 Read the daily ramblings of Pete Tiernan as he discusses the strategies behind BracketScience.
Efficiency numbers say four and five seeds will be dramatically weaker |
|
Tue,
Jan
31,
2012,
6:33 PM
|
|
0 Comments...
|
A month ago, I noted the low efficiency numbers of this year’s top ten squads compared to the previous eight seasons. I mentioned that this was an important development to track because there’s a correlation between the overall efficiency of the elite teams and tourney predictability. Basically, the numbers suggest that the worse the overall efficiency of the top ten, the more upset-laden the dance.
Last year’s tournament featured the worst top ten teams in terms of Ken Pomeroy’s Pythag efficiency calculations since 2004, when the data became available. And what happened? The dance tied a record for upsets (13) and broke the Madometer unpredictability record (19.8% madness). The 2010 dance saw the second most inefficient top ten—and it had the third highest upset and Madometer totals. The second craziest dance of the eight I studied was 2006—and its top ten teams also had Pythag numbers well below average.
I describe this correlation between Pythag ratings and tourney unpredictability in more detail at the end of the “2011-12 Trends and Flukes” article. In that article, I compared the average Pythag values—and defensive efficiency in particular—of the top six seeds since 2004. The results were even more closely correlated between efficiency and predictability.
So if these numbers convince you that Pythag is a good predictor of whether the dance will play chalky or go haywire, it pays to track them on a regular basis. When we did the analysis a month ago, the 2012 top ten owned an average Pythag of .9510—sixth best out of nine seasons. Today, the 2012 top ten is still just the sixth most efficient class since 2004—but their efficiency numbers are dropping. Take a look:
As you can see, last year’s ten best squads were the least efficient since 2004—and they played the worst in the dance, greatly contributing to the record madness. The 2010, 2009 and 2006 top tens also had below average efficiency numbers…and below average tourney records. Now, take a look at the red column. That’s the average Pythag of the current top ten most efficient teams in the country. It’s down to .9494, dangerously close to dropping below both 2009 and 2010. That does not bode well for an orderly “by-the-seeds” tourney.
Let’s think of it another way. I compared the average Pythag of all the top teams since 2004 with this year’s current top Pythag team (Ohio State), then the average Pythag of all the second best teams with this year’s number two (Wisconsin), and so on for the top 20 teams. What I discovered was that this year’s six best squads compare favorably to the top six of the last eight seasons, but the rest of the top 20 is markedly weaker than their predecessors. Take a look:
Ohio State is actually a better top team than average; Wisconsin is a worse number two; and Kentucky, Kansas, Michigan State and Syracuse are pretty much average third- through sixth-ranked teams.
Then things get ugly. From #7 Missouri to #12 Wichita State, the teams are markedly weaker than their predecessors. After that, the gaps get even wider. From Duke all the way to #20 Texas, this year’s squads are dramatically less efficient than comparative teams of the last eight years. Here’s another way to look at it: the average 15th best team from 2004-2011 would be the ninth best team this year. And Duke, this year’s 13th best team, would be just the 23rd most efficient team in years past.
What does that portend for the 2012 tourney? If this curve stays substantially the same shape as it is today, it would mean that, while top seeds might match historical quality, two and three seeds will be slightly weaker. And four and five seeds will be especially soft. Considering that 4v13 and 5v12 opening-round matchups are already fertile ground for upsets—more than two per tourney—this could presage outright craziness. We’ll continue monitoring the efficiency numbers as the season winds down.
|
|
0 Comments...
|
|
Usual suspects on champs list—but only three qualify based on adjusted scoring and margin |
|
Mon,
Jan
30,
2012,
8:34 PM
|
|
0 Comments...
|
There are eight stats you can use right now to narrow your list of potential champions. Every single one of the last 10 champs have:
• earned a one, two or three seed
• come from a Power conference (ACC, Big East, Big Ten, Big 12, Pac-10 or SEC)
• been led by a coach with more than five tourney trips and at least one Elite Eight run
• either went to the previous year’s dance or had an All-American
• averaged more than 73 points per game
• allowed fewer than 73 points per game
• owned an average scoring margin of at least seven points per game.
