Saturday, August 8, 2009

Dove Entrails, Tea Leaves and Returning Minutes...and Pythagoras

So I Was Surfing the Web, Minding My Own Business...
Ok, not really. I logged into VBTN late last week and found a link back notation from the guys over at the UM Blog, who (as desparate for hoops discussions as I am) decided to apply returning percentages to the Big Ten teams to see if anything interesting shook out. The tidbit that peaked my interest (and inspired this post) was their reference to the guys over at The Only Colors Blog (they are a Michigan State blog) who crossed each (Big Ten again) team's returning minutes (percentage) with their previous season's Pomeroy rating (the team's Pythagorean Winning percentage actually) to produce a pretty interesting scatter chart, which plots the conference's eleven members on an X (returning minutes) and Y (Pomeroy's Pythagorean-based rating) Axis. Their chart may be problematic since they used Pomeroy's data for their returning minutes (a bit incomplete, given that Pomeroy does not include players whose %Min are <10%). The idea for the chart came from this diagram over at the Big Ten Geeks Blog, a blog I added to the blog roll last August (when they started up). The Geeks are on hiatus for the summer, but hopefully they will be back for this season. Where did The Only Colors get the idea of crossing returning minutes with Pythagoras? I can't be certain, but on my way over to see their scatter chart I found a reference at Rival's The Dagger Blog (his Extra Onions series, a kind of "walk around D1 basketball" feature) that contained a link (embedded in a rant) to the Vegas Watch Blog's post on using Pomeroy's Rating with returning minutes to project that team's rating for the next season.

A Reliable Metric?
Vegas Watch regressed the previous season's Pomeroy Ratings and returning minutes from every team affiliated with the six major conferences, and determined there was a statistically significant relationship -- strong Pomeroy Ratings and high returning minutes generally translated into strong Pomeroy Ratings for the next season. There are of course outliers, but not many. The Only Colors' idea seems like a good one, and frankly I was interested to see how the conference would align, but before I looked at 2009 numbers, I decided to look at the Pythagorean percentages and returning minutes for 2008. The quadrants were organized like this:

Pomeroy: Below Average
Ret. Mins: Above Average
Pomeroy: Above Average
Ret. Mins: Above Average
Pomeroy: Below Average
Ret. Mins: Below Average
Pomeroy: Above Average
Ret. Mins: Below Average

It was not a neat, 4 team per quadrant breakdown, but the breakout was interesting. Not just for those predictable (good teams in the good quadrant, bad teams in the bad quadrant) team, but also for the outliers, those teams that seemed to run against the assumptions. I ordered the teams within each quadrant by highest Pomeroy Rating.

Seton Hall
Notre Dame
De Paul
South Florida
St. John's
West Virginia

The X-Y intersection (average for returning minutes & Pomeroy Rating) is 0.689 and 0.8361. Finding De Paul, South Florida and St. John's in the lower left quadrant is not surprising. Even finding Cincinnati in that company is only a little surprising, since the Bearcats' conference record had not changed from 2008. Providence registered a good improvement over their 2008 finish (12th up to 8th), while Seton Hall compiled an identical record (& finish -- 11th). Though Rutgers had a very high "return" rate (about 89%), the Scarlet Knight's Pythagorean Winning Percentage was (by far) the worst in the conference (0.495). The Scarlet Knights actually lost ground in the conference win column (3 down to 2), but finished one place higher (16th to 15th). The "blue" quadrant looks like a winner, except for Georgetown. The Hoyas' 2009 results seem to conflict with the other teams' results, until I compared returning minutes. Both Pittsburgh and West Virginia returned over 60% of their 2008 minutes (66.3% & 61.0% respectively). Georgetown returned 43.8% of their minutes. If I had an nickel for every analyst who projected the 2009 order of finish according to the "green" quadrant I could probably afford a better skin for VBTN (or at least a nicer vacation destination next summer). But Notre Dame will drive us all to the poorhouse (if we gambled and if we bet on Notre Dame). Of the teams in the "green" quadrant, Villanova returned the highest percentage of minutes, which might explain how the Wildcats finished above the Orange. Marquette's loss of Dominic James is a reminder of how a season can turn on a critical injury.

How About 2010?
Mapping 2010 returning minutes (at least what we know so far -- see Disclaimers... below) with 2009 Pythagorean Winning percentage (from Ken Pomeroy's Big East Page -- many thanks to Ken Pomeroy for making this data available) yields a scatter chart that looks like this:

X-Y intersection in this data set is 0.583 and 0.8088 -- a modest drop in both measures. Declining returning minutes makes sense when you consider the graduation of seniors and the exodus of NBA talent. The drop in the Pomeroy Rating is more paradox than contradiction. The standard deviation for Pomeroy Ratings in 2008 was 0.1407, for 2009 it was 0.1789. The teams at the top of the conference pounded the teams at the bottom. 75% of the teams sorted out to the "blue" and "yellow" quadrants this time. A hint that next season's conference will be more competitive than last season's? Most likely I suspect. Taking up sole residence in the "red" quadrant suggests a long season ahead for Providence fans. Among the "yellow" quadrant teams, Seton Hall and Cincinnati are positioned, relative to the Y Axis, much like Providence was going into last season. Upward mobility in the conference may be in the offering for both. St. John's has been mentioned as another candidate for upward mobility. I confess I am not sure what to make of the Johnnies' location on the chart however. Rutgers and South Florida track very closely, can a large separation in finishing ranks next season be chalked up to coaching? Like St. John's, De Paul sits in an "outlier" location on the chart. Rutgers had a similar spot (low Pythagorean WP, relatively high returning minutes), which translated into a slight rise in rank, but a decline in wins.

Like Seton Hall and Cincinnati, Louisville sits in a "blue" quadrant spot similar to where Pittsburgh was going into last season. The Panthers finished in a 1st place tie with the Cardinals in 2009 and ran to the Elite Eight. I suspect, should it come to pass, that Cardinal fans would not be too disappointed. With their three guards and Dwight Burke moving on few would argue that Marquette will struggle in 2010. Location on the chart suggests the Golden Eagles will fall more than a spot or two in conference standings. Connecticut fans (and more than a few in the media) anticipate the Huskies will compete for the conference title. Pittsburgh & Syracuse no doubt have hopes too. The chart could hint however, at a very brisk competition for spots #4 through #8 in the conference, where experience (but not winning track records) will contend with winning track records (but less experience). Among the "green" quadrant teams, Georgetown's location is a bit of a surprise. After dropping nearly 8 spots in the Big East standings in 2009, I was frankly a little surprised to see the Hoyas place so well. Those anticipating Georgetown will again compete for conference honors may not be far off base.

Rosters for 5 programs have not been made public. Coupled with suspensions to Mountaineer players Joe Mazzulla and Truck Bryant (though if Bryant's participation in a European tour is a hint, he will most likely be reinstated to the team at some point in the next 3 or so months), West Virginia's returning minutes may be understated. Pending resolution of Reggie Redding's case, Villanova's returning minutes may be overstated. Radically alter those two teams and the scatter chart changes. The unpredictable events -- injuries, staff conflicts, chemistry problems within the team -- are impossible to graph.

No comments: