Tempo Free? Importance of efficiency to Groce, Illini

Tempo Free? Importance of efficiency to Groce, Illini

A look at what John Groce means when he talks about efficiency.

By now, you've no doubt heard Illini basketball coach John Groce talk about "efficiency" when he discusses how well his team is playing.  You'll often see writers mention "tempo free" statistics when talking about how efficient a team is, both on offense and defense.  These types of statistics have been around for a long time.  Dean Smith, the hall-of-fame coach at North Carolina, was the first to really use stats like points per possession to measure how well his teams were playing.  Smith started asking the question, "how well are we playing on each possession?" to try and make his teams more efficient (and to keep the opponents from being efficient) on offense.  Because tempo in each game can fluctuate wildly (picture playing Indiana vs. playing Wisconsin), Smith started looking at his team's performance on a per-possession basis.

If you're looking for a good primer on what tempo free statistics you'll frequently see discussed, here is a good one from the folks at the Michigan State fan site theonlycolors.com.  It's appropriately titled "Tempo-Free Stats for Dummies," and gives a good summary of what stats are used, why they're used, and some good benchmarks to separate good performance from bad.

Finished reading?  Good.  As KJ mentions in that article, there are some really good resources online if you'd like to keep up with how the Illini are doing in the "tempo free" world.  One of them happens to be from a University of Illinois alumnus, John Gasaway.  He's been working with tempo free stats for quite some time, starting with the Big Ten Wonk website several years ago.  If you were able to follow along with the "Dummies" article above, then take a look at John's more detailed look at tempo-free stats, "This is TFS: tempo-free stats."  He's since moved to Basketball Prospectus, where he writes regularly.  Much of his work requires a subscription (totally worth it, by the way), but he has a weekly column called "Tuesday Truths" that is available for free.  Here's the most recent edition, which he posted this Tuesday 2/12.  Some of Gasaway's columns are also available at ESPN.com, but only for "Insiders."

The other guru of tempo free stats is Ken Pomeroy, who you can find at his website kenpom.com.  His home page gives you a snapshot of rankings for all of the teams in NCAA Division One.  As of today, after the Purdue win, he has Illinois ranked at #32, which is up nearly ten spots from where they were before last night.  Here's his brief explanation of his ratings system.  I suggest you read the other primers above before diving in too deeply.  Full disclosure: when he starts talking about things like "log5 formula", I tend to go a bit crosseyed.  But, for the more mathematically inclined among you, enjoy yourselves. 

Blog Photo

 

 

Meanwhile, here's a look at his rankings as of this morning.  He's got Florida as his best team as of today, with Indiana #2.  Just click to enlarge the screen shot, or you can go to kenpom.com to see rankings of all division one teams.

 

 

 

 

 

Blog Photo

 

As for how the Illini are faring in Big Ten play, the aforementioned "Tuesday Truths" from John Gasaway provide thebest snapshot there, and it's useful to see how the teams are faring in conference play, since non-conference schedules can vary so wildly.  Note that these rankings were done before the blowout win by Michigan State over Michigan, and before the Illini's destruction of Purdue last night.  Again, just click to enlarge, or click the link mentioned above to see John's complete article.

 

 

 

 

Blog Photo

UPDATE:  Thanks to regular HQ reader JimOATSfan, who pointed out a great stats site below in the comments, I'm able to update John Gasaway's Big Ten efficiency stats through games of last night.  If you compare Illinois' numbers from Tuesday to their numbers today, you can see that the Illini have made progress.  By the way, I'm not sure I mentioned this, but the EM you see in both John's article and in my table below stands for "efficiency margin," which is simply the difference between a team's Offensive Points per Possession and their Defensive Points per Possession.  You can see that the Illini offense has been slightly above average in Big Ten play, while their defense has been below average, though both numbers have improved a lot in the last week.  Click to enlarge.

 

 

 

Now, let's take a look at how the Illini did in these tempo-free stats in their game last night vs. Purdue.  As you might expect in a blowout win, Illinois did pretty well, but let's take a closer look.