• played a schedule among the 75 strongest in the country
At this point, only eight teams have the statistical chops to be champs. They are: Kentucky, Syracuse, Ohio State, North Carolina, Duke, Kansas, Michigan State, and Florida. Check out the numbers:
(PR = Pythag rating, CF = Conference, */11 = have a star or went to the previous dance, CO = coach criteria. All the rest of the abbreviations should be clear.)
This is the same list of suspects as last week. Since we relaxed the offense and margin requirements from last year to reflect the downward trend in scoring, more teams are meeting the criteria. If you read the previous blog entry, however, you know that I posited another way to evaluate a team’s champ qualifications.
The average tourney champ scores 9.6% more than the tourney field and owns a scoring margin 74.5% better. Even if we assume that the tourney field will have an all-time low 72-point scoring average and a margin one point below the 27-year 8.5 average, the percentage increases equate to a 78.9 scoring average and 13.1 margin. Of our eight stats champs, only two meet the adjusted criteria—North Carolina and Florida. One makes sense; the other is a surprise.
Each week, we also track the teams whose AP rankings don’t reflect their possession-based efficiency numbers. This is the invaluable “Pythag” rating that Ken Pomeroy tracks on www.kenpom.com. Right now, according to efficiency numbers, six teams don’t deserve to be ranked in the top 20. The biggest imposter is San Diego State (ranked 53th in Pythag), but undefeated Murray State (46), Creighton (27), St. Mary’s (22) and UNLV (21) aren’t playing as efficiently as their ranking would indicate either.
They’ve supplanted six more efficient squads: Wichita State (11), St. Louis (14), Florida State (16), California (20) and Texas (20).
|
|
0 Comments...
|
|
With scoring decreasing, are defensive-oriented teams faring better in the dance? |
|
Sat,
Jan
28,
2012,
1:44 PM
|
|
0 Comments...
|
ome of you have noticed that I’ve relaxed my scoring and margin credentials for the champion criteria this season. For the last four years, I’ve been warning that the trend toward lower scoring in college basketball will render some of the time-honored guidelines for tourney advancement obsolete. 2011, for instance, marked the first dance in 26 years that the tourney champion didn’t average more than 75 points per game. UConn’s scoring average was 73.4. You have to go all the way back to the first tourney of the modern era to find a lower scoring team. In 1985, Villanova won with a scoring average of 71.5.
|
|
0 Comments...
|
|
Still only eight champs left; Kansas, MSU and Florida at most risk to drop off |
|
Mon,
Jan
23,
2012,
5:49 PM
|
|
0 Comments...
|
After a week in which several top teams—notably Syracuse and Duke—suffered big upsets, the list of teams with tourney champ credentials hasn’t changed. If you haven’t been following this segment, there are eight stats you can use right now to narrow your list of potential champions. Every single one of the last 10 champs have:
• earned a one, two or three seed
• come from a Power conference (ACC, Big East, Big Ten, Big 12, Pac-10 or SEC)
• been led by a coach with more than five tourney trips and at least one Elite Eight run
• either went to the previous year’s dance or had an All-American
• averaged more than 73 points per game
• allowed fewer than 73 points per game
• owned an average scoring margin of at least seven points per game.
• played a schedule among the 75 strongest in the country
At this point, only eight teams have the statistical chops to be champs. They are: Kentucky, Syracuse, Ohio State, Kansas, North Carolina, Duke, Michigan State, and Florida. Check out the numbers:
(PR = Pythag rating, CF = Conference, */11 = have a star or went to the previous dance, CO = coach criteria. All the rest of the abbreviations should be clear.)
