Fear Not, Fair Readers…

Posted by rtmsf on June 25th, 2007

We’re back from vacation.

The Shining

Wendy, I’m home!

Yes, we know you were waiting with baited breath for the next installment of our pithy commentary and analysis, but unlike certain charlatans working for large four-lettered sports news organizations, we don’t have enough friends (nor funds) to hire a guest blogger. Nor do we simply “forget” to finish posts, but that’s another issue for another time.

There’s a ton of things to get to, so for now, it’ll have to suffice with a simple agenda for the week. Here’s what’s coming:

  • Thursday night is the long-awaited 2007 NBA Draft, also known in common parlance as the Oden/Durant Sweepstakes. Everyone has an opinion on whom should be taken where, but we really don’t do projected drafts. What we will do, however, is take a look at the history of the NBA Draft through the prism of the many colleges that the players attended. Out of the more than two thousand players drafted in the first two rounds of the NBA Draft from 1949 to present, which schools are best represented? Who are the surprises and who are the disappointments? Which schools have trended up and down throughout the six decades of the draft? The first part of this snapshot analysis should roll out early this week and continue up through draft day.
  • We’re also nearly ready to introduce Part III of our NCAA Tournament Analysis, which will take a closer look at the raw numbers relating to the success of each conference during the 65 (64) team era. Building from this, Part IV will examine the conference overachievers and underachievers (assuming there are any), using a similar methodology to the one used for the schools themselves. Depending on how the NBA Draft material plays, these two posts should be coming out over the next week.
  • Finally, we’ll also try to find time to make some noise about some of the news that we’ve missed over the past week or so, including more of the coaching carousel, Indiana getting a new barn, and an NC State fan who told us over the weekend that “Sidney Lowe is going to be a beast.”
Share this story

NCAA Tourney Overachievers and Underachievers of the 65 (64) Team Era

Posted by rtmsf on June 15th, 2007

We’ve received a resounding response on the post Tuesday evaluating school performance during the 65 (64) team era, so let’s waste no time in presenting Part II of our statistical profile of the NCAA Tournament during that period. As we alluded to in that post, looking at raw data in a vacuum has its limitations. From that data we got a reasonable sense of which programs have had the most success during the period, but it didn’t really give any insight as to which programs have overachieved, underachieved, or simply performed about as expected.


Which Schools Reach Their Potential?

Today’s post will address that issue. You might want to bookmark this page for next year’s bracket-filling, because we feel that what we’re showing today has significant value in determining those schools that tend to embrace expectations and those that tend to wither in the face of the red-hot lights of March. Stat geeks like ourselves should enjoy this post – hopefully it won’t bore the rest of you.

The first consideration in trying to determine which programs have overachieved and underachieved is how to answer that question. Traditionally, you often hear commentators and writers talk about “playing to your seed,” which essentially presumes that a #1 seed should make it to the Final Four, a #2 seed to the Elite Eight, and so on (see Table A below). Using this model, a team that wins any games beyond the expected wins of its seed would fairly be said to have overachieved as to the expectations of the seed (and vice versa).

Table A. Standard Model

NCAA EV by Seed (Presumed)

So let’s try to first answer that question using the Standard Model outlined in Table A. But don’t give up on the post after you review the Standard Model results, as we have another model below it that may bring up new and interesting considerations. The Standard Model as applied to the 65 (64) Team Era is Table B below.

Table B. Standard Model Applied to 65 (64) Team Era

Notes: the table is sorted by “+/- per App,” which represents the number of games won above or below the expected number of wins for that seed per NCAA appearance (1985-2007). The rows colored in green highlight schools that are in the top twelve spots of both Tables B and D. The rows colored in yellow highlight schools that are in the bottom twelve spots of both Tables B and D.

NCAA EV by Seed - Detailed v.3

Inside the Numbers (Table B):

Overachievers. What first jumps out at us using this method is just how (except Florida) there are no big names near the top of the list. Tulsa, Seton Hall, GW, UTEP, Temple? Without taking away from their performance – after all, every one of those teams should be commended for overperforming versus its seed – this immediately suggests to us that the Standard Model often used by the pundits actually favors lower-seeded teams who manage to win a game or two every couple of years in the NCAA Tournament. Using the schools above +0.30 as a natural break point, only Florida, Louisville and Temple have won more than twenty games in the NCAAs during this period, and the average seed of these thirteen schools was 7.0. Are these teams chronically underseeded or are they overachieving as a general rule? Whatever the case, we’re going to think a little harder about picking Tulsa the next time they make the bracket.

Tulsa Logo

Are the Golden Hurricanes the Biggest NCAA Tourney Overachievers?

Underachievers. A brief look at the names at the bottom of the list – Oklahoma, Cincinnati, Wake Forest, Stanford, Purdue, etc. – tells us that we’re on to something here. As our good buddy MK says, “Wake and Purdue can effin’ bite me. They kill my bracket every year.” Or something like that. Using -0.42 as a natural break, the eleven teams at the bottom averaged a seed of 4.8 during this period, which is a marginally better seed than the overachievers on this list. Oklahoma, Arizona and Illinois – what do you have to say for yourselves? Each of you has had at least 18 appearances during this era, averaged a solid if not spectacular seed between #4 and #5, and consistently played below that seed to the tune of -0.42 wins per appearance or worse. This matches Tuesday’s analysis with Oklahoma as a disappointment; but under this model, Arizona and Illinois (second tier elite performers on Tuesday) have some explaining to do.

Oklahoma celebration

You May Want to Avoid These Guys in your Brackets

You’ll See Them Again on the Long Road Back to the Middle. The Big Four from Tuesday – Duke, UNC, Kentucky and Kansas – find themselves in the middle of the pack. It’s notable that Duke averaged a seed of 2.32 during this era and still overachieved by +0.09 wins per appearance relative to seed. UNC and Kentucky are verrrry slightly underachieving, whereas KU’s recent vintage teams (2005 & 2006) probably contribute to its slightly lesser position at -0.23 wins per appearance. UConn should be lauded for overachieving at a clip of +0.14 wins per appearance, which somewhat mitigates their overall lack of F4s – they do win a lot of NCAA Tournament games. Syracuse, Georgia Tech, Memphis, Iowa St., California, Princeton and Notre Dame – all of these programs have played exactly to seed in the last 23 years. Be sure to remember that fact next year when filling out your brackets.

Criticism of the Standard Model. The primary problem with evaluating NCAA Tournament success based on “playing to the seed” is that it puts an unrealistic burden on highly-seeded teams because over time they never play to their seed. In the knockout crucible that makes the NCAA Tournament so exciting, a corollary result of using this criteria is that higher seeds (#1-#4) are largely set up to fail, and lower seeds (#9-#16) are significantly overachieving simply by winning one game. Additionally, we feel that using the Standard Model assumes too much at the squishy margins – that a #4 seed should win two games, a #5 seed should win only one game, and that a #8 seed should always beat a #9 seed – when both teams are probably very similar in ability. For this reason, we now offer an alternative model, one that considers actual historical performance of the seeds.

To that end, see Table C below for a historical snapshot of how well each seed actually performed in the 65 (64) team era.

Table C. Historical Model

NCAA EV by Seed - Historical

Considering Table C, we wanted to point out for a moment that we have a couple of thoughts. First, we found it interesting that #6 seeds have traditionally performed better than #5 seeds. This at first seems anathema to conventional wisdom, as a #6 likely has to get past a #3 in the second round, whereas a #5 must only defeat a #4. That is, until you remember that a #5 has the dreaded 5-12 match-up that knocks out so many higher-seeded teams (excepting 2007, of course). Also, it’s interesting that a #10 seed has a better historical expected number of wins than a #9 seed. This shows again that a better draw is to avoid the 8-9 match-up, even if you end up as a lower (but more likely to succeed) #10 seed.

Using the 1985-2007 Historical Model as a baseline expected value for each seed, our list of 64 schools is now featured in Table D below.

