I got a lot of comments on my piece about Ilya Kovalchuk’s and Jody Shelley’s shooting talents to the effect of “who cares?” And I understand where they’re coming from: watch Kovalchuk and Shelley for two seconds and you can tell who has more shooting talent. Of course, no less than Paul Holmgren was fooled by randomness and signed Jody Shelley to a long-term contract. But assuming the rest of you around fooled, let’s look at something that’s less clear – Andrew Brunette (career Sh% = 18%) vs Dany Heatley (15.7%). Here’s their distribution of goals per 100 shots:
Which one of those guys is a better shooter? Well, we know Brunette works around the net, while Heatley is more of a slot guy. We see this when we look at their observed shooting performance:
Remember that shooting performance takes goal-scoring and adjusts it for where a player shot from on the ice. Brunette took a lot of higher percentage shots – so while he scores just as often (per shot) as Heatley does, his performance is just a little bit more likely. But those distributions don't tell us much – even over five seasons, we can only be 60-70% confident that Heatley is a better shooter than Brunette. Fortunately, there's another option…
What does their shooting talent look like if we start them with the same talent distribution and update the distribution forward over their last five seasons?
These distributions imply that it's about 90% likely that Heatley is a better shooter than Brunette. In other words, we probably don't need to concern ourselves with the possibility that Brunette is better…
