No, this is not a rare and dangerous condition that occurs when you spend too much time hunched over a keyboard (although whatever that is, I have it).
No, I mean I have DangerousFenwick WOWY data.
Before I go into more detail, be warned that this is a fairly numbers-heavy topic, so if that’s not your thing, my feelings won’t be hurt if you bail … no blood, no foul! Just like the officiating standard applied to Taylor Hall.
Otherwise, read on …
1 – What is a WOWY and why should I care?
WOWY stands for WithOut With You.
The idea is to look at pairs of players and see how they did when on the ice together, and then look at how they did individually without each other. By doing that, you often get a clear(er) picture of which players are zooming others, and which are being zoomed.
The most common statistic for WOWY is the old reliable, on ice shot attempts, aka Corsi.
For example, we could look at Lauri Korpikoski and Matt Hendricks:
|Korpikosi with Hendricks||267:04||42.3%|
|Korpikoski w/o Hendricks||259:29||41.3%|
|Hendricks w/o Korpikoski||257:39||48.1%|
Seems pretty clear, doesn’t it? Together, they really struggle. But apart, Koriposki still stinks … while Hendricks without Korpse pops up close to breakeven. That’s not the end-all be-all (who they play with when apart obviously makes a huge difference) … but it’s certainly a very good clue.
You can do this with any pair of players, though personally I find WOWYs indispensable when looking at defensemen, as fancystats can struggle when it comes to assessing them. I find WOWYs are one of the fancystats tools that seem to give reasonably good results.
2 – OK, but aren’t WOWYs fairly widely available already?
Why yes. Yes they are. I like stats.hockeyanalysis.com and the modernized version of it, puckalytics.com for pulling them. There are others.
Generally speaking, the only statistics available for WOWY are Goals and Corsi.
Not* the more sophisticated Score Adjusted Corsi, nor any of the danger weighted metrics like War On Ice’s Scoring/High Danger Scoring Chances, or DTMAboutHeart’s xG. Or my own DangerousFenwick for that matter.
*please do correct me if I’m wrong
Until now that is!
I now have WOWY data for my DangerousFenwick statistic, which weights all unblocked shots by the probability of their scoring (which itself is calculated from a 5-year analysis of shot probability by distance and shot type).
Does Player X get lots of shots but you suspect he’s a perimeter player? It will show clearly in the DangerFens if he is. Similarly, a defensemen who gives up lots of dangerous close in shots will also take a hit.
As an added bonus, I’ve also calculated Score Adjusted Corsi WOWYs.
Here’s the same table as before, this time with DangerousFenwick:
|Korpikosi with Hendricks||267:04||41.1%|
|Korpikoski w/o Hendricks||259:29||38.8%|
|Hendricks w/o Korpikoski||257:39||47.3%|
The numbers are reasonably close (as you’d usually expect them to be), but you can see the one thing that this does make clear is just how bad Korpikoski craters without Hendricks.
Signals Get Crossed and the Balance Distorted
This is my first full publication of the data, so it’s quite possible there may be some errors embedded in it still. I have validated it against puckalytics data, and it’s generally quite close most of the time.
With a large dataset, some small differences are to be expected as we all have slightly different rules for what we include as Corsi events, and also have different techniques for calculating shared time on ice. There’s also the minor difference of update times (if one dataset is a game ahead of the other, which I believe this dataset is right now, obviously that makes a difference).
Generally, my TOI and CF% are pretty much within a few minutes and half a percent of the puckalytics data, so I’m confident that I’m close, if not there!
The data for the Oilers is available as a CSV file (it’s a Windows zip file, with the csv and a readme file providing a glossary for the columns – link at bottom).
At some point, I may try to make this a live site like some of the others – but that’s a big undertaking and not one I want to pursue right at the moment.
Until then, I will periodically update these and make them available (other teams by request).
I hope that it proves interesting and useful. If you try it out, please leave comments below if you find any errors, or have suggestions, questions, or comments.
Thanks, and enjoy!
Download the data file here.