| Passes attempted | Passes completed | Pass difference | Pass completion % |
Swansea | 1809 | 1467 | 342 | 81.09 |
Man. City | 1663 | 1348 | 315 | 81.05 |
Man. United | 1640 | 1329 | 311 | 81.03 |
Chelsea | 1782 | 1444 | 338 | 81.03 |
Arsenal | 1592 | 1273 | 319 | 79.96 |
Wigan | 1218 | 953 | 265 | 78.24 |
Liverpool | 1653 | 1281 | 372 | 77.50 |
Tottenham | 934 | 717 | 217 | 76.77 |
Fulham | 1446 | 1071 | 375 | 74.07 |
West Brom | 1228 | 907 | 221 | 73.86 |
Wolves | 1334 | 965 | 369 | 72.34 |
Aston Villa | 1292 | 920 | 372 | 71.21 |
Everton | 938 | 667 | 271 | 71.11 |
QPR | 1250 | 864 | 386 | 69.12 |
Newcastle | 1132 | 781 | 351 | 68.99 |
Bolton | 1265 | 871 | 394 | 68.85 |
Norwich | 1062 | 730 | 332 | 68.74 |
Sunderland | 1054 | 706 | 348 | 66.98 |
Blackburn | 1159 | 774 | 385 | 66.78 |
Stoke | 1056 | 698 | 358 | 66.10 |
Clearly, congratulations are in order for Swansea City, who have stayed loyal to the true heart of football. Look at United and Chelsea there, separated only by pass difference. And isn't it great to see plucky Wigan Athletic challenging for a place in next season's Pass Completion Champions League?
Data collected from Guardian Chalkboards, but any error is probably mine.
That's brilliant actually. good hints on who to avoid for your Fantasy League midfielders, too.
ReplyDeleteThe metric for separating teams is not really the best. If the percentages are the same then using the "pass difference" will favour the team that has made fewer successful passes when it seems clear that 80% pass completion over 2000 passes is better than 80% over 1000. In this example the teams would have a "pass difference" of 400 and 200 respectively and so the 1000 total pass team would be higher.
ReplyDeleteTechnically this relates to the variance of estimators. The "pass completion %" can be seen as an estimate for the teams actual probability of completing a pass, with the larger sample size giving a more consistent estimator.
I'm sure it's not something that will actually come up that often if you take the percentages to a greater accuracy though.
Thanks for the feedback Steven. I was aware of the limitations of using "pass difference" - for instance, Tottenham and Everton have a markedly better PD than everybody else because they have played one game fewer. The point you raise about number of passes is an excellent one as well.
ReplyDeleteStatistical analysis is not really something I do (obviously the post above is not intended to be taken entirely seriously) but I intend to update this table regularly so I figure it's worth doing it properly. From now on I'll take the completion percentages to a greater degree of accuracy, and if you have any other suggestions about what I could do with the data then I'd be delighted to hear them.