RealGM

All teams play at an average pace throughout the course of the season. However, some teams are better at controlling the pace of the game than others, playing to a speed and style that ultimately suits them. I'd like to capture this effect statistically. One way I've thought the output might look is when trying to predict the pace of a game, one team's (the better, controlling team) average pace will get weighted higher than the other, lesser team.

So a prediction for the pace of an ATL-IND game might look like:

--IND---|--ATL---|--Game
(.6)(93.4) + (.4)(95) = 94.04

But how to capture those weights?

One way I've thought about is checking variation from a team's average pace, and then indexing every team in the NBA's variation from their own average pace on a 1-100 scale. The teams with a tighter variation are the ones who impose their own style more often, and will have a higher score. When it comes to getting the weights, you simply add up the two index numbers, let's say, for the sake of argument that Pacer's score 100 and the Hawks 50, and you divide each's score by the total combined score. In this case, it'd be the Pacer's .667 and the Hawks .333. These would be the weights.

Would that be accurate??

Another, related question. Is there a difference between "team pace" and "average game pace?" For example, if I average the pace for all games in a teams advanced game log from b-r (http://www.basketball-reference.com/teams/ATL/2014/gamelog/#tgl_advanced::none) it comes out differently than the pace rankings.

Look forward to replies.

You are going to want to look at offensive and defensive efficiency . The wider the spread between them, the less reliable PACE is as a baseline. And bad data in, bad data out.

Also, you need to consider margins. Teams that play from behind speed up, teams comfortably ahead tend to milk the clock. See Miami and its 23rd rated PACE.