ShotCaller: Beautiful Struggle
By Matt D'Anna
[ED Note: This post is a follow-up to Matt’s Shot Caller introduction, attempting to integrate predictive analytics with patterns of shot selection.]
As promised, every prediction we publish this season will be evaluated — for better or worse, no shame in our game. Here’s how the first one panned out.
The original prediction:
During the 1st quarter of the New York Knicks season opener against the Chicago Bulls, Carmelo Anthony is expected to take 5 shots; those shots are expected to occur within each of the red ovals.
The result:
Well that doesn’t look so good… but let’s break this down into its parts:
Melo only took 3 shots, instead of 5. Three is an interesting number, since over the course of his career Carmelo Anthony had never taken exactly 3 shots during the first quarter of the first game of the season. In 9 of his 11 seasons Melo has taken 4 or more shots in these games:
Needless to say, 3 is low; but by no means extraordinary low.
Where the 3 shots came from. One of the shots (the one at the rim) is a total bulls-eye. This shot is arguably the easiest and least interesting to predict — but a match is a match! The shot near the elbow is 6ft from two of the predicted areas, in a triangulation kind of way. Less than ideal, but not necessary terrible. However, ‘terrible’ is a great descriptor for the third shot — way off. I had zero indications of that shot occurring.
Upon deeper inspection, consider the results from this perspective:
If we momentarily ignore my prediction areas, we can see just how anomalous that infamous third-shot-from-the-top-of-the-key really was. The shot at the rim was in the strongest cluster of Melo’s activity, at the intersection of the three unique datasets involved in the original analysis — last season’s Hunting Grounds; career Game 1, 1st quarter Hunting Grounds; and last season’s 1st quarter activity versus Chicago (this was an attempt to include macro- and micro- levels of information). At the rim was the only area where all three intersected. The second shot occurs at an intersection between the two types of Hunting Grounds — one of six areas on the court where the purple meets the green. And then there’s that third shot – it’s near some of the layers, but overlapped with none. It was rare and out of character — I swear!
My running ShotCaller metric: 20%
This number captures the bulls-eyes, relative to the number of areas I identified. I’ll update this with each new prediction evaluation.
For a first pass I don’t consider this terrible, but there is certainly room to grow as the season progresses.
Data and photo support provided courtesy of NBA.com, Basketball-Reference.com, and Darryl Blackport.