pick stays on the public record.

The High Stake added +5.76 units in October and is 7 of 9 in the Nations League (+33.9 units). The Draw Martingale hit two of the three draws it went for: Bosnia-Sweden (3.60) closed a cycle that came after seven misses, and Greece-Germany (3.80) closed the next one on the first step.

🏁 Window wrap-up: 5 of 6 free picks and 5 of 6 High Stake

The Nations League returns on Thursday 1 October, but the September window already has its balance sheet. This is how Gambeta's three products did over matchdays 1 and 2:

Matchdays 1 and 2 of the Nations League are done and the AI's record is this: of the six free predictions, five won. The only miss was Poland, who could not get past a 0-0 against Bosnia and Herzegovina. Staking 1 unit on each pick, the six returned +1.64 units, 27% on the amount staked.

The human analyst's VIP picks, the High Stake, went five for six in the Nations League, all in shots and corners markets, where the gap between national teams shows more clearly than in the result.

The last one, on Tuesday 29, did not land: the High Stake for Czech Republic vs England (Czech Republic over 8.5 shots, at 1.76, 9/10 stake) was the only miss of the window in the Nations League. Like every pick, it stays in the public record with its result, and the next High Stake goes on sale on the High Stake page.

With each pick's stake (from 8 to 10 out of 10), the six High Stake picks added up to 56 units staked and returned +28.1 units: a 50% return. With a flat 1 unit per pick the balance would have been +2.94 units (+49%). The AI's free picks are always measured at 1 unit per pick.

The Draw Martingale showed the other side of the window: it won on Thursday 24 (Netherlands-Germany, at 3.75) and then none of the following draws landed (Turkey-France, England-Spain, Racing-Boca, Sweden-Poland and, on Tuesday 29, Spain-Croatia at 7.00). That is how the system works: each step raises the stake so that a single draw covers what was lost, and open cycles carry on with the next pick.

What the window left us

• The clear favourites delivered. All six free picks were on the match winner, between 1.51 and 1.73, and five won. The only slip was a 0-0 (Poland-Bosnia), not an upset by the opponent.
• Shots and corners, the High Stake's gold mine. All six of the analyst's picks were on shots or corners for the home side, where the gap between national teams shows more than on the scoreboard: five out of six.
• Draws were the exception. Of the five the Martingale went after in the Nations League, one landed, at 3.75. That is why it is played as a cycle and not as single bets.

✅ Update, 25 September: the Max Stake and the draw Martingale both won

Thursday 24 in the Nations League: two VIP picks, two winners.

Thursday's classic went our way twice. The Max Stake 10/10 on Netherlands vs Germany was over 13.5 shots for the Netherlands, at odds 1.86: won ✅. And the pick of the draw Martingale was the draw in that same match, at odds 3.75: won ✅. Anyone who bought the Martingale cycle for $0.99 staked $10 at step 1 and collected $37.50: the cycle closed in profit on the very first pick.

With this result the High Stake record stands at 20 wins and 16 losses, with a 5.4% ROI weighted by stake, every pick public on the High Stake page. The AI's free picks for Friday are still pending, and three more have been added for the weekend:

AI Nations League predictions at a glance

The 2026-27 UEFA Nations League kicked off on Thursday 24 September with 46 matches across two matchdays, running until Tuesday 29. Gambeta's artificial intelligence reviewed every one of them and only published, for free, the ones that passed its filter. Here is the summary of this round's AI Nations League predictions, together with the two VIP products we have in play:

The odds are the ones at which each pick was published; the line can move until kick-off. The no-vig probability comes from the best market prices for the full 1X2 (draw included), once the bookmakers' margin is removed. Kick-off times are UK time (BST): add one hour for Central Europe (CEST), subtract five for New York (ET).

Turkey vs France prediction: AI backs France to win (1.51)

This is the AI's strongest pick of the round. France visit Turkey on Friday 25 at 19:45 UK time as one of the most consistent national teams of the last decade: world champions in 2018, finalists in 2022 and winners of the Nations League itself in 2021. The market makes them clear favourites even away from home: the best price on a French win is around 1.46, against 4.80 for the draw and 7.54 for Turkey. Without the margin that is roughly 67% for France, 20% for the draw and 13% for the hosts.

