Sports Reference Blog
Introducing Per 100 Possessions Statistics
Posted by sean on June 17, 2014
If you've been browsing recently, you've may have noticed that we added per-100 possession statistics tables to the player pages and team pages (and for both the regular season and playoffs). For now, we're able to estimate team possessions going back to the 1973-74 season.
We also have league-wide tables that correspond to those on the player and team pages, with stats for every player, and even totals for all individual team stints for those players that were traded, re-signed, etc. Sometimes users tend to not to find these but they're extremely useful for sorting or exporting and doing your own analysis.
Here's the easiest way to find those tables: Mouseover the 'seasons' link and click on Per Poss on the dropdown menu:
(Similarly, these are available for the playoffs, as well - corresponding links can be found in the 'playoffs' site section)
Per-Possession statistics are an easy way to nullify the distorting effects of pace and compare players more easily across teams and eras. For the record, we estimate league pace going back to the 1973-74 season using the formula provided in our statistics glosssary, and estimate it for individual teams going back to 1950-51 by estimating component statistics.
Love this feature! Would have saved me a lot of research time over the years!
I just counted and LeBron's page now has 17 tables. Wow. How many more tables are you going to add?
Tremendous work, thanks as always! A few questions:
1) Do you plan on expanding this to pre-73-74 seasons and playoffs soon, since you already calculated pace/ORtg/DRtg going back to 50-51?
2) Will you calculated Pace/ORtg/DRtg for pre-73-74 playoffs, and produce the same per 100 possessions numbers?
3) Are you considering adding per 100 possessions numbers to the Play Index and "leaders" tab (you recently added 2P%, which was another great category, I'm wondering if that'll find its way to "leaders" as well)?
Thanks as always for your hard work and dedication!
@Frank -
1 & 2) I hope to. It would be really useful to extend this back two decades, but probably won't be able to do this soon.
3) Definitely hope to integrate this into the PI and leaderboards, yes
Very call David, thanks for the response.
These numbers are pretty fun to play around with, thanks again for adding them.
I'm not criticizing the addition, but applying team pace factors to individual performances is utterly fallacious, hence the primary reason why PER is a fraud. According to per-100-possession statistics, for instance, Chris Paul is superior to Magic Johnson. The reality, however, is that the relationship between team pace factors and individual performance is not linear and can actually be inverted. At a faster pace, there are more possessions, but a player cannot dominate and milk the possessions to the same extent. Conversely, at a slower pace, there are fewer possessions, but that slower pace better enables a star to milk and manipulate the possessions. Thus Michael Jordan, Karl Malone, and John Stockton posted historic numbers while playing at some of the league's slowest paces, and thus playing at a slower pace can inflate Chris Paul's numbers at least as much, and perhaps more so, than playing at a faster pace.
Any chance you can estimate (or give out the formula) for NCAA basketball team possessions?
@KJOK -
0.5 * ((fga + 0.475 * fta - orb + tov) + (opp_fga + 0.475 * opp_fta - opp_orb + opp_tov))
David:
Thanks. For older years, NCAA data only has TOTAL rebounds instead of Offensive and Defensive rebounds. Is there a decent way to modify the formula when offensive rebounds are not available, and still get a reasonable estimate?