Today
All articles

Arsenal 21% vs Bayern 19.9%: Champions League Probabilities for Bettors

7 Sept 2026·11 min read

Champions League probability title card

Arsenal sits atop most 2026-27 title models at around one-fifth, with Bavaria Munich close behind near that level and Manchester City trailing in the low double digits. Those numbers come from Monte Carlo simulations run by outlets like Opta, not gut instinct, and bookmaker-implied odds don’t always match them. The gap between model output and market price is exactly where informed fans and bettors find value, and it’s worth understanding before you trust either number.

TL;DR:

  • Simulation models like Opta and Predixsport predict Arsenal as having roughly a 21-27% chance to win the Champions League, with Bayern close behind at around 20-25%.
  • Different models use varying assumptions such as recent form versus squad quality, which causes their probabilities to drift and should be considered when interpreting the results.
  • Knockout stage predictions are less precise due to small sample sizes, with about 65-70% accuracy in two-leg xG results, meaning underdog teams can still advance unexpectedly.
  • Bookmaker probabilities incorporate a margin that inflates total implied probabilities beyond 100%, so bettors should compare model estimates to market odds and identify significant gaps.
  • Live, continuously updated prediction tools like BetsyScore reflect real-time events, injuries, and line-up changes, providing more reliable guidance during the tournament.
Betsyscore
Track Probabilities As Matches Change
Compare live win probabilities with lineups, momentum, and instant match statistics across the Champions League and more.
Explore BetsyScore

Table of Contents

What Are the Current Champions League Win Probabilities?

The clearest way to read Champions League win probability right now is to place the major simulation outputs side by side. Opta’s supercomputer, which runs the tournament forward 10,000 times to generate frequency-based outcomes, currently has Arsenal winning the title in about 21.2% of simulated seasons, with Bayern Munich at 19.9%, Manchester City around 12%, and Paris Saint-Germain near 7.7%.

Champions League title probability comparison

Predixsport, which builds its model on an Elo rating system layered with expected-goals sampling, produces a different flavor of the same story. Its 10,000-simulation run has put Arsenal as high as 27%, Bayern at 25%, and PSG near 20% in one recent output. Neither model is “correct.” They’re built on different assumptions about how much recent form should outweigh long-term squad quality, and that’s why the numbers drift.

Here’s a snapshot of where things stand across the two major simulators:

  • Arsenal: around one-fifth (Opta) vs. somewhat higher (Predixsport)
  • Bayern Munich: close to one-fifth (Opta) vs. somewhat higher (Predixsport)
  • Manchester City: low double-digit chance (Opta)
  • Paris Saint-Germain: less likely according to Opta, higher in Predixsport

Bookmaker-implied probabilities move faster than either model, often shifting within hours of a squad announcement or a single result. Treat any snapshot like this one as a moment in time, not a fixed forecast. For live-updating numbers rather than a dated table, check BetsyScore’s Champions League predictions, which refresh as matches are played.

How Do Prediction Models Calculate Win Probability?

Every credible Champions League probability model runs on some version of the same core idea: simulate the tournament thousands of times and count how often each club comes out on top. The details of how that simulation works vary quite a bit between providers.

  1. Monte Carlo simulation. The model plays out the remaining fixtures, knockout draws, and final repeatedly, typically 10,000 times, using randomized but statistically weighted results for each match. The final win percentage is simply how often a club lifted the trophy across those simulated seasons.
  2. Elo and rating-based systems. These assign each team a numerical strength rating updated after every match, and matchups are resolved based on the rating gap. They’re computationally lighter but less sensitive to attacking or defensive quality shifts within a squad.
  3. Goals-sampling models. These simulate scorelines directly from each team’s expected-goals profile, capturing finishing quality and defensive solidity more precisely than a single rating number can.

Most systems blend recent form, head-to-head history, expected goals, injuries, and fixture congestion into the inputs. It means that across 10,000 simulated versions of the season, Arsenal lifted the trophy in about 2,100 of them.

Pro Tip: When two models disagree sharply on the same club, check whether one leans on Elo ratings and the other on expected-goals sampling. The disagreement usually says more about model design than about the team itself.

Why Do xG and Knockout Variance Shift the Numbers?

Expected goals, or xG, estimates the probability that any given shot results in a goal based on distance, angle, and the buildup that created it. Non-penalty xG (npxG) strips out penalty kicks to isolate open-play quality. Across a full season, xG tends to predict a team’s true attacking level better than the actual goals column, which is why simulators lean on it so heavily.

Illustrated football xG shot probability

Knockout football complicates that signal. A two-legged tie is only two matches, and two matches is a tiny sample. Combined two-leg xG predicts the actual round outcome with roughly 65 to 70% accuracy, which sounds solid until you remember that leaves a real chance the “better” team on paper simply loses.

That gap matters in practice:

  • A team can win a two-legged tie 2-1 on aggregate while being outplayed 3.4 to 1.1 on combined xG.
  • League-phase data has already shown clubs like Liverpool underperforming their underlying numbers in ways that shift how models weight them going forward.

