Goals against average – how many goals are the goalies fault? #Moneypuck

8 Nov

The hockey world in Hungary has been a little slow this week.  Olympic qualifiers start tomorrow and the MOL League is on break until after this weekend.  I will be live Tweeting and providing game recaps for EuroHockey.com.

In the meantime, I thought it would be a good to start thinking about advanced hockey metrics.  I’ve done some readings on advanced metrics being used in hockey, but according to this TSN article, hockey remains behind the curve.

A quick disclaimer…I am a numbers guy.  However, I understand the limitations.  I’ve played sports throughout my life and I understand that there are some things you cannot measure.  I do not think statistics should substitute workouts, film review, interviews and the like.  I do think they have an important role together and statistical analysis can enhance and make player decisions more efficient.

So here is my statistical story…..

Goalies are considered to maybe be the most important part of a team.  If you have a hot keeper, the other team cannot score, it makes it much harder to lose a game.  Goalies can also become the goat very quickly…being the last line of defense, a bad play of the puck or an easy goal could lose the game for his/her team.

I was thinking of something else today though.  The team I am working for has been outscored 33-93 through 20 games.  If you watch the starting goaltender, you feel bad for him.  He is standing on his head a lot of the time, night in and night out.  I’ve seen games where he has taken 50 to 60 shots on goal.  There is no doubt he has a bad defense in front of him.

Say I was this guy’s agent and I think he is talented enough to play in a more competitive league.  His stats alone and team’s record might discourage a lot of teams from trying him out.  Or maybe he would have a shot to play on the national team, but the guy with better stats, who happens to have a better defense in front of him, but isn’t as good, continually gets the nod.  There should be a way to differentiate blame for the goals that go in.  The goalie will always get some of the blame, but it should also be spread to his defense and also the offensive prowess of their opponent should also get due credit.

My goal then is to construct an adjusted goals against average rating that looks at backing out the blame/credit of others and helps us understand the goals per game that went in the net based as closely as possible on what I will call the keeper’s skill.

Ok, so the math… let’s say a goalkeeper’s goals against average is a function of a) her skill; b) the quality of the defense in front of her, and; c) the quality of the opposing offense.  Formally: f (a, b, c).  What are the factors or how does one determine then the elements of a, b, c?

First, I will examine who was on the ice for each team when a goal was made, then….

A – opposing team’s offensive prowess.  I think we need to take an average of the line that scores plus rating.  The goals, assists or points that players score are not enough because they don’t pick up the play of those they don’t make the score sheet…the guys making hits in the defensive zone, getting turnovers on a forecheck, drawing a penalty, etc.  Thus, take all players on the ice for a goal, take their plus rating, then divide it by the number of games they played (this could be further calibrated into minutes on the ice, maybe.  I’ll use per game stats for simplicity).  Then sum all the player’s on the ice plus per game average and divide that by the number of players on the offensive team, excluding the keeper, for that goal.  This will give an average of an average….plus rating per game line average seems like a good name.

B – basically the same rating as above need to be put into the equation, but we are going to look at the minus rating per game line average.  This will help determine how often the line that was scored on is scored on every game.

C – keeper’s goals against average.

Ok-so these are the basics, but because it is more complicated than looking at the above, and I am trying to make an “advanced” metrics model, there are some other considerations.

1)      Power play/penalty kill multiplier.  Some teams are better on the penalty and worse on the penalty kill.  Either way, being on the penalty kill doesn’t make a goalie’s job any easier.  He shouldn’t be faulted for that.

2)      Bad day/Good day.  In economics, and probably some other fields, they call this a shock.  I think it is important to look at the days the offense and defense exceeds or does not stand up to their normal expectations.  So, when the plus or minus ratings are higher or lower than the average we should know: what happens more often – good or bad play – and the results of each type of play.  This will give us the probability of a bad or good day occurring for the scoring lines and also a range of blame or credit to pass along on a bad or a good day.  Keepers have good and bad days too, so it will be good to see how often they are outplaying their defense, how often they are hurt by their defense and how often a good offense is affecting their play.  This is a pretty easy thing to pin down, I think, but I think it could be pretty powerful.

3)      Home/away effect.  Self-explanatory…just a modifier for A, B and C for a home or away game.

4)      Shot difficulty.  I will probably leave this one alone for now.  I think it is important to look at breakaways, deflections, rebounds and routing shots from different places.  When I build the model, I will add this in somehow, but I’ll set it to 0 or 1 so it doesn’t change anything.

I plan to build the model tonight and then run it over the weekend using last year’s NHL stats.  It will be interesting to see the results…mostly which keepers relied more on their defense, how keepers faired against teams with different offensive levels and how much of their success (or lack thereof) is due to a goalie’s own ability.  In the end, I hope to assign a portion of the goaltender’s goals against average to the offense and defense at the time of the goal and then back that out of the keepers GAA.  I will give it a first go this weekend.  Wish me (and goalkeepers with awful defenses) luck!


One Response to “Goals against average – how many goals are the goalies fault? #Moneypuck”


  1. Wednesday #MoneyPuck post. Predicting winners and losers. « Sport Exec in Training - November 14, 2012

    […] week in my MoneyPuck post, I discussed trying to figure out how many goals were a goalie’s fault.  I determined […]

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: