The road to Budapest — decoded by data, not vibes
Our AI gives Paris Saint-Germain the highest probability of winning the 2025/26 Champions League at 34.1%, making them clear favorites ahead of Bayern Munich (24.7%), Arsenal (21.4%), and Atletico Madrid (19.8%). The model's most common simulated final is PSG vs Arsenal, with PSG winning 2-1 through a late Vitinha goal — which would make them the first back-to-back Champions League winners since Real Madrid in 2016.
The 2025/26 UEFA Champions League has served up perhaps the most brutal knockout bracket in living memory. Liverpool, Real Madrid and Barcelona are all gone. The team sitting unbeaten at the top of the Premier League survived Sporting CP by a single goal. The defending champions rolled into the semifinals on the back of a 4-0 aggregate demolition of Anfield. Bayern Munich traded six goals with Real Madrid and came out on top. Atletico Madrid outworked their biggest rivals on the continent.
With the first legs kicking off on April 28, we wanted to cut through the narratives and let the data do the talking. We fed terabytes of information into our AI — league phase xG, knockout round heat maps, home/away splits, closing line drift, squad availability, referee profiles, even travel load and rest days — and asked it one simple thing: who lifts the trophy at the Puskás Aréna on May 30?
Here is what the machine sees.
Paris Saint-Germain enter the semifinals as reigning European champions and as the AI's clear favorites to lift the trophy for a second straight season. Luis Enrique's side has been the most complete team in the competition — a blistering league phase finish, a round-of-16 statement, and a 4-0 aggregate dismantling of Liverpool in the quarterfinals that included a 2-0 win at Anfield.
Reigning Ballon d'Or winner Ousmane Dembélé has chipped in with four Champions League goals, two of them against Liverpool. But the story of this PSG team is depth. Khvicha Kvaratskhelia, Désiré Doué and Bradley Barcola rotate around him without the offense missing a beat, and the midfield of Vitinha, João Neves and Warren Zaïre-Emery is one of the best ball-progression units our model has seen in the Champions League era.
The AI assigns PSG a 34.1% probability of winning the trophy — the highest in the field. The concern? They run into a Bayern side that has beaten them in five of their last seven meetings. No team is scarier on a single-leg knife's edge.
Bayern Munich have the most potent front three in Europe — full stop. Luis Díaz, Harry Kane and Michael Olise have combined for 32 goals and assists across just 14 Champions League matches this season, a rate our model ranks as the most efficient attacking trident since the 2019/20 Bayern treble team.
Kane alone has been the revelation. Going into the semifinals he has 16 goals in his last 15 Champions League appearances, and he has scored in the away leg of Bayern's quarterfinal in each of the last three seasons — Arsenal, Inter, Real Madrid. The pattern is too consistent to be noise. Add in the thrilling 4-3 victory over Real Madrid at the Allianz Arena (6-4 on aggregate) and it is easy to see why Vincent Kompany's side is the AI's second favorite at 24.7%.
The tie against PSG is a rematch of the Covid-era 2020 final in Lisbon, and the two clubs have now met seven times since — Bayern winning five, including earlier this season in the league phase. Home-leg advantage swings to Bayern for the decisive second leg on May 6. That matters more than pundits typically price in.
Arsenal are the story of the European season. Mikel Arteta's side is the only remaining unbeaten team in the competition — ten wins and two draws from their 12 Champions League matches so far. That alone would make them fascinating. The fact that they are also last year's semifinalists returning with unfinished business makes them dangerous.
The concern is that the attack has sputtered when it matters most. The quarterfinal tie with Sporting CP finished 1-0 on aggregate after a nervy goalless second leg at the Emirates. Kai Havertz remains the reluctant number nine and Bukayo Saka is carrying the creative load. Our model's Expected Goals number across the knockout rounds is well below league phase output — a flag that the Gunners are grinding results rather than dominating them.
The AI gives Arsenal a 21.4% chance of winning the final. They get a semifinal against an Atletico Madrid team built to frustrate them. It will not be pretty, and the machine isn't sure it will be enough — but nobody wants to face Arsenal in a final in Budapest.
