Ok guys, get ready for an absolute crap-ton of charts coming your way. This isn’t an explainer, merely just a data dump.
The best way to find your team is to hit Ctrl+F and use the team name. Otherwise, everyone’s in alphabetical order.
Feel free to save and use. Consider this blanket permission to use in blogs or other posts as long as it links back to here or my twitter, @Classlicity.
Ok guys, get ready for an absolute crap-ton of charts coming your way. This isn’t an explainer, merely just a data dump.
The best way to find your team is to hit Ctrl+F and use the team name. Otherwise, everyone’s in alphabetical order.
Feel free to save and use. Consider this blanket permission to use in blogs or other posts as long as it links back to here or my twitter, @Classlicity.
The new charts I’ve started sharing on twitter are a natural evolution from the old Salary Bands. I wanted a clearer way to show how a player has performed relative to his peers, while still keeping the important metrics front and center.
If you follow me on twitter, you’ve probably seen a lot of charts about offense over the last few days. All of these graphs represent a player’s individual offensive contribution to their teams based on two stats.
First, is Individual Corsi For per 60 (iCF/60). This basically tallies all the shots on target, missed shots, and blocked shots a player has taken up until I pulled the data, and then turned it into a rate stat.
The second is Personal Shooting %, often just called Shooting %, which is the number of goals a player has, divided by the number of shots on goal they’ve taken.
The first chart (I’ll use Dallas as an example) shows just this year’s numbers.
As you can see, Roussel’s Sh% is really high and likely to come down, especially because his iCF/60 is lower than average. On the other hand, Sharp, Seguin & Spezza have Sh% within a normal range, and have really high iCF/60, so they could probably continue that pace.
Graph two is slightly different.
This shows how the player is doing vs their results over the last 5 years. I’ve excluded rookies because, well, they don’t have any data from the last 5 years. Here we can see Hemsky smack dab at average, indicating he’s doing exactly what he’s always done. Roussel is actually shooting less than he normally does, and both Spezza and Val are shooting a lot more. Klingberg’s shooting % is extremely deflated compared to his normal 5v5 number, which is a scary thought for the rest of the Central Division.
The most important thing to remember is that these graphs only tell you about offense. If you have a player who is sound defensively, but doesn’t take a lot of shots, they won’t look good on these charts.
Under the jump, I have all the teams in alphabetical order by abbreviation.
As you may know, I spent last weekend in western New York, attending the first ever Rochester Institute of Technology Hockey Analytics Conference (or RITHAC for short). And not just attending, but presenting!
There have been some excellent write ups of the conference so far, including one by a Guest Contributor, Sean, over at Today’s Slapshot. This recap includes all the links to everyone’s slides (like mine) and links to the videos of the conference, which are time stamped so you can pick out what you want to watch.
First I just want to say how honored I was to be asked to speak. I think a lot of it has to do with my background – not just being a business background, but my active avoidance of academia, especially math and science – but I often feel like I’m not really an “analytics” person. So when I was asked to participate and saw the list of speakers I was like “what, you want me to speak? But everyone else on this list is brilliant.”
Needless to say, I was a little intimidated. Here’s the best advice I can give to anyone afraid of public speaking – just pretend you’re Kanye West. There is no one on this earth more convinced (rightly so or not) that what he says is important.
It’s finally the one you were all waiting for, the last (for now) in my Clutch rankings – the Playoffs.
Methodology wise, these are done the exact same way as the regular season, with weighted metrics measuring individual effort, team effect, and efficacy in Tied & Trailing states vs the Leading by 1 state. If you missed it, Part 1 is where I outline the definition of Clutch and what I’m attempting to achieve, but the final methodology used is explained in Part 2.
There is one major difference between the Regular Season ranks and the Playoff ranks, and that is the sample of skaters. Because of the comparative infrequency of the 5v5 playoff minutes, these are a cumulative total of the last five seasons (2011-2015), with a minimum requirement of 50 5v5 minutes to get on the list. This gives us a list of 431 forwards, and 236 defenders.
As a follow up to Version 1, I wanted to go through and refine how I was ranking “clutch” because, as was rightly pointed out, comparing players to just themselves often overvalued 3rd & 4th liners, and undervalued guys who were good in all situations. Compare the following:
Player Phryne*:
Lead by 1: 2.0 P60
Tied: 2.0 P60
Trail by 1: 2.2 P60
Trail by 2: 2.3 P60Player Jack:
Lead by 1: 1.1 P60
Tied: 1.5 P60
Trail by 1: 1.4 P60
Trail by 2: 1.6 P60NHL Avg:
Lead by 1: 1.5 P60
In all scenarios, Phryne is the better player, however, Phryne’s rank would be much worse, because the change between her Lead by 1 P60 and other game states is smaller. However, if I ranked them based on change from NHL Average, Jack would be unfairly disadvantaged as “clutch” because while he has a large personal delta, he mostly is right at NHL average in P60.
What I ended up with was a Lead 1 average weighted by on Time on Ice. As mentioned in the first article, coaches don’t typically change their usage of players within the varying game states, so a fourth liner will always have less TOI than a star.
