Twitter

Tuesday, 23 June 2015

EPL 2014-15 Assists vs Key Passes p90

2014-15 Assists p90 vs Key Passes p90 (min of 750 minutes played)
R-squared = .505

Cesc, Di Maria and Leandro Bacuna?

Small samples can throw up some funny numbers. Bacuna finishing 3rd in Assists p90 is a surprise and we can safely assume this is unlikely to repeat itself. But what of Cesc and Di Maria? These numbers obviously wreak of over-performance but they still are 6th and 7th, respectively, in Key Passes p90 (see table below) while Bacuna is down in 63rd. 

The league average 'Assist per Key Pass %', in 2014-15, was 8.9%:

Leadro Bacuna: 29.4%
Cesc Fabregas: 18.9%
Angel Di Maria: 19.6%

Is it fair to expect (major?) regression** in terms of assists next season, for these 3 players?


Anyway, the Tableau is interactive and the table is sortable, so play around!

Cheers,
Clarke


** Here is a piece by Alex Oshansky (@atosoccer) on the repeatability of Key Passes: http://statsbomb.com/2013/09/shots-and-key-passes-are-better-than-goals-and-assists/

 
Top 50 Key Passers p90 >750mins played
Assist per KP% (League Avg)  = 0.089
Player Name A p90 KP p90 Assist per KP
Mesut Özil 0.242 3.389 0.071
Jesus Navas 0.320 3.163 0.101
David Silva 0.237 3.148 0.075
Dusan Tadic 0.306 2.971 0.103
Samir Nasri 0.342 2.966 0.115
Cesc Fabregas 0.561 2.963 0.189
Angel Di María 0.548 2.795 0.196
Marc Albrighton 0.263 2.717 0.097
Eden Hazard 0.240 2.696 0.089
James Milner 0.362 2.687 0.135
Bobby Zamora 0.202 2.475 0.082
Alexis Sanchez 0.244 2.472 0.099
Gylfi Sigurdsson 0.334 2.436 0.137
Christian Eriksen 0.057 2.408 0.024
Erik Lamela 0.274 2.349 0.117
Santiago Cazorla 0.331 2.346 0.141
Stewart Downing 0.217 2.309 0.094
Leighton Baines 0.292 2.271 0.129
Raheem Sterling 0.207 2.213 0.093
Joey Barton 0.115 2.144 0.054
Stephen Ireland 0.289 2.122 0.136
Alex Oxlade-Chamberlain 0.060 2.113 0.029
Aaron Ramsey 0.269 2.106 0.128
Steven Gerrard 0.040 2.093 0.019
Jason Puncheon 0.223 2.071 0.108
Gastón Ramírez 0.179 2.054 0.087
James Ward-Prowse 0.365 2.009 0.182
Juan Mata 0.156 1.990 0.078
Jack Grealish 0.110 1.988 0.056
Victor Moses 0.186 1.979 0.094
Marko Arnautovic 0.235 1.971 0.119
Charlie Adam 0.060 1.969 0.030
Bojan 0.074 1.935 0.038
Lazar Markovic 0.097 1.931 0.050
Riyad Mahrez 0.130 1.902 0.068
Philippe Coutinho 0.161 1.897 0.085
Jordan Henderson 0.249 1.853 0.134
Leroy Fer 0.076 1.827 0.042
Willian 0.112 1.826 0.061
Robbie Brady 0.103 1.795 0.057
Matthew James 0.048 1.762 0.027
Remy Cabella 0.092 1.757 0.053
Mark Noble 0.117 1.756 0.067
Mousa Dembélé 0.079 1.740 0.045
Diego Costa 0.130 1.729 0.075
Aiden McGeady 0.203 1.725 0.118
Stevan Jovetic 0.343 1.715 0.200
Sebastian Larsson 0.088 1.706 0.052
Sadio Mané 0.126 1.686 0.075
Kieran Trippier 0.105 1.686 0.063

Save % - A Reminder that Goal Keepers are High Risk

Twitter was buzzing with Arsenal signing Petr Cech, for £11m. An amazing sum for a 33year old GK. Plenty of work has been done by others in this field showing that their is next to nothing repeatable about the GK position. This lack of ability to narrow down any sort of skill, should make clubs afraid to spend, even a small amount of money, on this position.

Many have praised Arsenal for getting a proven and experienced GK who makes Arsenal title contenders:


Then there are others like @footballfactman, (probably the best GK analyst in the blogosphere)
who is less excited:

My point here is not about Cech but GK's in general. They are voodoo.

I have ran a few scatter plots to show the linear regression of team save% over various lengths of time and none show any signs that a team or keeper is worth 'betting' on both in the short term or long term, at least when looking at saves per shot on target against (save%).




These results will be nothing new to most of you but reiterates that teams and GK's lack a lot of control over goals scored against them. Analytically, it is pretty tough to endorse keeper x, y or z. What we see one year or even for 2 yrs in a row (unlikely) can change very quickly.

Arsenal are not a poor club and can absorb this move without missing much of a beat, financially, but based on linear regression the best we can predict is that Arsenal will have a save% close to league average, in the 2015-16 season (like every other team).

Cheers,
Clarke