Table D. Historical Model Applied to 65 (64) Team Era

Notes: the table is sorted by “+/- per App,” which represents the number of games won above or below the expected number of wins for that seed per NCAA appearance (1985-2007). The rows colored in green highlight schools that are in the top twelve spots of both Tables B and D. The rows colored in yellow highlight schools that are in the bottom twelve spots of both Tables B and D.

NCAA EV by Seed - Hist Detailed 3

Notable Differences. Using the Historical Model, the first thing that strikes us as interesting is that many of the heavy hitters of the era have jumped considerably, including some all the way to the top of the list. In fact, the following schools – Kentucky, UNC, Duke, Kansas, UConn, UNLV, UCLA, Syracuse, Ohio St. – all rose at least fifteen places from where they were situated in Table B. Each of these schools (excluding Kansas and Ohio St.) have overachieved at least +0.25 wins per appearance above the historical performance of their seeds. Since these schools represent a total of thirteen championships and thirty-eight F4s during this era, the Historical Model viewpoint makes more sense than what the Standard Model shows.

Who Dropped? Creighton, George Washington, UAB and UTEP all went from nice-looking overachievers in Table B to slight underachievers using this model. NC State, Xavier and Gonzaga looked fantastic using the Standard Model, but look rather pedestrian using this one. The explanation for this is simple – these teams collectively averaged a #9 seed throughout this era – if we use the Basic Model, they should never win a game, so when they do, it looks as if they’re significantly overachieving. However, using the Historical Model, we see that they actually win NCAA Tournament games roughly commensurate with how a #9 seed typically performs. This again illustrates why we feel this model shows a truer snapshot of performance.

More Importantly, Who Stayed the Same? Now this is where it gets really interesting. The rows shaded in green in Tables B and D represent schools that were in the top twelve in playing beyond its seed using both models. You could accurately say that, no matter how you slice it, Florida, Seton Hall and Louisville are the truest overachievers of the 65 (64) team era. (Incidentally, Florida and Seton Hall are in the top four of both tables, with each school averaging an extra win beyond its seed for every two appearances in the NCAA Tournament.)  These three schools consistently play beyond the actual and historical expected value of the seed that the NCAA Selection Committee gives them.  

The rows shaded in yellow at the bottom of Tables B and D represent the true underachievers. These schools were in the bottom twelve in playing below its seed using both models. What’s peculiar about this statistic is that there appears to be a much greater correlation between the tables with respect to the underachievers than with the overachievers. Whereas the overachievers only had three common schools between the tables, regular NCAA disasters such as Georgia, Oklahoma, Stanford, Purdue, Wake Forest, Pittsburgh and St. John’s find themselves in both at the bottom looking up. If you are at all like us and believe that the weight of history contributes as a predictor of the future, then anyone reading this post should be extremely hesitant picking these schools to play beyond its seed to its seed in the Dance in the years going forward.

Chicken or the Egg

And Now, the Chicken/Egg Argument. Whether you buy into the Standard Model or the Historical Model or neither, we can say without a doubt that some schools appear to consistently overachieve relative to its seed, while others consistently underachieve relative to its seed. The question we want to posit to the readers is what causes this – is this simply a statistical anomaly? Are the overachievers regularly seeded too low by the NCAA committee; or, are the underachievers regularly seeded too high? Is there a self-fulfilling prophecy at work here, where schools that are known as chronic underachievers (ahem, Oklahoma) place so much pressure on themselves that they tighten up in the clutch? Conversely, do the overachievers find their wits amidst chaos to consistently come through under pressure because they are expected to do so? These are questions that we cannot answer, but we figure that there are elements of each at play here. We’d certainly like to hear your thoughts on this and any other topic. Oh, and apologies for the long post today.

Coming Next: We’ll take a look at the raw numbers again, but this time we’ll examine it by conference, which isn’t as easy as it sounds. Do you realize how many conference re-alignments and mergers and dissolutions there have been in the last 23 years of NCAA basketball?




