Last week, I made a post comparing 3-point shooting rates in the All-Star Weekend 3-Point Shootout to the participants' rates during the season. I found that contest shooters have made 3,669 of 7,173 shots (51.2%) in the contest over the years, and if you weighted their in-season 3FG% by their contest attempts, their usual season % is 40.5%.

However, longtime commenter Jason J noted that in a way, the 51.2% figure is artificially inflated by the fact that the best contest shooters will have the most attempts -- guys that get 10 points in the first round aren't equally weighted as guys that get 20, simply because the 20-point guy will move on to Round 2.

The solution, of course, is to simply take the raw average of all competitors' shooting rates, unweighted by contest attempts. To that end, I took every player's contest shooting record, aggregated by year (instead of by round), and averaged them all together no matter whether they flamed out in Round 1 or won a tiebreaker in Round 2. The typical shooting %s now are 47.6% for the contest, while the in-season rate remained 40.5% -- this implies that having 25 wide-open shots in a row only raises your 3-point % by ~ 7%, not the ~ 10% we had initially believed. Here's the dataset:

However, just as last week I inadvertently introduced selection bias into the process by weighting rates by contest attempts, even the list above has a degree of selection bias in the sense that multi-year contestants are given more weight than one-time participants -- and it stands to reason that players who are asked to come back to the contest for a second time (or more) are naturally going to be those who performed well, or at least better than "average" in their prior contests. To remedy this, another alternate way to look at the data is to aggregate each player's career 3-Point Contest performance vs. his in-season rates for the years in which he participated, giving equal weight to players whether they enjoyed multiple successes in the contest or flopped in their lone appearance:

As you see, the average in-season rate still doesn't change, hovering around 40.5%, but the contest rate drops again, this time to 46.2%.

Ultimately, there really is no one "right answer" to the question of how much better a given player will shoot in a 3-point contest vs. in games, since this is only looking at a very specific subset of players (the best 3-point shooters in basketball) anyway. But it is useful to eliminate some of the biases and try to find how players in that subset will be affected by the absence of constant pressure in their face as they shoot. From these indications, being free of defensive impediments is worth 6-7 points of 3-point % compared to in-game rates.

This entry was posted on Tuesday, February 16th, 2010 at 4:03 am and is filed under All-Star Game, Statgeekery. You can follow any responses to this entry through the RSS 2.0 feed. Both comments and pings are currently closed.