PSG v Barça: our two models don’t pick the same favourite
Six points each, four away goals each on the last matchday, and two opposite readings. Dixon-Coles sees a slight edge for Barça; our player model makes PSG favourites by ten points. What each model looks at, and why they don’t see the same game.
By Lucas da Silva · 20 October 2026 · 6 min read · Models by Rondo Research
Paris Saint Germain won 4-2 at Manchester City, and 
FC Barcelona won 4-1 at Galatasaray. Both clubs have 6 points from 2 games, and they meet tonight at the Parc des Princes.
There are two ways to look at this game, and our two models have each taken one. The first sees a slight edge for Barça. The second makes PSG favourites by ten points.
Two models, two favourites
Dixon-Coles, our results model, calculated on 19 October, gives Barça 41.0%, PSG 38.4% and the draw 20.6%. Close to a coin toss, with a very slight Catalan lean, and a most likely score of 1-2.
Our player model, calculated this morning, flips the order: 44.6% for PSG, 34.1% for Barça, 21.3% for the draw. Its most likely score is 2-1.
| FC Barcelona | 41.0% |
| Paris Saint Germain | 38.4% |
| Paris Saint Germain | 44.6% |
| FC Barcelona | 34.1% |
On one point, though, they agree: this should be an open game. Over 2.5 goals stands at 74.1% for Dixon-Coles and 76.0% for the player model. Nobody should be expecting a 0-0.
Why they don’t see the same PSG
The gap doesn’t come from tonight’s data but from what each model judges.
Dixon-Coles judges a club on its results, goals scored and conceded, measured in its own league and then brought to a European scale by a level coefficient shared by the whole league. PSG are judged as a Ligue 1 club, Barça as a LaLiga club. It expects 1.90 goals from PSG and 1.97 from Barça: two very similar attacks, and home advantage isn’t enough to tip the balance.
The player model judges the eleven who should be on the pitch. Each player is measured on his own output, once the level of his teammates and of his opponents has been stripped out, which makes him comparable from one league to another. It expects 2.13 goals from PSG and 1.86 from Barça: with squads this close, home advantage makes the difference.
This isn’t a one-off disagreement. For the title, Dixon-Coles gives PSG 3.4% (7th of 36) and Barça 12.4% (4th). The player model puts them almost side by side, and it is on PSG that everything turns: from 3.4% to 12.5%.
| FC Barcelona | 12.4% |
| Paris Saint Germain | 3.4% |
| FC Barcelona | 16.2% |
| Paris Saint Germain | 12.5% |
These are probabilities of lifting the trophy. For the league phase, PSG have a 79.9% chance of finishing in the top eight according to the player model, and 55.4% according to Dixon-Coles.
What one absence costs, in this specific game
Neither club arrives at full strength. Nuno Mendes has had a dead leg since 18 October. On the Catalan side, Koundé (ankle), Eric García (hamstring) and Christensen are listed as injured.
We replayed the fixture twice, with and without Nuno Mendes, Lucas Hernández taking his place: same opponent, same venue, only the player changes. On PSG’s usual line-up, their win probability drops from 48.8% to 45.1%.
The figure only holds for tonight: against a weaker opponent the same absence would cost less, because a gap in probability shrinks when the outcome is already likely. And we don’t know whether Nuno Mendes will play: the club gives no return date.
What would prove the model wrong
The player model relies on a projected line-up, not the official team sheet, which comes out an hour before kick-off. It starts from the last eleven fielded in the Champions League and replaces each declared absentee with his backup in the same position. A tactical surprise, a player listed as injured who plays, or the reverse, and its figure moves.
Barça’s defence is their weak spot: three injured defenders, and it is also the line where the choice of replacement matters most, while the injury feed doesn’t tell us everything the coach will decide. Dixon-Coles knows nothing about line-ups: it sees a Barça side that has scored a lot in LaLiga, with or without its defenders.
Calculations of 19 and 20 October 2026: in the same game, one model judges results, the other judges players, and PSG is where they disagree most. The Parc des Princes will tell which of the two read this PSG better.
Frequently asked questions
Who is the favourite for PSG v Barcelona?
It depends on the model. Dixon-Coles, calculated on 19 October 2026, gives Barça 41.0%, PSG 38.4% and the draw 20.6%. Our player model, calculated on 20 October on the probable line-ups, gives PSG 44.6%, Barça 34.1% and the draw 21.3%.
What is the most likely score of PSG v Barcelona?
1-2 for Dixon-Coles, at 7.7%, and 2-1 for the player model, at 7.8%. Either way, more than nine times out of ten the final score will be another one: it is the most likely score, not a prediction.
What does Nuno Mendes’s absence cost PSG?
In this specific game, with Lucas Hernández in his place, PSG’s win probability drops from 48.8% to 45.1% on the usual line-up, i.e. 3.7 percentage points. The figure is specific to this opponent and this venue.
The figures in this piece are read from the latest published computation: they update with the model and may therefore differ from those of the publication date. That is deliberate — an article should not go on asserting what is no longer true.