This is the same list of suspects as last week. Since we relaxed the offense and margin requirements from last year to reflect the downward trend in scoring, more teams are meeting the criteria. Historically, however, tourney winners have averaged scoring more than 77 points a game with a margin of least 10 points. By that standard, Kansas and Michigan State would fall off the champ list. And if we restrict our picks to the top 12 teams (the squads currently most likely to bag top-three seeds), Florida is also at risk of failing to meet stats champ credentials. That leaves a handful of more likely winners: Kentucky, Syracuse, Ohio State, North Carolina and Duke.
Each week, we also track the teams whose AP rankings don’t reflect their possession-based efficiency numbers. This is the invaluable “Pythag” rating that Ken Pomeroy tracks on www.kenpom.com. Right now, according to efficiency numbers, six teams don’t deserve to be ranked in the top 20. The biggest imposter is Mississippi State (ranked 57th in Pythag), but San Diego State (55), Murray State (42), Michigan (39), Creighton (327) and Marquette (23) aren’t playing as efficiently as their ranking would indicate either.
They’ve supplanted six more efficient squads: Wisconsin (third best in Pythag—despite five losses), Wichita State (nine), St. Louis (13), Alabama (18), Kansas State (19 and California (20).
|
|
0 Comments...
|
|
Round-by-round advancement rates show one seed dominance, but low seed champs come in pairs |
|
Mon,
Jan
16,
2012,
6:49 PM
|
|
0 Comments...
|
I just posted the “2012 Seed Guide,” a six-part analysis of every match-up that’s occurred in every round of the tournament. The guide is one of the more valuable resources I provide for making your tourney picks. This year, I’m providing “at-a-glance” scoring margin charts for every tourney match-up as well.
There’s a ton of information in the guide—and it’s probably best absorbed after Selection Sunday, as you mull over your bracket. For now, it may be helpful to get some quick and simple guidance. Every year, I also put together the graphic below. It shows the round-by-round advancement rates of the top six seeds in the tournament. These are the most important teams to watch in the tourney. 101 of the 108 Final Four combatants and 26 of the 27 champions come from these seeds.
The animation gives a powerful illustration of the dominance of top seeds in the dance. More one seeds reach the Elite Eight than five or six seeds make it to second round. And nearly as many top seeds (45) reach the Final Four as four seeds make the Sweet 16 (46). Heck, top seeds have won 16 of the 27 dances in the modern era…while just 14 three seeds have made it to the Final Four.
Of course, last year’s champ UConn was a three seed, the first non-top seed since 2006, when Florida, another three, cut down the nets. It’s tempting to think that the odds favor a one seed taking this year’s title. But think about this: in the six other times that a lower seed succeeded a one seed in winning the championship, another lower seed won the following year four times.
Here’s a look at the history of lower-seed champions and what happened the following year:
1985 – Villanova (8), 1986 – Louisville (2)
1988 – Kansas (6), 1989 – Michigan (3)
1991 – Duke (2), 1992 – Duke (1)
1997 – Arizona (4), 1998 – Kentucky (2)
2003 – Syracuse (3), 2004 – Connecticut (2)
2006 – Florida (3), 2007 – Florida (1)
2011 – Connecticut (3), 2012 - ?
Here’s a funky little fact: in the only two years when a lower-seeded champion was followed by a top seed victor (1991 and 2006), that top seed was the previous tourney champ (1991 and 1992 was the Duke repeat, 2006 and 2007 was the Florida repeat). Hmm. Unless UConn finishes the season on an amazing run, the pattern won’t repeat this year. Could we have another non-top seed champ?
|
|
0 Comments...
|
|
Butler top upset victor, Duke top upset victim |
|
Fri,
Jan
13,
2012,
7:58 PM
|
|
0 Comments...
|
There have been 232 upsets in the 27-year history of the 64-team era. That’s about 8.5 upsets per dance. I define an upset as any game in which a team seeded four positions or lower than the favorite prevails. Some teams have a knack for springing surprises—and others are prone to being victimized.
Not surprisingly, given their last two surprise runs to the Final Four, Butler is the top Cinderella. The Bulldogs have sprung six upsets. So has Villanova, but the Wildcats have also been upset twice. Auburn and Richmond has five upsets without being victimized themselves. Tulsa has five upsets and was upset once. And Boston College and Gonzaga have five upsets but suffered two shockers.
