Save % Analysis - Part II

Mignolade, part II

Stinkin' Lousy Players...

Part I of this mini-series illustrated some remarkable differences in goalkeeping stats between EPL GKs and NCAA GKs.  In men's NCAA DIII we saw that 141 GKs posted save percentages higher than Petr Cech, David De Gea, Joe Hart, Simon Mignolet, Tim Howard, Brad Guzan and basically every other starting GK in the EPL in 2012-13.  Something doesn't mesh there.  Here we'll try to parse out why college GKs appear to have greater success relative to their pro counterparts, and what that says about the other 20 players on the field.

To bring this down to the youth level, this post tracked the number of shots and passes attempted by my U13 boys team...37 shots led to 5 goals.  This was in a mere 40 minutes of indoor play, but all the same, at that rate we'd have taken just over 83 shots in a 90 minute game!  Without getting into shot conversion rates (and that was pretty interesting stuff- read the post for more) the main theory is that young kids play in way too big a rush.  Which is mostly OK; the problem is we don't see a lot of college teams behaving that differently from my U13s.

EPL teams put 1/3 of their shots on target- 1,961 on goal out of 5,891 attempts (33%) through 22 matches this year.  10% of all shots end up in goal in the EPL (a number proven steady over many years, and first noted by one of soccer's earliest and most notorious statisticians, Charles Reep).  Interestingly, as you'll see below, college teams at all levels and both genders convert about 10% of all attempts at goal, too.



Consider SUNYIT again.  SUNYIT put nearly 50% of their shots on target in 2013, which is a remarkable difference from the EPL clubs (SUNIT's opponents shared a 69% save percentage...).  But the data bear this out as typical, not remarkable at all as compared to other DIII schools.  Here are some (rough) numbers for the season past :

• NCAA DIII men averaged 45% of all shots on target - 12% of all shots are goals... 
• In men's Division I, 41% of all shots taken were on target - 10% of all shots are goals.   
• In women's DIII 52% of all shots were on target - 12.5% of all shots are goals, &: 
• In women's DI 46% of all shots were on target - 10% of all shots are goals.

Teams with a lot of wins and goals will probably trend toward SUNYIT's numbers, but with everyone so far above the EPL average, it looks fairly normative to have 50% of all shots on target.  It is so hard to get a decent shot off at the pro level and even if a player gets a shot off, many good looks are blocked by committed defenders.

760 games of EPL shots, color coded by shot
conversion percentage.  Red is 50%...nowhere
other than inside 6 yards are shots converted
at or near 50%...

College players by contrast do well getting chances on target, but they are such weak attempts the GKs save the vast majority- they settle for any look they can get.  The goal is freaking huge, so it's not that unreasonable to think that a decent top-level amateur can at least hit 192 square feet of surface area every other attempt.  But with anything approaching an athlete defending that space, well, the effective scoring area is reduced considerably.  EPL athletes truly try to hit specific spots, trying to hit the wide areas of the goal. College shooters are perhaps trying to do the same, but their misses stay on target more often than the pros. This indicates less confident shooting (shooting not to miss, rather than to score) and so we see the amateurs hitting the highly unlikely portion of the goal easily covered by the GK.  Higher save percentage explained...sort of.

This conclusion would simply be corrected if players worked on their shooting more.  And that's certainly part of it.  Individual technical ability is always in need of refinement.  No player is ever done improving.

If we go a little further in the investigation, however, what is really of concern here is a systemic inefficiency in college-level soccer.  The fact that at the macro level we see minimally 41% of all shots on target would lead us to expect a correlating higher percentage of shots to be goals; well over 10% of all shots to be goals...but that's not what happens.

760 games of EPL shots by location (inc. goals, saves,
blocked attempts and misses) Clearly EPL shooters
are trying to get to that crucial dark red area; but per
the previous graphic, we see that even from the red area
there's a less than 50% likelihood of scoring. 

There's a structural problem here, and that's an issue that falls at the feet of the leaders of the various clubs. The question is, do college players take the same quality of chance as the pros or do the college players make any view to goal into an attempt?

If we accept second premise (and while I can't, yet, make an empirical argument on this, it feels very, very true), then the player's ability may well not explain away a 10 percentage point difference in terms of converting shots on target.  An inefficiency that big is effectively tactical: too many players too often are allowed/encouraged to shoot from highly improbable settings.  Look at the graphic to the left; chart your team's shot locations over the next couple games and see how the pattern matches up.

Final post to come...why college coaches are holding the game back (and blaming youth coaches for it).