Pittsburgh Penguins

Taking the next step in merging the eye test with data analysis
April 4, 2017, 12:22 PM ET [107 Comments]
Ryan Wilson
Pittsburgh Penguins Blogger • RSS • Archive • CONTACT
One of the biggest rivalries in hockey doesn't involve players or teams. It involves data driven analysis versus the eye test. While data analysis' value has grown acceptance over the years there are still people who dismiss the idea that numbers are better than eyes.

Relying solely on either method is a sure fire way to know that you aren't doing the analysis thing correctly. However, the problem with the eye test is that the human brain cannot and will not remember everything you see. In fact, it won't come close to remembering the amount of details necessary to make it useful analysis on its own. Ideally, we would be able to see things and then have them documented for reference so that we can lean back on that useful information we saw, but could not remember.

Thankfully, Ryan Stimson and his army of trackers have been working diligently on having the eye test flow in concert with the data driven approach. This takes an enormous amount of work because the things that are useful to track have to be done manually. The NHL isn't up to speed with leagues like the NBA and MLB with player tracking technology. Ryan is working towards changing that one game at a time. Here is his latest work which is awesome.




Corsi was never meant to be a catch all statistic. It was never meant to be the "god particle" of analysis. Nobody responsible uses it that way despite the straw man arguments you hear. However, it was and still is to an extent the best we have to go on. That is until Ryan's sample size of data grows. Here are the things that Ryan has looked at with his tracking project:

The metrics are below. Each metric’s per-sixty minute rate is multiplied by the percentage of the total on-ice events that player creates:

• Shot Index (individual shots)
• Shot Assist Index (primary shot assists )
• Build Up Index (secondary and tertiary shot assists)
• Transition Index (controlled entry assists)
• Danger Index (Shot Contributions from below the end line or across the slot)
• Influence Index (Total shot contributions)
• Pass Index (Total shot assists)


These are all things that we see with our eyes, but don't accurately remember to fall back on. Now we have a growing data base of these ultra important events to reference. For me this is one of the best breakthroughs in analyzing hockey. We have data and eye test working in concert with one another which was always the goal.

Ryan has also created an interactive web tool so that we can visualize these findings. For example, Ryan and his tracking group have 52 games of Penguins data from last year and this year to pull from.

The Sheary-Crosby-Guentzel line is all the rage right now as it should be, but this data doesn't have Guentzel's sample size. As a result I am going to use Ryan's data to focus on Sheary-Crosby-Hornqvist which was one of Mike Sullivan's go to lines in this sample size. This visual can help tell us who is doing what when they are on the ice. Is a player well rounded? Are they limited? What are their strength or weaknesses? Here is a look at Ryan's visual tool:

source: imgur.com
*Crosby is in green, Sheary is in blue, Hornqvist is the red outline

Sidney Crosby is great at everything besides his shot volume and even then it is in the 75th percentile among his peers. He does the heavy lifting. He is amazing.

Patric Hornqvist is a limited player. Within that limit he excels, but there isn't much outside of it. His shot volume is high and that is his strength. This isn't a surprise to anybody who has watched him play. However, the amount of areas where his contributions are at or below the 50% threshold in relative to his peers are high. The reason he works on a line with Sidney Crosby and Conor Sheary is because those two players are doing all the heavy lifting. There are a lot of candidates who could play the high shot volume game and succeed with a player like Sidney Crosby. A chart like the one above can paint a picture of why Hornqvist has been removed from the Crosby line in favor of Jake Guentzel. Sid wants player(s) like Conor Sheary (and I assume Guentzel) to excel at the other stuff that contributes to offense. Conor Sheary's contributions on the ice are both impressive and diverse when you look at the graphic. These are things like contributing to transition, building up plays, passing ability, primary shot assists etc... These variables are what keeps plays alive. They are what drives possession and offense.


I have written/podcasted about the limitations in Hornqvist's game in the past, but that was subjective analysis which is always open to criticism. Now with the work Ryan is doing I have statistical evidence to back up my original hypothesis. This is a great example of eye test and data merging to form an analysis of a player. This is the kind of information you are going to want to pay attention to when you look to make decisions on a players future IE: Hornqvist's expiring contract.

I do want to be clear that Patric Hornqvist is not a useless player. Within his limits he excels. You just need to know what those limits are and who can get the most out of him. He is a defensively responsible shot volume specialist who shoots at 9%. You need to be able to look at that role and compare it to other players on the team to best allocate ice time.

The example above is just one way on how to use the data. In the future I will be looking at many other avenues of analysis with this new information.

Today was a victory for both data and eye test folks alike. This is the future of hockey analysis.