  • Illinois PPP (points per possession): 1.19 (1.01 is average in the Big Ten this season)
  • Illinois effective FG% (takes into account higher reward on 3pt shots): 47.7% (average is about 49%)
  • Illinois turnover percentage: 16.5% (average is about 20%)
  • Illinois offensive rebounding percentage: 50.0% (average is about 33%)

Now, how did the Boilermakers do in their tempo-free stats?  Um, not so well.

  • Purdue PPP: 0.90
  • Purdue eFG%: 43.6%
  • Purdue turnover percentage: 25.9%
  • Purdue offensive rebounding percentage: 36.8%

Illinois did a great job of taking care of the basketball and of getting offensive rebounds last night, giving them many more possessions than Purdue had last night.  When you add in the Illini's four-percent edge in effective FG%, you can see how the Illini were able to win this game comfortably.

So, does any of this make sense?  Do you care?  I'll try to post a quick tempo-free wrapup after each Illini game the rest of the season.  Looking at these stats from time-to-time has given me a bit better understanding of the "why" and "how" behind some of the results for the Illini this season, and I hope it can do the same for you.

Of course, stats are only part of the story.  No stat can be as much fun as this:

Or this:

Thanks for reading.  Talk to you soon.

bmoline@wdws.com

Comments

News-Gazette.com embraces discussion of both community and world issues. We welcome you to contribute your ideas, opinions and comments, but we ask that you avoid personal attacks, vulgarity and hate speech. We reserve the right to remove any comment at our discretion, and we will block repeat offenders' accounts. To post comments, you must first be a registered user, and your username will appear with any comment you post. Happy posting.

Login or register to post comments

JimOATSfan wrote on February 15, 2013 at 6:02 am

Hey Brian:

Very interesting post.  I saw Tempo Free stats for the first time earlier this month on The Champaign Room blog for the Illini on SB Nation.

Just two days ago I discovered this STAT filled website: http://statsheet.com/mcb/teams/illinois

Again stats like I've never seen them before. Thanks for the primer on efficiency & TF.

Cheers!

Brian Moline wrote on February 15, 2013 at 9:02 am
Profile Picture

Hey Jim,

Thanks for the feedback, and thanks for the link to that stats site.  That will be helpful.

Cheers!

Brian

Update:  It's already been helpful!  I was able to update the Big Ten efficiency stats through games played last night thanks to that link!

IlliniMike05 wrote on February 15, 2013 at 11:02 am

Maybe it's just my generation, but a beginner's guide to efficiency stats in 2013 seems a little odd. This stuff has been pretty mainstream for awhile.


The thing I love about basketball stats, though, is that they're easy to comprehend. As much as I love baseball sabremetrics (and I really, really do; I think the world can be divided into two groups of people: those who understand why Mike Trout should have been a unanimous MVP, and knuckle-dragging Neanderthals), I realize that things like WAR, wOBA, OBP+, ERA+, xFIP, etc, as much as they give us an infinitely better gauge of value than traditional, terrible baseball statistics, aren't intuitive or easily digested for all.


But advanced metrics in hoops? Extremely intuitive and easily understood. It takes the numbers we're already familiar with and actually puts them into proper context via removing pace. (My only gripe is that they're referred to, even by their inventors, as "tempo"- free stats, when pace is the word they're looking for. A team can play at an extremely brisk tempo while playing a low number of possessions per game. See this year's Michigan team or the '04-'05 Illini for perfect examples.) Converting rebound totals into percentages of rebounds grabbed on each end of the floor? Gives us a much better idea of what happened on the boards and is easily understood by anyone with a working brain. Same with turnover percentage, assist percentage, and anything else. Instead of looking at a million little incomplete details that add to the picture but by themselves don't tell us enough?* That's what we have adjO and adjD for. Now we know exactly how many points per possession a team is scoring, and allowing, and how that stacks up nationally. So instead of looking at a team like Michigan and coming up with subjective bromides for why they may not be able to win it all, we can simply look at their relatively mediocre defense and know that since Kenpom started compiling these numbers in 2003, a team that allows as many points per possession as Michigan hasn't won it all.


*I remember reading Lou Henson's book as a kid where he talked about the two main numbers he judged his defense by were points allowed and- more importantly- field goal percentage. Even as a kid, before anyone knew what tempo-free stats are, I thought this was insane. Field goal percentage? What about FT trips allowed, turnovers forced, defensive rebounding percentages and 3pt defense? Not to rip on Lou because no one knew this stuff that much back then. But I'm so glad to have a coach in Groce that has a high-level understanding of this stuff that he actually puts into practice. It's one of the main reasons I was so on board with him being hired even before he emerged as the top candidate.


 


 


 


 

Brian Moline wrote on February 15, 2013 at 3:02 pm
Profile Picture

Mike, you're right.  This stuff has been out there for a while now.  For whatever reason, no one has really been writing about it regularly here, so I thought I'd take a stab at it.  I'm a little late to the party, as I really started looking at the efficiency stats about 4-5 years ago.  Totally agree with you that they're easier to follow on the basketball side than on the baseball side...much more intuitive.

Thanks for reading.  I appreciate the feedback!

Brian

IlliniMike05 wrote on February 15, 2013 at 4:02 pm

It was a really good writeup, don't get me wrong. Sorry if I came off like a jerk. I probably take it for granted because I've been following this stuff for a long time. It was such a revelation for me because it put into words (and numbers) a lot of concepts I was sure plain stats weren't giving us all of. So hopefully the light turned on for some people reading this as well.


One thing I disagree with is about being "late to the party" on it. No such thing, in my opinion. Any time you can expand your knowledge about basketball (or, well, anything) is good timing. And really, I don't remember this stuff becoming widely available on mainstream sites like ESPN.com until a few years ago, so I definitely wouldn't say you were late on it.


Definitely good lookin' out on getting the advanced metrics some space on IlliniHQ. For those that don't follow numbers obsessively, it was an excellent tutorial.

Brian Moline wrote on February 15, 2013 at 5:02 pm
Profile Picture

Mike,

Not at all.  Thanks again for reading, and I hope to make this a regular feature.  Now that I've got the basics out of the way, I plan on having a short write-up after each game talking about the efficiency stats.  I'll probably have the game stats up the same day/evening, and update the year-to-date stats the following morning.

aaeismacgychel wrote on February 15, 2013 at 2:02 pm

I'm a numbers guy and I've enjoyed following the work of Pomeroy, Sagarin, and Gasaway for years. In fact, I'm currently a Pomeroy subscriber and have used his site fairly readily for the past 8 or so years. He brings a very unique perspective to college basketball and I highly recommend a subscription as he has detailed efficiency and statistical information on every player and game. And the statistical analyses covered in his blog are tremendous. Just a bit of a word of warning though for anybody not really all that knowledgable of statistics or advanced metrics- All systems have their positives and drawbacks, and it's good to know what these are in each system before relying too fully on them. For example, RPI and standard SOS unfortunately are dragged down by the worst teams you play and effectively those values have higher weight simply because they're so low (For example, there is no difference between playing 4 10-10 teams and 2 19-1 + 2 1-19 teams which isn't true at all) and they don't take into account road vs. home. Efficiency based statistics on the other hand are almost the opposite in that they are biased in favor of teams who blow out cupcakes in the non-con schedule. Beating a bad team 94-30 for example is much better than beating the #1 team in the country 72-70. So you just need to know what you're getting into ahead of time. It's a lot of fun to play around with. Just so you know where Illinois ranks in the B10 right now (from Pomeroy):

Offensive stats:
Efficiency- 1.02ppp 6th (Indiana 1.16ppp 1st)
Effective FG%- 46.3% 7th (Indiana 54.9% 1st)
Turnover %- 17.6% 7th (Michigan 15.0% 1st)
Offensive Rebounding %- 34.0% 4th (Minnesota 41.7% 1st)
Free Throw Rate (FTA/FGA)- 0.276 11th (Indiana 0.511 1st)
2pt FG%- 48.5% 4th (Michigan 51.4% 1st)
3pt FG%- 28.6% 9th (Indiana 43.4% 1st)
Free Throw %- 76.0% 1st