The detail we like most about this prediction: the AI published it at 1.51 and the price has since shortened to 1.46. When the market moves towards your pick after you took it, it is the best sign that you got a good price; bettors call it beating the closing line.

Where is the risk? Turkey at home is one of the loudest settings in Europe, and a first round of international football always comes with some rust. If you want a safer version of the same prediction, the double chance «France or draw» pays much less but covers the draw. The artificial intelligence's verdict, though, is straightforward: France to win. The full prediction is in the Turkey vs France preview.

Sweden vs Romania prediction: AI backs Sweden to win (1.54)

Sweden at home are the favourites with every bookmaker we track: the AI took them at 1.54.

Sweden host Romania on the same Friday 25 at 19:45 UK time. The best market prices read 1.53 for the home side, 4.66 for the draw and 6.30 for the visitors: 64% Sweden, 21% draw and 15% Romania once the margin is removed. The price has barely moved since the AI published the pick (1.54 → 1.53), so the bookmakers' consensus is holding firm.

The AI's reading is that of a home side clearly superior on paper, with a forward line that can hold its own against Europe's best. Romania, on the other hand, are an organised side who know how to compete: they topped their group at Euro 2024. That is why the model does not go for handicaps or big-win markets: the Sweden vs Romania prediction is simply a Swedish win. The analysis is in the Sweden vs Romania preview.

Poland vs Bosnia and Herzegovina prediction: AI backs Poland to win (1.73)

The highest-paying of the three picks: Poland to win at home at odds 1.73.

This is the pick with the biggest odds of the batch. Poland host Bosnia and Herzegovina on Friday 25 at 19:45 UK time, with 1.70 on a Polish win, 3.94 on the draw and 5.40 on Bosnia at the best prices. Without the margin: 57% Poland, 25% draw and 18% Bosnia.

That is why it pays the most and also carries the most risk: roughly one in four matches with this profile ends in a draw. The AI publishes it because Poland are home favourites by consensus across all bookmakers and the published price (1.73) sits inside its «favourite without excessive risk» band. If you prefer some cover, the 1X double chance (Poland or draw) is the conservative version of the same Poland vs Bosnia prediction. The prediction is in the Poland vs Bosnia preview.

Albania vs Belarus prediction: AI backs Albania to win (1.55)

Albania host Belarus on Saturday 26 at 19:45 UK time. It is a side that played two of the last three European Championships (2016 and 2024), and at home it starts as the clear favourite by consensus across all bookmakers. The AI published the win at 1.55, inside its «favourite without excessive risk» band. The analysis is in the Albania vs Belarus preview.

Denmark vs Wales prediction: AI backs Denmark to win (1.53)

Denmark receive Wales on Sunday 27 at 17:00 UK time. Two nations with recent semi-finals in their history (Denmark at Euro 2020, Wales at Euro 2016), but the market has no doubt about the home side: the AI took Denmark to win at 1.53. The prediction is in the Denmark vs Wales preview.

Austria vs Kosovo prediction: AI backs Austria to win (1.51)

Austria, who topped their group at Euro 2024 ahead of France and the Netherlands, host Kosovo on Sunday 27 at 17:00 UK time. Kosovo is UEFA's youngest member (admitted in 2016) and arrives as a clear underdog. The AI's pick: Austria to win at 1.51. Details in the Austria vs Kosovo preview.

The AI Nations League accumulator: odds 4.02

Many readers ask us about this round's accumulator. If you combine the three free picks, the odds are 1.51 × 1.54 × 1.73 = 4.02. With today's no-vig probabilities, the chance of all three landing is around 24%: one in four. It is a high-price bet, so if you play it, keep the stake small; the picks are designed to be played as singles. To learn how to build accumulators sensibly, read the guide to AI parlay betting.

Netherlands vs Germany: this round's Max Stake (prediction on sale)

Thursday's classic has no free pick: it is the round's human Max Stake.

Thursday 24's big match (19:45 UK time, 20:45 CEST) is a European classic: Netherlands vs Germany, two nations that met in the 1974 World Cup final and in the Euro 1988 semi-final. Here the market has no favourite: 2.54 for the Oranje, 3.81 for the draw and 2.68 for Germany, or 38%, 26% and 36% without the margin. That is exactly why the AI does not publish a free pick: there is no clear favourite that passes its filter.

