I have been an admirer of the increased utilization and availability of advanced statistics in the NBA. I have tried to the best of my ability to incorporate them into my analysis of games in a relevant way. From time to time I have wondered about stats I would love to see such as assists at the rim as opposed to assists on that come from a long jumper or how many calories I burn yelling at the television.
The other night I was watching the barn burner between the Golden State Warriors and the Phoenix Suns and witnessed Anthony Tolliver play nearly every single second of the game, 47:28 to be exact. The thought struck me that 47:28 during that game, where the two teams combined to score 234 points, Tolliver was responsible for covering Amare Stoudemire, carried a significant load of his own team’s offensive burden and was bammed on like few have been bammed on before, was much more grueling than 47:28 of a game between to slower paced teams.
I wondered instead of talking about simply how many minutes a player played in a game, why not look at how many possessions he participated in? Instead of simply tracking who played the most minutes per game or in a season, why not determine who played in the most possessions? I am not sure if anyone has asked this question. As far as I can tell no one has in the manner I am suggesting it.
A simple way to determine possessions could be to divide pace factor by the percentage of minutes played. Using the formula 0.96*(FGA +(0.44*FTA)+TO-ORB) to determine pace for the Suns/Warriors game we arrive with a pace factor of 107.7. Tolliver played 47.42 minutes so he was on the floor for 106.5 possessions.
There are two problems with this idea as I see it. First, certain players or combination of players play at a faster pace than others and simply dividing by minutes is not necessarily an exact determination of possessions every player was on the floor for unless like Mr. Tolliver, they were on the floor for nearly all of them. Secondly, is participating in fewer possessions necessarily more work than having to defend for 16 or 18 seconds every possession even though you may have played in fewer possessions? After all not every player runs the floor during a fast break, but conversely at any one time there are two or three players standing around during a defensive possession and not exerting any energy either. The consensus seems to be that playing at a fast pace is far more strenuous than playing at a slow pace regardless of the quality of defense that is played. The Nuggets have relied on that fact to dominate at home for years.
Some could laugh at my little formula as there are legitimate advanced statisticians who probably already know exactly how many possessions a player participates in during every game. That information is required in order to calculate on and off court stats such as offensive and defensive efficiency ratings by player. As I mentioned above, I suspect no one has really cared much about documenting possessions per game or possessions per season. Let this be the call to look at possessions instead of just minutes and games as a measure of longevity or current service time.
What kind of information could we figure out based on possessions played instead of minutes played? Currently Gerald Wallace leads the league in minutes per game at 41.8 and Monta Ellis is second at 41.4. These two provide a perfect example for how pace can show the difference in the disparity in the possessions these two actually play every game. Using John Hollinger’s team stats the Bobcats are one of the slowest paced teams in the league with a pace factor of 92.9 while Golden State leads the league in pace at 102.7. When we calculate the possessions they would have participated in based on the minutes they played we see that while Wallace was a part of 80.9 possessions, Ellis easily surpasses him at 88.6 possessions per game despite playing four tenths of a minute less.
Apart from trivial things such as who plays more possessions could there be any value to it? Is that kind of information really significant? Bill Simmons talks about the anecdotal evidence that many players lose their legs after playing in 1,000 games. Could that number be lower or higher based on the style of play that player has been a part of during his career? What if a player plays a significant number of seasons on a fast paced team?
We have seen how a fraction of a minute can make a big difference in possessions per game with our example of Monta Ellis and Gerald Wallace. Over the course of nine or ten games Ellis will compile an entire game worth of possessions above Wallace. Over the course of a season Ellis might play in enough additional possessions equal to eight or nine more games than Wallace. Playing for Golden State for seven or eight seasons could possibly take nearly a full season off of a players’ career in additional wear and tear.
That is something a GM might be interested in knowing, for all I know maybe they already do. If not, maybe this can be my little contribution to advanced statistics.