Champions League Matchday 3: Model Estimates for PSG–Barcelona, Stuttgart–Atlético and Como–Manchester United
Matchday 3 of the Champions League league phase is played across Tuesday 20 and Wednesday 21 October, with 18 fixtures in two nights. Yesterday we ran Liverpool–Villarreal, Bayern–Arsenal and Real Madrid–Leipzig through the Dixon-Coles model (ρ = −0.13). This second piece covers the other three heavyweight ties — with strengths estimated from each team's domestic season so far. These are model estimates, not betting tips.
PSG vs Barcelona — Tue 20 Oct, 20:00 UK, Parc des Princes
The tie of the round: the holders against the only side with a perfect record in Europe's top five leagues. PSG sit top of the league phase after a 6–1 thrashing of Slovan Bratislava and a statement 4–2 win at Manchester City, and they have won six in a row in all competitions — including Saturday's 3–1 comeback at Strasbourg, where Luis Enrique rested most of his first-choice side. The cost of that game was Nuno Mendes, who suffered a thigh injury and faces several weeks out.
Barcelona's numbers are remarkable: nine wins from nine in La Liga, 39 goals scored, and wins in every game in all competitions this season — 5–2 at Real Betis on Saturday (Yamal, Rodri, Gabriel Jesus, Eric Garcia and Olmo on the scoresheet), 4–1 at Galatasaray and 5–1 at home to Feyenoord in Europe. But they arrive in Paris with a defensive crisis: Jules Kounde (ankle) and Eric Garcia (thigh, out three to four weeks) both fell in Seville, joining Andreas Christensen on the sidelines. Raphinha could make his first start since 19 September after a hamstring issue. History is dead even: 16 European meetings, six wins each, four draws, 29 goals apiece — though PSG have won the last three meetings in Paris by 10–3 on aggregate.
| Model output | PSG win | Draw | Barcelona win |
|---|---|---|---|
| 1X2 | 13.8% | 13.9% | 72.3% |
| Over/Under 2.5 | 89.4% / 10.6% | — | — |
| BTTS | 79.4% | ||
The model makes Barcelona clear favourites away from home: their attack rates 3.24× the La Liga average against a PSG defence of 0.94 (λH 1.68 vs λA 3.56), and the most likely scorelines are all Barcelona wins — 1–3 (6.7%), 1–4 (6.0%), 1–2 (5.7%). The honest caveat cuts against the model: those defensive absences happened three days ago and sit in none of the season averages. The 0.75 defence rating for Barcelona describes a team with Kounde and Garcia available. Without them, against Dembélé, Doué and Kvaratskhelia in transition, the true number is worse.
VfB Stuttgart vs Atlético Madrid — Tue 20 Oct, 20:00 UK, MHPArena
The table and the form guide point in opposite directions here. Stuttgart sit fourth in the league phase with six points from six — 3–1 at home to Viking, 3–0 at Slovan Bratislava, with Ermedin Demirović joint-top of the scoring charts on four goals. Domestically they are 13th in the Bundesliga with two wins from six, and Saturday's 1–0 defeat at Hamburg was their third loss in four league games. Injuries are piling up: Mittelstädt, Nartey, Al-Dakhil, Tiago Tomas, Zagadou and Diehl are all expected to miss out.
Atlético are the form team in Spain: six straight wins in all competitions, second in La Liga five points behind Barcelona, after Saturday's 1–0 win at Espanyol settled by Alex Baena's 89th-minute strike. In Europe they responded to a 2–1 defeat at Liverpool by beating Manchester United 2–0. Diego Simeone is without Koke (muscle) and Cristian Romero (adductor). The clubs have never met competitively — only a 2018 friendly that finished 1–1 — but Stuttgart's European record against Spanish opposition is poor: no win in their last seven Champions League games against La Liga clubs (D2, L5).
| Model output | Stuttgart win | Draw | Atlético win |
|---|---|---|---|
| 1X2 | 14.3% | 22.5% | 63.2% |
| Over/Under 2.5 | 57.7% / 42.3% | — | — |
| BTTS | 53.7% | ||
The model sides with the form team: Atlético's attack (1.58×) and defence (0.66) against a Stuttgart side scoring 0.82× the Bundesliga rate (λH 0.91 vs λA 2.09). Most likely: 0–2 (10.8%) or 1–1 (10.7%). The one number that flatters Atlético is that Stuttgart's European home record was built against Viking and Slovan Bratislava, the weakest opposition of the matchday.
Como vs Manchester United — Wed 21 Oct, 17:45 UK, Stadio Giuseppe Sinigaglia
A debutant against a giant in crisis. Como are fifth in Serie A (12 scored, 7 conceded in 7 games) and unbeaten in their last seven home games; their European campaign has been pure entertainment — a 4–1 demolition of RB Leipzig followed by a 3–2 defeat at Feyenoord — and Sunday's 3–1 win at Fiorentina kept the momentum going. Cesc Fàbregas has fitness questions over defenders Yan Couto and Trevoh Chalobah, both withdrawn against Fiorentina.
Manchester United arrive with problems at both ends of the pitch. Thirteenth in the Premier League on 7 points from 7 games, winless in six in all competitions after Sunday's 1–1 draw at Leeds. Europe has been their better stage — 4–0 over Sabah, then a 2–0 defeat at Atlético — but Michael Carrick's squad list is stretched thin: Leny Yoro is suspended, Manuel Ugarte, Patrick Dorgu, Benjamin Sesko and Tom Heaton are out, and Lisandro Martinez (thigh), Kobbie Mainoo, Luke Shaw and Ayden Heaven are all doubts. The clubs have never met.
| Model output | Como win | Draw | Man Utd win |
|---|---|---|---|
| 1X2 | 60.3% | 23.2% | 16.6% |
| Over/Under 2.5 | 59.3% / 40.7% | — | — |
| BTTS | 56.8% | ||
This is the model's clearest home lean of the week: Como's attack (1.41× the Serie A average) against a United defence conceding 1.04× the Premier League rate (λH 2.06 vs λA 1.01). Most likely: 1–1 (10.9%), 2–1 (9.9%), 2–0 (9.8%). The draw probability (23.2%) is the highest of the three ties — six winless games have taught the model what United are.
How we computed this
Strengths were estimated from each team's league games this season: goals scored and conceded per game, divided by the league-wide per-game averages from the predictor's presets (EPL 1.53/1.22, La Liga 1.565/1.113, Serie A 1.28/1.15, Bundesliga 1.78/1.46). Ligue 1 has no preset, so we used the average of all five presets as the proxy. For cross-league ties the home/away baselines are the average of the two leagues' presets — a transparent compromise, since the predictor has no Champions League average yet. Those strengths feed expected goals (λH, λA); the Dixon-Coles correction (ρ = −0.13) adjusts the four low scorelines for dependence.
| Fixture | Home attack | Home defence | Away attack | Away defence | λH | λA |
|---|---|---|---|---|---|---|
| PSG–Barcelona | 1.46 | 0.94 | 3.24 | 0.75 | 1.68 | 3.56 |
| Stuttgart–Atlético | 0.82 | 1.03 | 1.58 | 0.66 | 0.91 | 2.09 |
| Como–Man Utd | 1.41 | 0.82 | 1.04 | 1.04 | 2.06 | 1.01 |
Part one of this matchday preview covered Liverpool–Villarreal, Bayern–Arsenal and Real Madrid–Leipzig.
For informational and educational purposes only. Nothing here is betting advice.