Defensive stats:
Opp Efficiency- 1.05ppp 9th (Wisconsin 0.92ppp)
Opp Effective FG%- 50.5% 12th (Wisconsin 42.6% 1st)
Opp Turnover %- 21.0% 1st
Opp Offensive Rebounding %- 35.1% 11th (Wisconsin 27.3% 1st)
Opp Free Throw Rate- 0.436 11th (Michigan 0.230 1st)
Opp 2pt FG%- 47.5% 8th (Wisconsin 43.5% 1st)
Opp 3pt FG%- 37.9% 12th (Wisconsin 26.3% 1st)
Block %- 9.3% 5th (Minnesota 16.7% 1st)
Steal %- 10.8% 3rd (MSU 13.4%)

Brian Moline wrote on February 15, 2013 at 3:02 pm
Profile Picture

Thanks for the detailed comment with the Illini ranks.  Much appreciated! 

As you mentioned, all ranking systems have their drawbacks.  One thing many have noticed with Pomeroy's rankings is how well Wisconsin ranks relative to their W-L totals every season.  Much of that is their style of play, which lends itself well to these efficiency stats, though this season their defense has been outstanding, even better than usual.

 

IlliniMike05 wrote on February 15, 2013 at 5:02 pm

That's a great point about their style of play. Their consistently high Kenpom rankings is greatly helped by their consistently excellent defense (aside from 2011, when they bizarrely ranked No. 56 in AdjD, they've been in the top 20 pretty much every year since he started his site), but I do think their pace puts their efficiency numbers a bit beyond their actual performance. I said in another thread earlier this week that I don't think there's anything that brilliant or intricate about their offense. They just find guys that fit a system that's played at a pace that bores most teams to sleep. The numbers don't lie in terms of efficiency, but as consistently good as they are I do think they're the type of team these numbers overinflate.

IlliniMike05 wrote on February 15, 2013 at 4:02 pm

"Beating a bad team 94-30 for example is much better than beating the #1 team in the country 72-70."


Before I start, I wanna say this was a great post, but to nitpick I wanted to adress that quote; it's not that simple, because as I'm sure you know offensive and defensive efficiency ratings- at least Pomeroy's, which are still the gold standard in my opinion- are adjusted for schedule, hence why they're listed as "AdjO" and "AdjD". As I'm also sure you know, Pomeroy doesn't even list the regular, unadjusted ratings anymore.


But, yeah, you can end up with high efficiency ratings against bad teams that doesn't end up holding up. More often than not, though, those numbers end up being pretty effective in a predictive capacity. Take Pitt, for example. They got some buzz from the wonk community early on, rating highly in AdjO and AdjD despite not beating any top-level teams, then were scoffed at after starting slowly in Big East play. Now here we are with the majority of the regular season in the books and they're 20-5, still rated in the top 10 in AdjO and AdjD and No. 4 overall in Pomeroy's ratings.


Or take Illinois. Many were rightly skeptical after the 12-0 start because we still rated, at best, in the high 20's in Pomeroy's rankings. After a rocky January, we're now pretty much right where the numbers said we should be all along.


Pomeroy, Gasaway and the rest of their wonk cohorts would be among the first to say their numbers aren't perfect, but they work a lot better as both a predictive model as well as an evaluation of what's already happened than anything we had before.


So I think it's a *bit* oversimplified to say that efficiency stats are biased towards those who blew out bad opponents, but I get what you're saying about the minor flaws they have relative to the major flaws RPI has. For those that think margin of victory doesn't matter, you might not love efficiency ratings.

