Can Shot Statistics Lie to You? The Kevan Miller Story.

Analytics allow a person to “not watch the games.” In essence, it is impossible for a human to keep track of corsi (shots) numbers because the game is so fast, and remembering over +100 events in one game, let alone a season, would be something only a true wizard can do. That is why we use computers and NHL play-by-play sheets. That is not to say they tell the whole story, but they record a player’s output in some easy to understand statistics.

Kevan Miller seems to be one of those players that nobody cares what his numbers say, he is just bad. When you watch him, he can’t skate with the puck, can’t skate to the puck, and looks awkward in his own zone. However, his macro (shot) statistics are not too shabby. Take a look at the chart below. (Data from Corsica.hockey)

He even looks better than Torey Krug when we look at expected goals for percentage. So why do Bruins fans hate him when they watch him, even though he has great output numbers? Well let’s remember that these stats still have a relatively low R^2. xGF% is the highest R^2 of a shot statistic that we have with an R^2 of .28. That means that it explains 28% of the on ice events. You can read more about it here, but that is about all we can juice out of shots. The movement to micro (passing, entries, exits, etc.) statistics will be able to improve on these models later on. Right now we have smaller sample sizes for micro statistics, but we can try to use them to explain a little more about what is going on while a player is on the ice. Take a look at the charts below.

This is one of the first visuals of the passing project. I highly suggest you go play around with these. They show the path the puck travelled and approximately where the pass was received. We can see from the charts above that the preferred pass through the neutral zone is a move towards Miller. From there the offence has a greater chance on getting a shot on goal and eventually scoring. You can explain this how to want, but this may be due to his awkward skating and lack of hockey IQ.

However, let’s look at him compared to the team. If I showed you the all of the shots against that included a shot assist, it would be useless. I will use scoring chances only to show this. Here are the team’s home and away charts.

Here is Kevan Miller’s home and away charts.

One thing that pops right out is that the opposing team has the ability to penetrate that home plate area a lot easier with Miller out on the ice, and there are more passes received in that area too. Shot statistics don’t include this because the NHL doesn’t publically track it. This may or may not be due to Kevan Miller alone, but in the games tracked, Miller is certainly not that anchor defender that Julien makes him out to me.

Please don’t take this article and throw away the macro stats. Corsi, Fenwick, and Expected Goals are proven indicators, while we are just learning about passing data. I intended to show that there is some correlation to your eye test and Miller’s performance. If you really want to show how “bad” Kevan Miller is, then track more micro stats like entries and exits and tell us all about it.

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