New Tool: Player Usage Charts

Robert Vollman
April 01 2013 08:39AM

 

 

There's a fantastic new interactive resource out there for hockey fans that lets you put together  customizable Player Usage Chart.  I'm pleased to have been invited to tell you where you can find it, what you can do with it, and even a little bit about where it came from.

First, a little bit of background on Player Usage Charts. They were first introduced (by yours truly) in the 2011 off-season on Arctic Ice Hockey as a graphical representation of the player usage data the site's owner (Gabriel Desjardins) developed and hosts; offensive zone start percentage and quality of competition.

Since context is everything in the world of hockey statistics the idea then, as it remains today, was to show at a glance how a team is deploying its players – whether they're being used more in the offensive zone or the defensive zone, and against mainly the top lines, or more the depth lines.  It some ways it was a real eye-opener to see who was handling the tough minutes and who was being sheltered.

Though this first incarnation was quite primitive and poorly named (OZQoC Charts), the idea caught on quickly. The very next day, for example, famed hockey statistical analyst Eric T at Broad Street Hockey introduced a version that first included Corsi-bubbles – instead of a single point, Eric used a coloured bubble that was sized based on how well the team did with that player in terms of attempted shot differential (aka Corsi).

Eric's continued use of Player Usage Charts, or OZ Coke Charts as they were still known, caught the attention of his colleagues including Shutdown Line's Corey Sznajder, Derek Zona of Copper and Blue, John Fischer of In Lou We Trust, Rob Luker at Blue Shirt Banter, JP at Japers Rink, Josh Lile at Defending Big D, and many more. I myself used them as part of a weekly statistical update when working for the FlamesNation crew that season.

So in the 2012 off-season I put together the re-branded Player Usage Charts for every team, inviting them all to contribute their own analysis, and in the end a couple of dozen analysts, including TSN's Scott Cullen, helped bring it mainstream (within our little niche, at least). Later that same year, Player Usage Charts were also included in Hockey Prospectus 2012-13 annual book, and are already being slated for inclusion in additional publications next year.

Unfortunately Player Usage Charts were only available in the off-season, and mid-season updates had to be put together manually, which can be a time-consuming process. 

Enter visualization expert Robb Tufts, who had used a technology called Tableau to make an interactive and customizable Player Usage Chart for the St. Louis Blues. Amazingly, Tufts even included Tom Awad's early suggestion by actually sizing the Corsi-bubbles based on a player's average ice-time, and shading it based on how well the team did shot-wise. 

(Note: Fans of the original Player Usage Charts can still get them from Ninja Greg!).

Which brings us to the new tool, now available on Hockey Abstract, which allows you to generate your own Player Usage Charts any time you want.  Not only can you do all the basic things, like choose the team, filter on positions and games played, but you can size the bubble based either on average time-on-ice per game, or total ice-time, and shade the bubbles based on Corsi – either straight-up or relative to one's teammates, or Hockey Prospectus's GVT – either per-game or over-all total.

How to Use

For example, recently I used this tool to highlight how the Caps are actually struggling possession-wise with Mike Ribeiro on the ice despite his sheltered assignment (chart one), and again to show how the Edmonton Oilers potentially have enough top-four defensemen to pick up someone that would need to be more sheltered (chart two).

Labels can be added, players can be highlighted, sections can be zoomed in on, all the individual data components can be reviewed either in the sortable table below the chart or hovering over a data point, among many others bells and whistles.  When the Player Usage Chart is done it can be tweeted, linked, emailed, or even just saved to a file. 

Tufts even made a YouTube video which shows how to use them in greater detail, including many of the bonus features – it is worth investing the seven minutes to view the whole thing.

The Player Usage Chart tool has been promoted heavily since its introduction March 25th, on twitter, various web sites, and even on CBC's Hockey Night in Canada Radio, with Elliott Friedman and Kelly Hrudey, both of whom have become real believers in this type of work. Who knows - perhaps some day a game's opening line-ups will be presented in some form of Player Usage Chart.

Thanks to the involvement of so many people, Player Usage Charts have continually evolved, and hopefully soon will include the final important aspect of a player's usage – the quality of their linemates.  Even though players will be generally bunched together with their regular linemates, and with similarly sized and shaded bubbles, and it can therefore be guessed with whom everyone is playing, the tool will hopefully include this information explicitly.

The vertical axis is another intriguing area of improvement. Not only to make it possible to use something else instead, like quality of teammates, but to change the method by which the average quality of competition is measured. Currently is uses the most preferred readily available method, which is the average Relative Corsi of one's opponents, but perhaps some day one of the alternatives, like Eric's version based on the average ice-time of one's opponents, will be available instead.