But it is the match we chose for the human High Stake, and not at any level: it is a Max Stake, 10 out of 10, the highest confidence we give. This prediction is not made by the AI but by a human analyst, and it is sold separately for $4.99, with no subscription: you receive it by email and in your account before kick-off. And if the Max Stake loses, you get 14 days of High Stakes for free.

The High Stake record is public, wins and losses included: as of 24 September it stands at 19 wins and 16 losses, with a positive ROI. You can check it pick by pick on the High Stake page, where the Netherlands vs Germany pick is also on sale. If you are reading this after Thursday, what you will find on sale is the next High Stake: there is always one in preparation.

Result ✅: the Max Stake on Netherlands vs Germany was over 13.5 shots for the Netherlands at 1.86, and it won. It is now part of the public record with its odds and its stake.

Italy vs Belgium: today's Max Stake (prediction on sale)

After Thursday's win, the new Max Stake 10/10 is Italy vs Belgium, on Friday 25 at 19:45 UK time. As with Netherlands vs Germany, the AI has no free pick here because the market has no clear favourite: it is a pick made by a human analyst, sold separately for $4.99 with no subscription, and if it loses you get 14 days of High Stakes for free. It is on sale on the High Stake page until kick-off.

Draw Martingale for the Nations League for $0.99

The other product we have in play this round is the VIP draw Martingale, and its pick for Thursday 24 is precisely from the UEFA Nations League, at 19:45 UK time. We will not tell you here which match it is, because that is what is sold, but here is how it works:

• You pay $0.99 once and your cycle opens. It is not a subscription.
• The first draw pick reaches your inbox to stake $10.
• If it misses, the next one reaches you free and at double: $20, $40, $80…
• When the draw lands, the cycle closes in profit: draw odds of 3 or more cover everything staked before and leave a profit.
• The cap is step 11 ($10,000): if that one misses too, the cycle closes. A voided match does not count.

The Nations League suits this strategy very well, because between evenly matched national teams the draw is a common result: in England vs Spain, to take one example, the market gives it a 28% chance. The full details, the ladder of steps and the public record are on the draw Martingale page.

Result ✅: Thursday's pick was the draw in Netherlands vs Germany at 3.75, and the match ended level. Every cycle that started with that pick closed in profit at step 1: $10 staked, $37.50 collected. There is a new pick on Friday 25, also from the Nations League: to receive it you open a new cycle for $0.99.

England vs Spain and the matches where the AI publishes no pick

The rematch of the Euro 2024 final, which Spain won 2-1 against England in Berlin, is played on Saturday 26 at 19:45 UK time. It is the most searched match of the round and yet the AI has no prediction: 2.96 for England, 3.42 for the draw and 2.51 for Spain. When the favourite is not clear, the model prefers not to force a pick. The England vs Spain prediction is in the same situation as Netherlands vs Germany, Italy vs Belgium (Friday 25) or Norway vs Portugal (Sunday 27): huge matches, but none has a favourite that passes the filter.

It is a deliberate decision: we prefer to publish fewer picks and have each one well grounded. For the second matchday, from Sunday 27 to Tuesday 29, there are fixtures with a clearer favourite, such as Germany vs Greece, Spain vs Croatia or Czechia vs England. If any of them passes the AI's filter, you will see it first in today's predictions.

UEFA Nations League fixtures: matchdays 1 and 2 (24–29 September)

Six days, 46 matches: the first window of the 2026-27 Nations League.

These are all the matches of the first window of the 2026-27 Nations League that we track in the market, in UK time (BST). In bold, the ones with a Gambeta pick. Kick-off times can change: check them before betting.

