Fun With Shot Data (Eric Gryba, Erik Karlsson)

[Bear of Bad News]

http://www.the6thsens.com/2014-articles/march/fun-with-shot-data.html

Playing with scoring chances, possession time, and shot-based metrics.

Click to read more...

 

[Micklebot]

http://www.the6thsens.com/2014-articles/march/fun-with-shot-data.html

Playing with scoring chances, possession time, and shot-based metrics.

It does note that the leafs limiting quality scoring chances compared to shot totals seems to hold up, but they are still allowing a lot more than they get themselves.

I'd imagine Karlsson's lower Sc% has something to do with him taking a lot of shots while on the ice; his iCorsi is second only to Ovechkin. If you're only counting shots from within the homeplate, this would likely be a detriment for a defenseman. Gryba on the other hand is far more likely to pass the puck to the forwards.

 

[Wesleyy]

When I was developing LAEGAP the first source of shot location data I looked at was the Super Shot Search website, but when I tried to verify the data I found there to be inaccuracies and lots of missing data when compared to the same games over at NHL.com and their backend server, which is what I ended up using. I am not sure if whatever bugs were causing the inaccuracies are fixed now, but if not, the data might be further skewed, besides being skewed from the league's bad recording practices. Just wanted to share my experience.

 

[hatterson]

It does note that the leafs limiting quality scoring chances compared to shot totals seems to hold up, but they are still allowing a lot more than they get themselves.

It does seem that they're one of the best (2nd best by the numbers) in terms of scoring chances per fenwick against however the simple fact that we allow so many chances against means that we're still second worst in terms of overall scoring chance +/-

It's also worth noting that the Leafs are very middle of the road when it comes to taking "quality" shots (16th out of 30th) and slightly below the mean level.

 

[Master_Of_Districts]

Interesting.

According to the data linked to, the correlation between fenwick percentage and (estimated) EV scoring chance percentage is 0.82.

If the ability to out-fenwick was identical to the ability to out-chance, we'd expect to observe a correlation of about 0.89 (over a sample this large).

So I suppose the inference is that the ability to outchance is independent of the ability to out-fenwick to some small degree.

This is consistent with even strength shot quality data, where teams do consistently better or worse than average - beyond what would be predicted by chance alone.

If you look at subjectively recorded EV scoring chances, though, the ability to out-fenwick looks identical to the ability to outchance. Again, that's based on a comparison between the observed correlation relative to what would be expected.

 

[Bear of Bad News]

Fresh article, looking at the Ottawa Senators' player performance by zone:

http://the6thsens.com/2014-articles/assessing-player-performance-by-zone.html