aaeismacgychel wrote on February 16, 2013 at 10:02 pm

Granted.  I was mainly trying to make up an absurdley large score differential (maybe not so absurd, I was really pointing to Kentucky's game against Eastern Michigan though I misremembered the score a bit) to give a layperson an idea of what issues there are in Pomeroy's system (which I agree with you in it being arguably the best out there) which would still ring true (and that scenario would).  Technically, as you well know, it's all about how efficient you were against a given opponent as compared to how their average opponent performed against them in similar environments over the course of the season.  SOS of course is a factor that helps downgrade weaker schedules and raise rougher schedules, but when you think about a 90-30 blowout, the lesser opponent from a lesser conference will play much more competitively in conference play (more towards their average) and similarly the stronger opponent will play more towards their average in conference.  Thus, blowouts of that magnitude are rare and hence outliers.  Can it be possible that a 1 posession victory at home over a top opponent is a bigger boost for a team than a 60 point victory at home over a lesser team in Pomeroy's system?  Sure, it's possible if that top ranked team is absurdly good and that lesser team is bad in pretty much epic fashion, but in almost all cases, the blowout serves as a huge outlier that simply gains much higher weight because it's so far off the average.  A good example of this is that Illinois' big win over Purdue gave them almost the same boost in rankings as the combined wins vIndiana and @Minnesota did.

The reason I even brought it up though is to help explain why a team like Kentucky has an RPI of 41, an SOS of 41, and a Pomeroy rating of 18 while Illinois has an RPI of 28, an SOS of 6 and a Pomeroy rating of 33, all while Illinois is 5-3 against Top 25 competition whereas Kentucky has only beaten a decent but not close to Top 25 Ole Miss team.  One other difference being that Pomeroy includes preseason ratings in his formula worth a couple games values- it is statistically relevant as he showed, but something that would make the average person scratch their heads and question it).  The point is, all advanced analysis systems should be analyzed for their respective faults and all should be considered instead of thrown out.

In fact, while RPI does have "major faults" statistically speaking, it along with SOS probably still does the most decent job of analyzing teams based solely on resume, and I think comparing it to Pomeroy's or Gasaway's or Sagarin's stats and coming to the conclusion that it is "archaic" does a disservice to all methods simply because I think they look at completely different aspects to the same problem.  RPI is a good indicator of what teams have accomplished (again with the fault that it can be manipulated by playing a bunch of 3rd-5th ranked teams in lesser conferences than bottom teams in those conferences even though neither poses any realistic risk to win whatsoever) while Pomeroy is an excellent predictor of head to head performance and who the "best teams" are.  It's the age old who's the best vs. who's the most deserving argument.  There's no right or wrong necessarily, just two sides of the same coin.  Thank you for the response though and good discussion- Seems like those Illinois degrees are worth a little bit after all ;-)  What's your background by the way?  Engineer here.

Brian Moline wrote on February 17, 2013 at 6:02 pm
Profile Picture

Thanks to both of you for getting a nice discussion going here in the comments.  That's exactly what I was hoping this would do.  Hope to see it continue when I post about the Northwestern game later tonight!

IlliniMike05 wrote on February 19, 2013 at 10:02 am

Oh, don't get me wrong, I do agree that RPI still has a legit place in the discussion. It's still probably the best tool for grouping teams as a basis of comparison and as you said, looking solely at resume; Pomeroy's numbers do a better job of gauging true performance, but RPI is still a very useful tool for grouping and gauging actual results, which is what I meant by RPI having "major" flaws opposed to Kenpom/Gasaway/etc; I find measuring actual performance more telling than raw results that can fluctuate wildly from performance based on nothing but luck a lot of the time.  As we know, teams can sometimes wildly over/underperform in terms of W-L compared to their statistical resume/Pythagorean record/etc. (Case in point from another sport: 2005 White Sox. On the surface a dominating 99-win season in which they had the best record start to finish, capped off with one of the all-time dominating playoff runs in any sport. But with an OK-but-not-great run differential of 96 on the season, their Pythagorean record was merely 91-71. So that dominating season should've, theoretically, ended with them not making the playoffs.) But at the end of the day, for tourney purposes it doesn't really matter if you don't get the W.


Strangely enough, I have a journalism degree that I haven't used all that much. I have no real background in mathematics whatsoever. If I could do it over again, especially knowing what kind of place advanced math would have in the future of sports journalism, I very likely would have pursued something in that field instead. I was something of a math prodigy up until I was 12 or 13 and just totally stopped caring about it when I got to high school and decided I wanted to be a writer. I've always been obsessed with statistics (mostly in sports, of course). I wish I was making this up, but even all the way back to Little League I kept track of my on-base percentage instead of batting average and, I swear this is true, kept a spray chart of my hits.


So, yeah, I wish I'd stuck with something in math in hindsight.