Cornell @ Clark

[Tom Lento]

The reason it focuses on the Flyers is that one of the bloggers at Broad Street Hockey tracked every zone entry and result of that zone entry for the entire (season)

This sort of thing makes me feel well adjusted.

I do believe SB Nation pays their bloggers, but its still insane.

With the right tools it might not be that much harder than watching every game for the purpose of writing a game-by-game season diary. Which would also be insane, but somehow feels a lot more normal for a sportswriter.

Personally, I love the fact that they did this. Based on a quick read I think there are some flaws in their analysis, and it seems like they need better data and more of it to correct the most obvious failings (e.g., shots on goal instead of a direct measure of scoring). Because I'm also crazy, I've been thinking about ways to collect better data, and I believe a fairly simple tablet app could work. It might be a fun side project, build a UI for tracking turnovers, zone transitions, shot locations and outcomes, build in an automated uploader to collect data from anybody crazy enough to do this for even one game, and do some analysis... but I doubt anyone would ever use it.

According to the guy who did it, he just used a DVR.

I'd be surprised if he attempted to do this live. How did he track the zone entries and shots? Pen and paper? Spreadsheet? Custom software?

Pen and paper Methodology here

This summary account of the methodology and overall conclusions hints at potential flaw in the analysis.  I don't see mention of the quality the opponents defenders as an independent variable.  The skills of the individual Flyers defenders is acknowledged as contributing the neutral zone activity.  I wonder whether, on the average, better opposing defenders reduced the Flyers forwards' ability to carry the puck in and also reduced their ability to score once the puck was in the zone.  That would generate the reported correlation.

That *is* insane.

They should be able to get the opponent's d-men, at least approximately, based on the same shift charts they used for offensive data. Rating those d-men might be another painful task, but it'll be interesting to see if they do it.