David de Gea’s goals prevented at home and away
, ,
Against his share of what his opponents usually allowed goalkeepers, David de Gea was 0.13 goals under at home and 0.22 goals under away in 2026-27, 5 games against 4. The plain averages are -0.13 and -0.22, a difference of 0.09.
4 games away is still too small a sample for the statistics to say much, but maybe you can find something if you squint.
goals prevented a game over expected
at home
−0.13
away
−0.22
-101
A +0.09 difference in goals prevented over expected at home vs away is about what you’d expect.
plain
-0.13
-0.22
-101
A +0.09 difference in goals prevented at home vs away is about what you’d expect.
The games
at homeaway
| Side | Date | Game | Result | Prevented | Expected |
|---|---|---|---|---|---|
| Aug 24, 2026 | @ AS Roma | L | -0.49 | 0.00 | |
| Aug 29, 2026 | vs Frosinone | L | -0.26 | 0.00 | |
| Sep 5, 2026 | vs Torino | L | -0.52 | 0.00 | |
| Sep 11, 2026 | @ Venezia | W | -1.12 | 0.00 | |
| Sep 20, 2026 | vs Napoli | T | 1.77 | 0.00 | |
| Oct 10, 2026 | @ Genoa | L | -0.46 | 0.00 | |
| Oct 18, 2026 | vs Como | L | -1.20 | 0.00 | |
| Oct 25, 2026 | @ Internazionale | L | 1.19 | 0.00 | |
| Oct 29, 2026 | vs Atalanta | T | -0.43 | 0.00 |
Method
- Expected goals are this site’s model of each shot’s chance, from where it was taken, whether it was a header, and how it was set up (a cross, a through ball, a headed pass, or no pass at all), fit on the 105,024 non-penalty shots of the five leagues and the Champions League since 2024-25 and within 3% of the goals scored in each season. Penalties are left out.
- Expected is what the opponent allowed goalkeepers in its other 2026-27 games, plus 0.00 for the home team (half the league’s home edge for goalkeepers that season), nothing at a neutral site, scaled to his share of his team’s goalkeepers’ goals prevented in his games, 0%.
- Each expected number is then fit to the season’s games: a game bears out about -1% of the difference the model predicts.
- The shuffle keeps the season’s shape: the games are cut into stretches, one a 10 games, where the dates leave the least variance, each game is read against its stretch, and the sides are shuffled only inside each stretch.