Share this story

Top NCAA Performers of the 65 (née 64) Team Era

Posted by rtmsf on June 12th, 2007

GMU Cartoon

Since there’s absolutely nothing going on this week, this is a good a time as any to start rolling out some of the data that we’ve been hoarding. First, a respectful tip of the hat goes to Florida Gator fan Louis Frank, who allowed us unbridled access to the detailed work in his NCAA Tournament database. Over the next week or so, we’re going to be presenting some descriptive statistics on the 64/65 team era of the NCAA Tournament sliced and diced in various ways.

Our first focus will be on individual team performance, viewed through the raw numbers and then with some analytical twists; then we’ll turn our attention to conference performance using the same parameters. The basic question we seek to answer is which teams and conferences tend to over- and underachieve in the NCAA Tournament since it expanded to 64 teams in the 1985 season? The reason we start with that somewhat arbitrary season is because from that point until now every championship team has had to win six games against seeded teams, with no exceptions. It also provides a tidy way of reviewing the data with a substantial sample of seasons – twenty-three – which also happens to coincide perfectly with the rise in popularity of NCAA basketball and the ESPN era.

NCAA Tournament Success (1985-2007)

Notes: the chart is sorted by winning percentage (minimum: 8 appearances) from 1985-2007. The green shaded rows represent schools that have won a national title during this period.

NCAA Tournament 1985-2007 v.3

Inside the Numbers:

Elite Eight. Of the 267 schools that have been invited to the NCAA Tournament during the last 23 years, the 64 listed above are the chronic repeat performers, each having made the Dance on at least eight occasions. Thirty-nine of those sixty-four have winning (> .500) records, but only a handful, eight, are elite (> .700 winning pct.). Suffice it to say that those eight elite programs account for 14 of the 23 (61%) national championships and 39 of the 92 (42%) Final Four teams during this era (programs with a national title are denoted above in green shading). These eight programs are: Duke, Connecticut, UNC, Kentucky, UNLV, Kansas, Florida and Michigan. Incidentally, Georgetown is the only school of the top 13 who did not have a title from 85-07, but dumb luck led to its 1984 title team being omitted from this list – apologies to the Hoyas.

Coach K b/w

You Have to Give the Devil His Due

The Krzyzewski Era. This era also neatly coincides with the rise of Duke as a basketball powerhouse – Coach K’s first Final Four was in 1986, and his string of success particularly from 1988-92 exceeds by itself almost every other school’s performance on this list. Duke has the most #1 seeds, the most Sweet 16 appearances, the most Final Four appearances, the most wins, the best winning percentage and the most national titles during this period. In several of those categories it leads by comfortable margins. We’ve made note that the current era of Duke basketball might be slipping a tad, but with numbers like the above to sustain, that may be an impossible task even for Krzyzewski. By these numbers, you’d have to go with North Carolina in second place and Kentucky a close third. Each has very similar statistics (appearances, #1 seeds, sweet 16s, titles, wins, winning pct.) in all but one category, Final Fours. Given the importance that the college basketball community places on reaching the final weekend, Carolina’s seven F4s to Kentucky’s four must trump, all else being relatively equal.

Traditional Powers. With Duke, UNC and Kentucky taking the top three spots by the raw numbers, how do the other three traditional powers of UCLA, Kansas and Indiana fare? Kansas is closest to the top group. The Jayhawks mirror UNC in many categories (including F4s), but its winning percentage is a little lower and it lacks that second national title that would vault it into the top three. UCLA experienced a couple of down periods during this era, but now appears to be on the rise again with two F4s in the past two seasons. Still, its top ten winning percentage (.667) and its national title in 1995 keep it in the second tier of performers over this era. Indiana has largely struggled since Bob Knight was forced out of Bloomington, but their consistency in making the tournament and winning a game or two (.604 winning pct.) – plus that national title in 1987 – probably keeps it in the second tier as well. There should be no question, though – if any of the traditional six powers were slipping in favor of one or more of the nouveaux riche, Indiana would be the choice here.

