The Future of Basketball Stats, Continued


By Jon Nichols

Earlier I wrote a little piece about how far basketball statistics have come and where I think they’re headed.  To recap, I think the next ground that will be broken is in the area of combining scouting and coaching strategies with useful and complex stats.  In other words, instead of focusing on rating a player’s overall abilities with one single number, smart folks who deal with stats will be able to break a player’s game into small parts and help teams make decisions that were more difficult in the past.  Think of the numbers at on steroids.  Coaches and front office personnel will be able to make objective decisions better than they ever have. 

Beyond that, I’d like to discuss a few other frontiers.  I’ve said that we’ve made a lot of improvements when it comes to measuring individual defense, but there is still progress to be made.  Play-by-play data has improved things considerably, but it’s still incomplete.  We can figure out which players are on the floor at all times, so we can guess who’s guarding who.  But we never know for sure, and that poses significant problems when trying to calculate a player’s impact. 

I e-mailed Bradford Doolittle and Kevin Pelton of Basketball Prospectus after they released a lot of their advanced stats recently, and one of the responses I got back was that they needed this sort of data.  Based on the way games are currently logged, this data is unavailable.  You could monitor these things on video if you took the time to look.  But it’s simply not possible to watch and log every NBA game that’s ever played unless you have a very large group of people doing it.

Besides exact matchup data, game tracking such as the work Schuyler Davis is doing would be very useful.  That’s because defense is very much a team thing.  We need to know who rotates well, who draws charges, who forces turnovers by reading the offense, etc.  I would guess a lot of teams record this kind of information about their own players, but I doubt anyone possesses this information about every player in the league. 

Beyond defensive stats, I think the college game is another relatively undiscovered area.  We do have some advanced stats about college basketball such as at  However, it still pales in comparison to what we know about NBA players.  It’s not just knowing how players are performing in college.  Eventually we’ll know what stats in college are transferrable to the pros.  This kind of information would be invaluable to college scouts.

In the future I personally will try to explore some of these ideas.  Some of them (college stats, player interactions) are doable, while others (matchup data) are currently not.  It should be fun trying to tackle these statistical questions.





Copyright © 2009