clock menu more-arrow no yes mobile

Filed under:

Mid-December Goaltender CSSI Look

When we completed last year's CSSI stats, one of the persistent issues we had with it was in trying to attain a more-perfect goalie-rating system. While points and plus/minus did a very good job telling the story of the skaters, last year's goalie totals lacked a bit of context and had to be manually adjusted.

Going into this season, we decided to track a few new metrics. While doing CSSI last year, we had a system of goalie plus/minus where the netminder would be given a score based on big saves made minus bad goals given up. We discovered that this system essentially turned a +2 rating into an average game. For this season, we've done away with counting the big saves, but not the bad goals. We've also added a "goals forgiven" category for those which sufficiently fluked their way past the goalie or otherwise could be blamed completely on other forces. Aside from that, goalies are being tracked on two systems now.

Head-to-Head Rating: A rating of -1, 0, or +1 given to a goalie as a measure of how well he played in comparison to the opposing goaltender. A goalie can play extremely well and get a -1 if the guy on the other side of the ice plays that much better. Meanwhile, a losing goalie can get a +1.

Overall Game Rating: A rating on a scale of -3 to +3 (0 being average) given to a goaltender for his performance during a game. Ratings on the severe end of the scale are used very rarely. This rating is meant to cover a goalie's overall impact on the outcome of the game. Playing as a non-factor,a goalie will earn a 0 rating regardless of the outcome. If a winning team has to overcome soft goals, their goalie will earn a negative rating. If a losing team is held into the game through his work, their goaltender will earn a positive rating.

With the new ratings system in mind, let's take a look at how Jimmy Howard has done in net for the Red Wings this season and try as best we can to compare that to last year's performance.

Last season, since goals forgiven did not exist, we can't really use those for a comparison. Everything else is fair game though. Since we're looking at what a goalie is going to do for a team over the course of a game and comparing different-sized time periods, we're going to be using rate stats for comparisons. For that, we simply Take the number of times a goalie has earned a specific stat divided by his ice time and then multiply it out to get how often he'd be expected to do it over the course of 60 minutes.

2010-11 Season 2010-11 Rate 2011-12 Season 2011-12 Rate
TOI 3,615:15 1,664:15
Shots Against 1830 30.37 739 26.64
Saves 1662 27.58 686 24.73
Goals Against 168 2.79 53 1.91
Save % 0.908 0.908 0.928 0.928
GAA 2.79 2.79 1.91 1.91
Bad Goals 27 0.45 7 0.25
Overall Rating* +24.5 0.41 +20 0.72

You don't exactly need a microscope and abacus to see that Howard is playing better this season than he was last year. The official stats tell this story very well on their own. However, let's take the bad goal stats a bit deeper.

Shots per Bad Goal

1/(Bad Goals / Shots Against)

Last Season: 67.78
This Season: 105.57

Howard is seeing fewer shots per game this season, but he's also stopping way more of the ones he needs to be stopping. So far this season, he's gone over 100 shots against between giving up a soft one. The rate difference here is severe.

Bad Goal Frequency

(Shots per Bad Goal / Shots-Against-Per-60)

Last Season: 2.23
This Season: 3.96

This further utilizes the difference in shots per game and bad goal rate to show how much better Howard has been. Last season, you could expect him to give up one bad goal for roughly every 130 minutes. This season, he gives up about one every four games.


(What if Howard played this year like he played last year?)

Let's say that everything else is equal except for Howard's Bag Goal rate. Let's look at what his stats line would show this season:

Shots Against Saves Goals Against Save% Goals Against Average
This Season (Official) 739 686 53 0.928 1.91
This Season (Adjusted) 739 675 64 0.913 2.31

That's pretty impressive all by itself, wouldn't you say? These numbers are taken strictly from calculating out for the new bad goal rate. Of course, numbers are rounded where they have to be (it would be 10.9 bad goals instead of 7, so I rounded up to 11). I don't want to take guesses here, but I can imagine that the Red Wings' record would not look quite as good as it does today if Jimmy Howard simply let in more goals that he shouldn't.

For reference, if Howard somehow managed to go the entire season to this point without giving up a single bad goal, his save % would be 0.938.

The other part of the stats system, the Head-to-Head rating doesn't work as well for comparison, but it does indicate that Howard has been very good in rising to the level of a challenger. His +7 rating is a +0.25 rate stat. Considering this statistic will always fall between -1 and +1, this on the correct side of and a good distance from center on an expected curve of scores.

What it all tells us about the play of Jimmy Howard so far this season is that there isn't one cause of the jump in numbers. It would be foolish to pretend that the Red Wings' defense hasn't improved over last season (so far), but that doesn't account for the whole story of how Howard is doing the little things right with much more consistency. After all, a bad goal doesn't always happen when a guy just whiffs on a fluttering puck. Many of them are counted by him badly playing rebounds, jumping out to unwise challenges, recovering slowly from puck movement, and other small mistakes. Jimmy Howard is making far fewer of these this season than he did last year. that is a good sign for our All-Star Ballot snub.

[*The Overall Rating category for the 2010-11 season was adjusted. While the overall ratings systems are judged completely differently between this year and the last, it was well-established that a 0 rating for this year equaled a +2 rating last season. With that in mind, the +24.5 overall rate was gathered by taking the per-60 rate of 2.406, subtracting 2 from it, and re-multiplying it out by the season's numbers. I am very satisfied with this adjustment despite its imperfect nature for comparison]