Atletico Madrid are the chaos variable. Diego Simeone — still in charge since 2011 — has quietly tweaked his tactical DNA this season. The low block remains, but Ademola Lookman and Julián Álvarez give Atleti genuine transition speed for the first time in years. They have scored 34 Champions League goals this campaign, at least 10 more than in either of the last two.
Their path here has been earned. A 3-2 aggregate win over Barcelona in the quarterfinals was the kind of result that Atletico used to specialize in before the post-Griezmann-Costa-Felix rebuild muddied their identity. This feels like the full-circle return to what made them Champions League finalists in 2014 and 2016.
And yet, the AI is lukewarm. Atleti enter as the longest shot at 19.8%, largely because the model does not love their goals-conceded profile against elite transition teams. If they can drag Arsenal into a rock fight, anything is possible. If Arteta's side plays its game, the tie is Arsenal's to lose.
Run the simulation 10,000 times and the most common final pairing the model produces is Paris Saint-Germain against Arsenal — a rematch of the league phase fixture that PSG won 2-0 at the Emirates back in October. The second most common is PSG vs Atletico Madrid. Either way, the road to the trophy runs through Paris.
The model projects PSG to edge Bayern over two legs in a tie decided by a narrow xG advantage in the second leg. Arsenal's unbeaten record survives Atletico by the thinnest margin, with a late goal at the Emirates on May 5 tipping an otherwise goalless tie.
The final? PSG 2-1 Arsenal. Dembélé opens the scoring, Saka equalizes from the spot, and a late Vitinha winner delivers back-to-back Champions League crowns to Paris — something only Real Madrid have done in the past 36 years. The betting markets currently price PSG around +220 for the outright, which — if you trust our model's calibrated probabilities over the consensus — implies genuine value betting edge at those odds. For tracking whether that edge is real, closing line value over the two weeks of semifinal action will tell the story.
Four teams. One trophy. A 71-year-old tournament that has never delivered quite this much parity at the business end. The AI's verdict is that PSG's depth and current form make them the team to beat, but the margin over Bayern is thinner than the headline favorite status suggests — and Arsenal carrying their unbeaten run to Budapest is exactly the kind of plot-twist narrative that football tends to reward.
As always, the model is a tool — not a crystal ball. Check our daily AI predictions for live match-by-match projections through the semifinals, and see if PSG can become just the second team in 36 years to successfully defend the Champions League crown.
Based on our AI model's 10,000 simulations of the remaining ties, PSG are assigned a 34.1% trophy probability, Bayern Munich 24.7%, Arsenal 21.4%, and Atletico Madrid 19.8%. PSG's higher probability reflects their superior depth, league phase dominance, and recent knockout record. Bayern's elite attacking output (32 goals and assists across 14 matches from their front three) makes them the most dangerous opponent in a single-leg tie.
AI Champions League knockout prediction uses two-leg aggregate simulation. The model calculates expected goals (xG) for each team in each leg based on squad quality, home/away factors, opponent defensive metrics, and historical head-to-head performance. It then simulates goal distributions from those xG estimates, accounts for away goal scenarios and extra time, and runs thousands of iterations to produce win probability distributions for each club across all possible paths.
AI football prediction models perform best in high-data environments like the Champions League, where detailed metrics including xG, pressing intensity, possession sequences, set piece efficiency, and player tracking data are available. Accuracy is highest for predicting two-leg aggregate outcomes (more data points reduce variance) than single matches. Our model's Closing Line Value analysis on Champions League markets shows consistent positive CLV, indicating genuine predictive edge versus bookmaker pricing.
Our model assigns PSG a 34.1% win probability. At current bookmaker odds of approximately +220 (implied probability ~31.3%), there is a 2.8 percentage point positive expected value gap — sufficient to represent a value bet under standard edge-filtering criteria. Bettors should validate this against Closing Line Value analysis through the semifinal legs to confirm whether the market moves toward or away from the model's estimate before committing final positions.