# Shots, Qualitatively: Part 1

I think it’s time we revisit the shot quality. More specifically, quantifying shot quality. I think using distance as a way of adjusting for the quality of shot has a lot of value that has previously been overlooked. Tom Awad showed in this famous article that, yes, shot quality is statistically significant, and with a few little refineries here and there I’m going to take another shot at measuring shot quality, and shed some light on this rather dark area of hockey analytics. In 2004 Graeme Jones created something called ‘weighted shots’. Michael Parkatti has done some work and the THOR ratings by Michael Shuckers also weight shots.

There’s been a lot of good attempts over the years, but I’ve found most attempts unsatisfactory or unscientific, so my hope is to create something that can use everything we have at our disposable to give us the most accuracy possible and, more importantly, something that can be universally accepted in hockey. I hope fully laying out the process here can help the latter.

If we only look at the bucket of shots within the 30-70% win expectancy group, which I’ve previously proved is a better way of calculating ‘close’, we see that there really is a strong correlation between shot distance and shot quality.

From distances 5 feet and out, shot distance and the average shooting percentages have a correlation of 0.99. We can use the equation of the polynomial line from this graph to give us a rough estimate of the ‘expected sh%’ by distance from the net. The first 4 feet didn’t fit this line very well, so they are treated separate from the equation (2 ft, 46.6%, 3 ft, 37.8%, 4 ft, 27.2%).

Using the described method to calculate 5 on 5 close shooting attempts, we’ll start this series off by looking at how shot quality has affected goalies.

Adjusted’ sv% is calculated just like regular sv%, except that shots are weighted by the shot quality (SQ), an estimate of the ‘expected sh%’ as calculated by the method above. The rank indicates their place in the league, and change the difference in those rankings. Minimum 300 shots faced. I realize this snuck in some backups. Whatever.

First our guys who’ve had their save percentage under represent their puck stopping ability

And our goalies who’ve been most negatively effected by adjusting their save% by distance of the shot.

Jonas Hiller, Cory Schneider and Corey Crawford have all had save percentages inflated by a low average distance of shot.

I’m pretty confident these numbers better reflect the quality of each individual shot. I mean, they better reflect the quality in that they do at all. Sure, it’s not perfect, but it’s definitely a step in the right direction. Basically what I’m saying is, it’s true that distance doesn’t perfectly represent any particular shots ‘real’ chance of going in, but isn’t that the similar much-reviled argument against other fancy stats like Corsi?

Stay tuned for part 2!