Bob Knight IU

IU is Showing post-Knight Slippage

Nouveaux Riche. Of the elites, Connecticut and Florida are clearly the party-crashers. Prior to 1985, UConn had four wins and Florida zero wins in the NCAA Tournament. Each now has two national titles and a winning percentage of greater than .700. The question is whether these programs will be sustainable whenever Calhoun and Donovan decide to move on (Calhoun, to retirement; Donovan, as Christine’s full-time house-b*tch). The 64/65 team era is already littered with similar riches-to-rags stories such as UNLV, which fell hard when Tarkanian was indicted retired; and, Michigan, who also dropped out of the college basketball landscape once the gravy train of athlete peddler Ed Martin ended. Arkansas is yet another example – all three programs have a national title and multiple F4s to its credit, but long periods of poor teams and inconsistency places them in the second tier of the era.

Others in Second Tier. Several programs were consistenty excellent over this era, but their numbers weren’t as eye-popping as some of the above schools. Syracuse, Michigan State, Maryland and Louisville all claim a title to go with multiple F4s. Who knew other than Orange fans that Jim Boeheim’s squad never claimed a #1 seed during this era – that seems hard to believe. True, though – Syracuse’s best seeds were five #2s – during the glory days of Pearl Washington, DC, Billy Owens and company – 1986, 1987, 1989, 1990, and 1991. It looked like Michigan State was ready to become a top tier program 6-8 years ago, and they still are an excellent one, but its winning percentage needs to improve a little more to reach that level.

Lute Olson

The Silver Fox has had his Ups and Downs

Whither Arizona? Arizona is the only school that was invited to the NCAA Tournament each year of this era. Yet Arizona’s success in the postseason leaves something to be desired for a program of its stature – multiple F4s and a title, but near the bottom of the championship-level schools in winning percentage. The Wildcats are a team to keep a watchful eye on when we present our over- and under-acheivers list later this week.

Rising Stars. Several programs to observe closely as we go deeper into this era are rising stars Georgetown, Ohio St., Memphis, Texas and Gonzaga. None yet has a title during this era, but each except Gonzaga has been to a F4, and all five are knocking on the door. These programs have the facilities and coaching in place to continue to rise up this list in the coming years.

Disappointments. Again, basing these observations on nothing more than raw numbers, you’d have to say that Oklahoma, Illinois, Purdue and Stanford have been the biggest disappointments. Collectively, these schools have had fourteen #1 seeds with only five F4s to show for it (obviously, zero titles as well). Although most of these programs have been consistently invited to the NCAA Tournament during this era, none has a winning percentage topping .600.

Quin Snyder Norm Stewart

What did these two do to Missouri?

Embarrassments. We’ll leave the mid-majors like Xavier alone here, but we wanted to save special mention for some of the BCS schools who have managed to get invited multiple times, but really didn’t do much when they got there. Georgia‘s one sweet sixteen in eight appearances and its .333 winning percentage doesn’t say much for a program that always seems to be rebuilding; Bob Knight’s Texas Tech doesn’t fare much better (two sweet sixteens). But the real winner of the most pathetic NCAA-caliber program award, in our estimation, has to belong to Missouri. The Tigers have been to the tournament fourteen different times during this era, even once as a #1 seed, and have only managed three sweet sixteen appearances, two elite eights and an overall losing record (.462). Serious congrats are in order for Norm Stewart and Quin Snyder. Mike Anderson has his work cut out for him. The saddest part is that Mizzou traditionally likens itself as a basketball school!

Ivy Sadness. The last word goes to Ivy stalwarts Penn and Princeton, two schools who show up every year (21 of the last 23 NCAA Tournaments) at the right time and venue, battle hard for about thirty minutes against a superior athletic opponent, then go back home and lick their wounds for another year after inevitably wearing down to the size and strength of its opponent. They may be a collective 3-21 (.125) in the Dance, but who will forget when they pull the big upsets, like Princeton 43, UCLA 41 (1996), or Penn 90, Nebraska 80 (1994). Ok, maybe beating Nebraska isn’t a big upset after all, but we still love the UCLA upset.

Coming Next: now that we’ve analyzed the raw numbers of the 64/65 team era of the NCAA Tournament, we’ll next be taking a look at the over- and under-achievers during the same period. After seeing the above, can you project who the best and worst will be? You might be surprised at some of the results.  View Overachievers and Underachievers here.

Share this story