Save Percentage Analysis - Part I

Mignolet's Slump-
Making Lemonade From Lemons

(Mignolade?)

In the never-ending search for intersections of useful data and applications to improve the game, some recent events have prompted the thoughts below.

I was reading Whoscored's nice piece on Mignolet's troubles as late which used this graphic (left) to show his form over the course of this season.

TOP TEN in Save Percentage BPL 2012-13
Image taken from Barriesview.com

28 Goals Against...1.27 GAA and 3 errors leading to goals in the past five games.  Not terribly impressive numbers and worrying for any Liverpool fan.  However, if we dig a little deeper we see that his save percentage is on track for his fine run of form over last season (which saw him finish in the top ten in save percentage).  So what gives?  We have to accept that even good GKs have stretches where they hurt their own cause (see De Gea/Hart the past couple years) but they typically bounce back. Mignolet ought to follow that pattern. This may all be a tempest in a teacup; Mignolet is legit, and so long as his coaches support him and he has good resilience, he'll be back at the top of the current class of GKs in short order.

So what's the connection between Mignolet and lousy offenses in NCAA Division III?  In looking at his stats, something jumped out at me.  For being such a talented guy, Mignolet's save percentage seemed low at 73%...which got me to looking into other BPL GKs numbers which showed me that save % in the EPL is surprising low across the board.  And then, inevitably, I had to ask what the numbers would show at the level at which I work in Division III.
Start by looking at the top ten BPL GKs by save percentage above again.  They all seemed low, particularly when only considering the players who play the most -Mucha of Everton saved 92% of the shots on target...but as any reader of this blog knows, he was buried by Tim Howard.  Ditto for Tremmel; he's played a lot this season for the Swans, but spent last year in Michel Vorm's immense shadow.  '13-'14 would have been a repeat but Vorm has been hurt this season.  So it's really the #2-8 spots we're interested in, and those are surprisingly low save percentages. One in three shots on frame beats a BPL GK!


Now, take the two NSCAA All-American GKs from NCAA Division III this past season (left). Malick Faye and Thomas Bull of SUNY IT and Amherst, respectively.  Note the save percentages. Way, way above everybody's mark from the BPL.  It's not bad when a kid is more than 5 percentage points better than Petr Cech (Bull) ... or a slightly better 13 + (Faye)!

For sure, Cech's opponents are a bit better shooters...but he's at least equally more competent than these two fellas.  No, the main concern that comes out of these numbers is the sloppy nature of college level attacking.  The utter lack of confidence in attacking play, the inability to play possession in the final third to any degree, and therefore create chances that actually have a likelihood of going in.

Evidently, there are 140 GKs who are better shot-stoppers
than Petr Cech in NCAA Division III!  Great news...

To the left is the point in NCAA DIII where we see save percentages start to match Cech's 77% mark...Additionally, there are 86 GKs who posted 80% or better in 2013! Since you asked, there are 71 Division I GKs who posted 78% or better save percentage this fall...

2012-13 BPL...Everton and Arsenal skew the curve, 
along with Tremmel.  Starters don't break triple digits.
Faye and Bull are doubling up Joe Hart - huh?!

But look at Minutes per Goal Conceded (right; courtesy of barriesview.com) and compare to the All-Americans' numbers.  Let's assume for a second that like the building trade, demolition is far faster/easier than the actual building; defending is easier to master than attacking.  No argument there.

By comparison, we might take the ACC, one of the toughest leagues in men's soccer and the league with the most GK minutes played in 2013 (15 schools & 3 made it to the final four). The GKs in that league played more than 22,930 minutes & conceded 373 goals. That's a minutes per goal ratio of 61 minutes and some change played per goal conceded.  Uh-Oh.  The PAC-12 (6 schools) GKs played the least soccer of all this fall (11,464 minutes) and conceded 203 goals: 56 minutes played per goal conceded.  Gulp.

The ACC featured a shot every 7 minutes & a goal every 61 minutes.  Division III saw a shot every 6.5 minutes & a goal every 54 minutes.  The EPL 2012-13 saw a shot every 6.7 minutes & a goal every 66 minutes.

Put it all together: EPL GKs get beaten at a considerably higher percentage but NCAA GKs give up goals ~1.5 times as frequently.  Pros: fewer shots & fewer saves.  NCAA: more shots & more saves.  Seems intuitive.  It's interesting to note that the Division I save percentage is about 1/2 way between the DIII and EPL numbers.  Half the number of GKs above the pro save percentage in DI as in DIII...seems to indicate a little bit of the degree to which DI is the "next level."

If we concede that amateur players can defend better than they can attack, we might not be surprised to see such remarkable minutes-per-goal stats.  But then again, given that the defenders at the BPL level will be that much more athletic, skilled and tactically astute, what possibly explains the massive difference in save percentage?

Part II Coming Up...