Statistic Callout: Two-leg xG calls the correct advancing side roughly two times out of three. The other third of the time, the scoreline and the underlying performance simply disagree, and that’s usually where a model’s probability and the final result diverge hardest.

How Should Bettors Use Win Probability Numbers?

A model probability and a bookmaker’s implied probability are not the same thing, and confusing them is the single most common mistake casual bettors make. Sportsbooks bake in a margin, sometimes called the overround, so implied probabilities across all outcomes typically add up to more than 100%. That margin is the house’s built-in edge, and it needs to be stripped out before comparing a market price to a model’s output.

  1. Convert the odds. Divide 1 by the decimal odds to get raw implied probability, then adjust for the overround if you’re comparing across multiple outcomes.
  2. Look for the gap. If a model gives a club a 25% title chance but the market implies 18%, that gap is where potential value lives.
  3. Update after results. Re-check probabilities after every matchday rather than clinging to a preseason number that’s gone stale.
  4. Size stakes conservatively. No single match should swing your bankroll, since knockout variance means even a strong favorite loses regularly.

Pro Tip: Never treat one bad result as proof the model was wrong. A 21% chance means the club fails to win 79% of the time; that’s the model working as designed, not failing. For a deeper look at where fans and bettors go wrong, BetsyScore’s guide to common prediction errors is worth a read.

What Data Powers BetsyScore’s Champions League Predictions?

BetsyScore builds its match predictions from the same category of inputs the major simulators use: expected goals, recent form, and head-to-head history, refreshed continuously rather than published once and left to age. Live scores update every few seconds, and a minute-by-minute momentum read shows which side is controlling a match as it happens, not just after the final whistle.

Coverage runs across the Champions League and more than 200 other competitions worldwide, with detailed player profiles, tournament leaderboards, and instant stats layered on top of the win-probability outputs. The result is a prediction feed built for readers who want the number and the context behind it, not a static forecast frozen at kickoff. BetsyScore’s own explainer on how its AI predicts Champions League matches breaks down the modeling approach in more detail.

How I Read the Numbers Before a Knockout Tie

Trust the model until a lineup sheet contradicts it. If a starting striker is suspended or a manager gets sacked midweek, the simulation hasn’t caught up yet, and your judgment has to fill that gap. I watched a heavy favorite’s implied probability collapse within a single leg last season after an early red card, which is a reminder that no percentage survives contact with 90 minutes of actual football untouched. Check BetsyScore’s live predictions for the updated read once the whistle blows.

— Aria

Where to Verify These Probabilities Yourself

Model outputs are only as trustworthy as your ability to check them, so bookmark a few primary sources rather than relying on secondhand summaries.

  • Opta’s supercomputer projections run 10,000 season simulations and publish updated per-club title percentages, giving a rating-independent baseline.
  • Predixsport’s title-odds page shows the Elo-plus-goals-sampling alternative, useful for spotting where models disagree.
  • UEFA’s official result simulator lets you test your own scenarios interactively, matchday by matchday.
  • BetsyScore’s Champions League performance data guide explains the xG and expected-points metrics feeding into most modern models.

Cross-checking two or three of these takes a few minutes and will tell you more than any single headline number.

Get Live Champions League Predictions on BetsyScore

Static probability tables go stale the moment a match kicks off. Some platforms provide win-probability percentages that update continuously, drawing on expected goals, current form, and head-to-head data rather than a number frozen after last week’s simulation.

Betsyscore

That live-update model fits exactly the use case this article has walked through. Instead of manually recalculating after every knockout surprise or injury update, you get a momentum read and refreshed probabilities as the match unfolds, alongside live scores across the Champions League and more than 200 other competitions. Head to BetsyScore’s AI predictions and pull up the next Champions League fixture on your watchlist to see how the model’s current number compares to what the bookmakers are offering.

Sources

FAQ

Who’s Most Likely to Win the Champions League?

Arsenal and Bayern Munich currently lead most simulation models, with Opta’s supercomputer placing Arsenal at roughly 21.2% and Bayern at 19.9% for the 2026-27 title.

Does Real Madrid or Another Club Have a Higher Chance of Winning?

Model outputs vary by provider, and Real Madrid’s exact figure shifts depending on which simulation you check, but as of the latest runs, Arsenal and Bayern Munich rank ahead of most other single clubs in win probability.

How Accurate Are Champions League Prediction Models?

Combined two-leg expected goals predicts the correct knockout-round outcome about 65 to 70% of the time, which shows models capture real signal but still leave meaningful room for upsets.

Why Do Model Probabilities Differ From Bookmaker Odds?

Bookmaker odds include a built-in margin, so implied probabilities across all outcomes add up to more than 100%, while simulation models report a cleaner frequency-based estimate without that markup.

How Often Should I Check Updated Win Probabilities?

Check after every matchday at minimum, and especially after injuries, suspensions, or managerial changes, since those events shift probabilities faster than the model’s normal weekly cycle. Live feeds like BetsyScore’s predictions page update continuously rather than on a fixed schedule.

Arsenal 21% vs Bayern 19.9%: Champions League Probabilities for Bettors