Thursday 24 September

• 17:00 · Andorra vs Malta
• 19:45 · Netherlands vs Germany (VIP Max Stake)
• 19:45 · Austria vs Israel · Norway vs Denmark · Serbia vs Greece · Kosovo vs Republic of Ireland · Liechtenstein vs Lithuania · Portugal vs Wales

Friday 25 September

• 17:00 · Armenia vs Latvia · Georgia vs Northern Ireland
• 19:45 · Turkey vs France (AI: France to win) · Sweden vs Romania (AI: Sweden to win) · Poland vs Bosnia and Herzegovina (AI: Poland to win)
• 19:45 · Italy vs Belgium · Montenegro vs Cyprus · Hungary vs Ukraine

Saturday 26 September

• 14:00 · Slovenia vs Scotland
• 17:00 · Bulgaria vs Luxembourg · Iceland vs Estonia · Faroe Islands vs Kazakhstan · San Marino vs Finland
• 19:45 · England vs Spain · Czechia vs Croatia · Albania vs Belarus · Slovakia vs Moldova · North Macedonia vs Switzerland

Sunday 27 September

• 14:00 · Lithuania vs Azerbaijan
• 17:00 · Serbia vs Netherlands · Denmark vs Wales · Austria vs Kosovo · Gibraltar vs Andorra
• 19:45 · Germany vs Greece · Norway vs Portugal · Israel vs Republic of Ireland

Monday 28 September

• 17:00 · Georgia vs Ukraine · Armenia vs Montenegro · Latvia vs Cyprus
• 19:45 · Belgium vs France · Turkey vs Italy · Sweden vs Poland · Romania vs Bosnia and Herzegovina · Northern Ireland vs Hungary

Tuesday 29 September

• 19:45 · Spain vs Croatia · Czechia vs England · Scotland vs Switzerland · Slovenia vs North Macedonia

How Gambeta's AI predicts the Nations League

The AI does not guess: it reads the consensus of around 20 bookmakers and only publishes when the reading is solid.

Gambeta's artificial intelligence does not guess results. For every match it compares the odds of around 20 bookmakers, strips out their margin and calculates the real probability the market assigns to each outcome. In club leagues, that reading is also crossed with recent form, expected goals and each team's context.

The Nations League has one particularity worth stating plainly: the 2026-27 edition has only just started, so there are no matches of this competition yet from which to measure form. That is why, in this first round, the AI only publishes clear favourites by market consensus, with odds between 1.50 and 1.75 and one pick per match. It is a rule of prudence: once the competition has a few rounds played, the full model comes into play.

Every pick is recorded with its odds before kick-off and is settled automatically when the match ends: win or lose, it goes to the AI's public track record. We do not delete losses. To dig deeper, read the guides to AI for sports betting and AI football predictions, and the comparison AI vs tipster.

Tips for betting on the Nations League with AI

• Check the line-ups. In international windows coaches rotate and an absence weighs more than at club level: confirm the starting XI before betting.
• Home favourites are the most stable picks. Evenly matched games are better for draw or goals markets than for the winner.
• Be careful with long accumulators. Three favourites together are already a one-in-four bet.
• Fixed stakes and only what you can afford to lose. No prediction, from an AI or a human, is a sure thing.

Frequently asked questions about AI Nations League predictions

How did the AI Nations League predictions do?

In September the AI won 5 of 6 free picks, all on the match winner: France, Sweden, Albania, Denmark and Austria. In October it won 1 of 6, with picks on goals and the draw (only Under 3.5 in Netherlands-Serbia landed). In total, 6 of 12 and −2.46 units at 1 unit per pick.

Are the AI predictions free?

Yes. The artificial intelligence's picks are free and published before kick-off with their odds. What is sold separately are the human predictions: the High Stake ($4.99 per pick) and the VIP draw Martingale ($0.99 per cycle).

Did the Netherlands vs Germany Max Stake win?

Yes. It was over 13.5 shots for the Netherlands at odds 1.86, and it won. The draw Martingale pick in the same match (the draw at 3.75) won too.

How did the High Stake do in the Nations League?

Seven of nine: 5 of 6 in September and 2 of 3 in October, almost all on shots and corners. With each pick's stake it adds up to +33.9 units. Every pick stays on the public record with its result.

Why does the AI have no pick for England vs Spain?

Because the market has no clear favourite: 2.96 England, 3.42 the draw and 2.51 Spain. The AI only publishes when the reading is solid, and it prefers not to force a prediction.

How does the draw Martingale work?

You pay $0.99 per cycle. You start with a draw pick at $10; if it misses, the next one reaches you free at double, until the draw lands or until step 11 ($10,000). The first cycle of this Nations League closed in profit at step 1.

Where can I see whether the predictions won?

In Gambeta's public track record: every AI pick is settled automatically when the match ends, with its score and its result, win or lose.