Another feature under development is the ability to create Player Usage Charts that aren't team-based. For instance, a Chart with all the rookies, or all the top-line defensemen, or everyone on the trading block.  Though it's amazing how far these charts have come in such a short while, it's also exciting about what is yet to come.

On that note, watch Robb's video, head on over to use the Player Usage Chart tool yourself, and share what you come up with.

781445e1560a4e1d0d641bc270a7eedc
Rob Vollman of www.HockeyAbstract.com is a regular feature writer on ESPN Insider, co-author of Hockey Prospectus 2010-11 and 2011-12, and regular contributor to NHL Numbers, Flames Nation and Arctic Ice Hockey. Innovator of Player Usage Charts, Quality Starts, GVS (Goals Versus Salary), the Snepsts Projection System, and known for work in League Equivalencies (NHLE). Twitter: @robvollmanNHL
Avatar
#1 Dawn
April 01 2013, 09:39AM
Trash it!
0
trashes
Props
0
props

Was just playing around and looked at the Kings. Interesting observation re team (rather than individual) analysis - Just about everybody is on the O-zone start side. Not sure if I'm interpreting this correctly. But, it appears that as a team, they must not have many D-zone faceoffs. I know they're among the highest in possession metrics. So I guess it follows that they would be spending a lot of time in the O-zone. It will be interesting as the body of data builds to see what the profile of the cup winning team looks like from year to year.

It would be sweet to add a function to go back to previous years....

Avatar
#2 Dawn
April 01 2013, 10:06AM
Trash it!
0
trashes
Props
0
props

Playing around some more...

With a quick look by eye, cut off about 6 players on one side alone.

Majority O-zone starts - Hawks, Wings, Panthers, Kings, Rangers, Sens, Yotes, Sharks

Majority D-zone starts - Flames, Columbus, Oilers, Wild, Preds, Flyers, Leafs, Caps

With a few notable exceptions (Panthers, Leafs, Wild), it seems to play out that the better performing teams start in the O-zone a lot more. Seems self evident, I know. The exceptions interest me though. Is it PDO (aka puck luck)? Is it shooting percentage? Something else or all of the above?

Going back to play...

Avatar
#4 Dawn
April 01 2013, 02:31PM
Trash it!
0
trashes
Props
0
props

So I would expect over a full 82 game season, the exceptions would tend to disappear. Generalizing: good possession teams = more successful teams = player bubbles all clustering in the O-zone side of the chart

I guess that's a splash of cold water on my beloved Oilers playoff hopes if it were a full season. ;)

Avatar
#5 tony
April 01 2013, 07:54PM
Trash it!
0
trashes
Props
0
props

this is cool! i notice that the wilds first line that scores most of their goals sits int he upper right with koivu, parise and coyle looking really good and the rest of the team starts in the defensive zone.

Avatar
#6 tony
April 01 2013, 07:56PM
Trash it!
0
trashes
Props
0
props

consistent with the PDO numbers for the ducks they have not a single player in the + corsi+ offensive zone starts portion of the graph.

Avatar
#7 Ty
April 02 2013, 01:32PM
Trash it!
0
trashes
Props
0
props

Might be that teams with a higher fenwick have a higher O-Zone start and teams with a lower fenwick have a higher D-Zone start. Basically, teams that possess the puck more are going to have more scoring opportunities, while taking away the scoring opportunities of the opponent, resulting in more wins. You can't win if you don't possess the puck.

Avatar
#8 Graphic Comments
April 05 2013, 02:21PM
Trash it!
0
trashes
Props
0
props

First off, this is awesome. Thanks for putting it together and putting it out there for anyone to use. Top notch stuff.

Second, it might be just me, but I would love to see the option for dynamic charts that show how player usage evolves over the season. Not sure if that's an option that is possible within the Tableau platform, but something similar to the Google Motion chart I used when I did the analysis of the Forbes team data way back when (way down at the bottom of this post):

http://nhlnumbers.com/2012/9/17/winners-and-losers-under-the-current-cba-part-3

Would be really interesting to see this and pick out how coaches adjust strategies or how teams adapt to injuries, trades, etc.

Avatar
#9 Jonathan Happy
April 07 2013, 05:30PM
Trash it!
0
trashes
Props
0
props

Robert what happened to the Black Box's that you were doing last season on FN? I really enjoyed those.

Avatar
#10 Jonathan Happy
April 07 2013, 05:40PM
Trash it!
0
trashes
Props
0
props

Wow this stuff is absolutley sick. Thanks alot for putting it up

Comments are closed for this article.