Hockey analytics at a glance

On Tuesday, I posted the first of what will be a weekly update on the Everett Silvertips. In one section I mentioned that defenseman Jake Christiansen looks initially like the best offensive weapon on the blue line because of his transitions.

“Jake Christiansen is a good candidate to lead the defense in points due to a strong transition game. After tracking zone entries on Saturday, he was the only defender to successfully carry the puck in, and accounted for nearly half of all entries from defensemen.”

I realized shortly after writing this that perhaps not everyone is familiar with some of the advanced statistics that are currently popping up in the hockey world. Many of these stats aren’t officially tracked by the NHL or any other league, but instead tracked by hand by some extremely dedicated fans. Most of the work being done is strictly on the NHL, but there are a few people working on different leagues. Mitch Brown (@MitchLBrown) is in year 2 of his CHL tracking project, but even that is focused more on just the best NHL prospects. I’ve made it my goal this year to track some of the new stats for the Silvertips this year myself, and if I find the time ideally I will be able to track some other WHL teams as well in order to get a better idea of what league average is in these statistics. Since I will be using some of these in my writing this season, I decided to give a quick overview of some of these stats.

Important note: All data, unless specifically stated otherwise, is being tracked only during 5 on 5 play. Power plays are excluded because the game changes so much with a man advantage and because the majority of hockey is played at 5 on 5.

Zone Entries

Offensive zone entry data is what I looked at when I commented on Christiansen. Hockey analytics legend Corey Sznajder (@ShutdownLine) does a great job of defining zone entries in his All Three Zones Project:

“Zone entries show how often each team and player entered the offensive zone and how often they did it with or without possession of the puck.”

Sznajder painstakingly tracked entry (and exit) data for every NHL team in 2013. I won’t be able to do that, but I will be keeping track of how well Silvertips players enter the offensive zone with control of the puck. Why is it important to enter the zone with possession as opposed to the old tried and true dump-and-chase method? Here’s Sznajder again:

“Through the work of Eric Tulsky, Geoffrey Detweiler and Bob Spencer, we were able to determine that zone entries and being able to enter the offensive zone with possession of the puck is a major factor in out-shooting your opponent during five-on-five play. Their work showed that entries done by possession lead to twice as many shots as compared to dumping the puck in.”

It seems fairly intuitive that being able to carry the puck into the offensive zone would generate more shots, but I don’t think anyone expected it to generate twice as many shots as dumping it in. That’s a huge difference, and tracking which players do this best can give us great insight into who is helping the team the most.

Defining zone entries is one thing, but a statistic doesn’t really help you out if you don’t know how to analyze the actual numbers. In general, the number of zone entries in any given game can vary greatly. They are also very dependent on how much ice time the player is getting. Let’s use the zone entry chart I made for the Silvertips home opener as an example.

The column titled “5v5 Entries” is a count of the number of times that player moved the puck into the offensive zone. This includes times he carried it in as well as when he dumped it in.* The next column is a count of all the shots generated from these entries. The next column, “5v5 Carries,” is a count of only those zone entries which the player carried the puck into the offensive zone himself. The shots column after that is a count of the shots generated from those entries. Finally, we have Carry %. This is the column I’m looking at the most.

*Occurrences where the player dumped the puck in with no attempt at retrieving it, such as when the team is making a line change, are excluded

Let’s pull a couple examples. This game unfortunately was filled with penalties (20 minutes between the two teams) which means there wasn’t as much 5 on 5 time as usual, so the numbers here are a bit lower than average. Bryce Kindopp brought the puck into the offensive zone 6 times, 2 more than the right wing ahead of him on the depth chart, Martin Fasko-Rudas. But of those 6 entries, only 1 time was the puck carried across. The majority of his entries were dump-ins, which we know result in fewer shots. Fasko-Rudas, on the other hand, carried the puck in on all 4 of his entries. His carries only resulted in 2 shots, so it may be worth looking into what he’s doing with the puck once he gets across the blue line. But that could also just be small due to the sample size.

Next, check out the numbers for the defensemen, specifically the last 3 columns. Of all 6 defensemen, only Christiansen entered the zone with possession of the puck. The carry % on all other defensemen is 0 compared to his 66.67. It’s difficult to classify what number would be considered “good” and what would be “bad” for entries and carry %. Oftentimes I simply compare it to his teammates. Again, the sample size is small, but Christiansen clearly outperformed his teammates in this area of the game. It will be really interesting to see if this continues as the season rolls along.

Now, when Sznajder, or other hockey analytics people who are much smarter than me, talk about “outshooting” your opponent, they are not talking about shots on goal. They’re talking about everyone’s favorite analytics heckling target:

Corsi

Corsi is a strange name for a fairly simple statistic. In its most basic form, Corsi is simply a measure of shot differential. It’s how many shots your team took compared to the other team. The unique part is that Corsi accounts for every shot, regardless of if it was on target or not. Standard hockey shots on goal don’t take into account shots that get blocked, miss the net, or even shots that hit the post. Corsi counts all of those, and when taken as a whole can give you a good measure of which team controlled the puck more in any given game. It often gets broken down into percentages and relative percentages, which can be used to compare how much your team was shooting the puck with player X on the ice vs when he was off the ice. Old Time Hockey Guy™ will tell you that Corsi doesn’t matter, that it’s just a bunch of computer geeks trying to turn hockey into numbers. “You can’t measure GRIT with Corsi, nerd!”

https://deadspin.com/old-gm-yells-at-cloud-1825600773

And in some ways they are right. Corsi is by no means a perfect statistic, and it alone won’t tell you which player or team is better than the other. The main reason I want to talk about it here is that when I am tracking zone entries, I’m also tracking how many shot attempts are generated with each entry. When I track a shot attempt, I’m using the Corsi definition rather than the standard shots on goal metric that hockey people are used to seeing.

How do we know when a Corsi score is good?

Generally, Corsi is depicted as a percentage of the total shots. If a player has a Corsi score of 50, that means that his team accounts for half of the total shots taken while he is on the ice. Higher than 50 means his team is outshooting his opponent, and conversely a score lower than 50 means his team is being outshot. Typically Corsi scores will stay in the 40’s and 50’s. The highest Corsi % in the NHL last season among players with at least 25 games was 57.7 by Matthew Tkachuk.

If you want to read more about Corsi and all its offspring, Second City Hockey did a great introduction in a three part series found here.

This little introduction barely scratches the surface of the in depth data analysis going on in the hockey world right now. If you want to keep looking around for more advanced stuff like Expected Goals or GameScore, I highly recommend checking out Ryan Stimson’s (@RK_Stimp) passing project data and Micah Blake McCurdy’s (@IneffectiveMath) www.hockeyviz.com. If you’ve got any other questions or if you just want to tell me math is dumb, let me know either in the comments or on twitter.