Manchester United played the 2025-26 season under two different managers. Rúben Amorim took charge from the start, before being replaced by Michael Carrick, who saw the season out. With both managers handling roughly 20 matches each, there’s enough data to make a proper comparison.
This piece uses Welch’s t-test and Cohen’s d to identify which metrics were meaningfully different between the two regimes. The goal isn’t just to compare averages — it’s to check whether those differences are statistically reliable. Welch’s t-test tells us whether the gap between two groups is likely real or just noise, while Cohen’s d measures how large that difference actually is in practical terms.
[ how we’re reading data ]
Rather than drawing a single cutoff line, this analysis uses a tiered approach based on the combination of p-value and effect size.
- Core difference: p < 0.05 and Cohen’s d ≥ 0.8.
- Statistically solid and practically significant. These are the metrics where the two managers diverged most fundamentally.
- Moderate difference: p < 0.10 or Cohen’s d ≥ 0.5.
- Less certainty, but a real enough gap to be tactically relevant.
- Tendency: Cohen’s d ≥ 0.5 but a higher p-value.
- Directional, but not something to lean on too heavily. Worth noting, not concluding from.
[ core differences ]
These are the metrics where the gap between Amorim and Carrick was both statistically clear and practically meaningful.
( Duels )
- The most consistent gap across the entire dataset sits in the duel metrics. Under Amorim, United averaged 107.15 total duels per game. Under Carrick, that dropped to 90 — a difference of more than 17. That gap runs through every subcategory: ground duels (69.55 vs 63.82), aerial duels (37.60 vs 26.18), and duels won (53.35 vs 47.71). Amorim’s United fought more, won more duels in absolute terms, and did so across both ground and aerial situations.
- The metrics that flip are duel win rate (49.8% vs 53.0%) and aerial duel win rate(47.1% vs 54.2%), which favoured Carrick. Amorim’s side fought more often; Carrick’s side fought more efficiently. That contrast turns out to be more important than it first looks — something we’ll get into properly in Part 2.
( Pressing )
- PPDA — which measures pressing intensity, with lower values indicating more aggressive pressing — was 9.04 under Amorim and 11.65 under Carrick. That’s a significant gap. Opponent build-up success rate backed this up: 77.6% under Amorim compared to 84.1% under Carrick, meaning United allowed opponents to build out of the back far less frequently when Amorim was in charge.
- The direct output of that pressing showed up in Opponent Danger Possession Lost: 24.3 per game under Amorim versus 18.94 under Carrick. Amorim’s press was forcing turnovers in dangerous areas at a noticeably higher rate.
( finishing efficiency )
- Big chances to goal (%) was the single biggest numerical gap in the entire dataset: 50.7% under Carrick versus 29.1% under Amorim — a difference of 21.6 percentage points. On the surface, this looks like a dramatic statement about Carrick’s attacking quality. But taken at face value, it’s also one of the most misleading numbers in the dataset. The full picture requires understanding what was happening around that metric, which is what Part 2 is for.
( defensive actions )
- Blocked opponent shots averaged 4.41 per game under Carrick against 2.95 under Amorim. Goalkeeper punches told a similar story: 0.71 per game under Carrick compared to just 0.15 under Amorim. Both metrics point in the same direction — under Carrick, United’s defensive work was happening deeper and closer to their own goal.
[ moderate differences ]
- Shots off target were higher under Amorim (6.15 vs 4.71), consistent with a side that attempted more overall. Long balls were also more frequent under Amorim (53.05 vs 46.82 per game) — a statistically significant gap that becomes important context when reading the pressing numbers. Fouls won in the final third favoured Amorim (2.25 vs 1.41), as did ground duels in absolute terms (69.55 vs 63.82) and aerial duels won (17.9 vs 14.18).
[ tendencies ]
- Interceptions were higher under Carrick (9.53 vs 7.85). Pass accuracy also favoured Carrick (83.3% vs 81.2%). The defensive line sat higher under Amorim (45.72m vs 43.14m). Opponent high turnovers — measuring how often the opposition lost the ball in advanced areas — were more frequent under Carrick (5.06 vs 3.70), which is a number that sits in interesting tension with the PPDA data. And goal difference per game was noticeably better under Carrick (0.88 vs 0.20).
[ What This Tells Us — and What It Doesn’t ]
- The numbers draw a fairly clear picture of two different styles. Amorim‘s United pressed harder, fought more duels, conceded possession less freely, and took more shots. Carrick‘s United played with better pass accuracy, defended deeper, and converted big chances at a much higher rate.
- But reading these numbers in isolation creates as many questions as it answers. Why did Carrick’s side convert big chances at nearly twice the rate? How does a manager who pressed less aggressively end up with more opponent turnovers? And what does it actually mean that duels were so much higher under Amorim?
- These aren’t quirks or noise in the data. They’re the fingerprints of two genuinely different tactical systems operating through the same squad. Part 2 gets into the correlation analysis that starts to explain why these numbers look the way they do.
| Core Differences | Amorim | Carrick | Diff. |
| Big Chances to Goal (%) | 29.074% | 50.744% | 21.670% |
| Throw-Ins | 19.95 | 14.06 | (5.89) |
| Total Duels | 107.15 | 90 | (17.15) |
| Duels won | 53.35 | 47.71 | (5.64) |
| Duels won(%) | 49.809% | 52.994% | 3.185% |
| Aerial Duels | 37.6 | 26.18 | (11.42) |
| Aerial Duels won (%) | 47.073% | 54.218% | 7.145% |
| PPDA | 9.04 | 11.65 | 2.61 |
| Opponent Build-up (%) | 77.6% | 84.1% | 6.5% |
| Opponent Danger Possession Lost | 24.3 | 18.94 | (5.36) |
| Blocked Opp. Shots | 2.95 | 4.41 | 1.46 |
| Yellow Cards | 1.45 | 2.06 | 0.61 |
| Punches | 0.15 | 0.71 | 0.56 |
| Moderate Differences | Amorim | Carrick | Diff. |
| Shots off Target | 6.15 | 4.71 | (1.44) |
| Long Balls | 53.05 | 46.82 | (6.23) |
| Fouled in Final Third | 2.25 | 1.41 | (0.84) |
| Ground Duels | 69.55 | 63.82 | (5.73) |
| Aerial Duels won | 17.9 | 14.18 | (3.72) |
| Tendencies | Amorim | Carrick | Diff. |
| Accurate Passes (%) | 81.184% | 83.3% | 2.116% |
| Goal Difference | 0.2 | 0.88 | 0.68 |
| Def. Line (m) | 45.72 | 43.14 | (2.58) |
| Interceptions | 7.85 | 9.53 | 1.68 |
| Error Leading to Shot | 0.45 | 0.82 | 0.37 |
| Opponent High Turnovers | 3.7 | 5.06 | 1.36 |
| High Claims | 0.7 | 1.12 | 0.42 |
Raw per-match data was collected from Sofascore and MARKSTATS.
Aggregate metrics — including per-game averages, cumulative totals, and derived indicators such as Accurate Passes (%) and Total Duels — were constructed independently by the author.
All statistical analysis, including Welch’s t-test and Pearson correlation, was conducted by the author using this dataset.
Leave a Reply