Friday, September 17, 2010

What I'm Using: Team Metrics

Previously:
What I'm Using: Individual Offensive Metrics
What I'm Using: Individual Defensive Metrics
A Note About Points Per Attempt

Offensive and Defensive Efficiency (OEff and DEff)
The efficiency stats are simply a measure of how many points a team scores (or allows) per 100 possessions. Ken Pomeroy shows us how to approximate possessions:

FGA-OR+TO+0.475xFTA
I’m not really sure why ‘per 100 possessions’ became the norm; it must be some sort of metric thing. If you want points per possession just move the decimal point two spaces to the left. These are probably the best ‘first glance’ stats to look at when figuring how two teams will match up. Here’s how the MIAA shaped up last year in terms of team offensive and defensive efficiencies (sorted by OEff):

TEAMOEffDEff
Hope106.897.8
Olivet104.899.6
Calvin103.797.8
Albion99.895.7
Adrian98.195.2
Alma97.2112.0
Trine93.9103.1
Kalamazoo89.8108.5
Average100.0100.0

I fudged the average numbers a little bit to make them nice and round. Actually, I averaged the average OEff and DEff and then fudged it to be 100 even.

These numbers give us a much better picture of each team’s offensive and defensive abilities than points per game does.

As an example of this, let’s compare last year’s Alma Scots squad with the Trine Thunder. Alma scored 63.5 points per game, which was 1.2 points worse than Trine, but they played a slow four corners style offense which limited possessions (both for themselves and their opponents). The tempo-independent efficiency numbers show that Alma’s offense was considerably better at turning possessions into points than Trine was.

The difference between PPG and OEff (or DEff) would be even more pronounced for teams that run the “Grinnell system,” which, contrary to popular belief, is not very fun to watch.

Shooting, Rebounding, Turnovers, and Free throws (eFG%, TO%, ORb%, FTr)
Statistician Dean Oliver contends that these rates, the four factors, contribute most to a basketball team winning or losing (listed in order of importance.)

Effective field goal percentage (eFG%) is calculate the same way as it was at the individual level: [FGM + 0.5 x 3PM] / FGA. Again, all this does is account for the fact that three pointers are worth an extra point.

Turnover rate (TO%) is simply the rate at which possesions end in a turnover. TO / Possessions.

Offensive rebound percentage (ORb%) is calculated by dividing a teams offensive rebound total by the number of possible rebounds. TeamORB / [TeamORB + OpponentDRB].

Finally, free throw rate (FTr) is the number of times a team gets to the line per field goal attempt. FTA/FGA. Oliver suggests that while free throw percentage is important, the number of times a team simply gets to the line is the more variable (and valuable) skill.

Here's how the four factors compared last season's MIAA teams (sorted by eFG%).

TEAMeFG%TO%ORb%FTr
Hope0.5200.1780.3050.331
Albion0.5030.2230.3170.407
Calvin0.4990.1970.3230.457
Olivet0.4890.1700.3310.301
Trine0.4760.1990.2640.293
Alma0.4660.2250.3420.300
Adrian0.4370.1860.3480.362
Kalamazoo0.4280.2150.3310.291
Average0.4850.2000.3200.360

Please note that Calvin was the only team to be league average or better in all four categories.

These same four factors can (and should) be applied to defenses as well. A great second level comparison of two teams can be done with comparing Team A's offense with Team B's defense (and vice versa) with the four factors.

Here's a look a last season's defenses:

TEAMDeFG%DTO%ORbA%DFTr
Olivet0.4640.1800.2860.385
Kalamazoo0.4780.2320.3600.352
Adrian0.4790.2240.2860.443
Hope0.4800.2180.3040.372
Calvin0.4890.1750.2760.267
Trine0.4920.1970.3290.372
Albion0.5050.1920.3150.469
Alma0.5320.1910.3820.314
Average0.4850.2000.3200.360

Here we see Calvin fall below the average mark in DeFG% (barely) and DTO% (significantly), but they made up for it by dominating the defensive glass and preventing opponents from reaching the free throw line. No team was above average in all four defensive factors.

I think that's all I have to say on team metrics for now. When comparing teams for game previews and the like, these stats will be pretty much all I'll use. I may eventually toss in team versions of the Points Per Weighted Shot, Block Rate, and Assist Rate stats, but for now I'll leave those alone. We'll see how that goes.

No comments:

Post a Comment