Speaking of shockers, Duke is the all-time victimized team. The Blue Devils have been upset nine times (more than one-third of their appearances) without springing a surprise of their own. Syracuse has succumbed to a darkhorse eight times (including the first loss to a 15 seed—Richmond—in 1991) and hasn’t posted an upset either. Oklahoma and Kansas are the other two eight-upset schools, but both have pulled two stunners of their own.
Here’s a list of the top upset victors and victims. I’ve ranked them by their net upsets (the shockers they’ve pulled minus the ones they’ve suffered).
|
|
0 Comments...
|
|
Efficiency stats say slower-paced teams have worse luck—but is it true? |
|
Tue,
Jan
10,
2012,
10:29 PM
|
|
2 Comments...
|
If you haven’t dug into Ken Pomeroy’s efficiency statistics (at www.kenpom.com), you may not know that he has a calculation for luck. Ken defines luck as “the deviation in winning percentage between a team’s actual record and their expected record using the correlated gaussian method. The luck factor has nothing to do with the rating calculation, but a team that is very lucky (positive numbers) will tend to be rated lower by my system than their record would suggest.”
|
|
2 Comments...
|
|
Defending champ UConn drops off champ list; nine teams left |
|
Mon,
Jan
9,
2012,
7:10 PM
|
|
0 Comments...
|
There are eight stats you can use right now to narrow your list of potential champions. Every single one of the last 10 champs have:
• Earned a one, two or three seed
• Come from a Power conference (ACC, Big East, Big Ten, Big 12, Pac-10 or SEC)
• Been led by a coach with more than five tourney trips and at least one Elite Eight run
• Either went to the previous year’s dance or had an All-American
• Averaged more than 73 points per game
• Allowed fewer than 73 points per game
• Owned an average scoring margin of at least seven points per game.
• Played a schedule among the 75 strongest in the country
At this point of the season, only nine teams have the statistical chops to be champs. They are: Syracuse, Kentucky, North Carolina, Ohio State, Michigan State, Duke, Kansas, Georgetown and Florida. Check out the numbers:
(PR = Pythag rating, CF = Conference, */11 = have a star or went to the previous dance, CO = coach criteria. All the rest of the abbreviations should be clear.)
UConn made the grade last week, but they fell a measly tenth of a point short in offensive scoring. The 73 points-per-game was lowered this year because of the downward trend in offensive output in college ball. Last year, the champ criteria called for at least a 77-point scoring average. If that were still the barrier this year, three more teams would drop off this list: MSU, Kansas and Georgetown.
The other thing we’re tracking is which teams’ AP rankings don’t reflect their possession-based efficiency numbers. This is the invaluable “Pythag” rating that Ken Pomeroy tracks on www.kenpom.com. Right now, according to efficiency numbers, six teams don’t deserve to be ranked in the top 20. The biggest imposter would be Mississippi State (Pythag ranked 56th), but UConn (34), Murray State (32), Virginia (31), Michigan (27) and Louisville (25) aren’t playing as efficiently as their ranking would indicate either.
They’ve supplanted six more efficient squads: Wisconsin (second best in Pythag—despite five losses), St. Louis (12), Alabama (14), Gonzaga (18), California (19) and Wichita State (20).
As an aside, something is fishy with the Pythag numbers when a five-loss team like the Badgers, who looked dreadful in their loss to Michigan on Sunday, is the second most efficient team in the land. One explanation is that they have bad luck, 324th worst in D-I out of 345 teams. But I notice a trend: most of the “bad-luck” teams are also slow-tempo squads. I’m going to dig into this phenomenon in a separate blog later this week.
|
|
0 Comments...
|
|
Scoring margins coming soon for every seed matchup in all rounds |
|
Thu,
Jan
5,
2012,
8:17 PM
|
|
0 Comments...
|
As part of my research for the “2012 Seed Guide,” I’ve put together charts of the average scoring margins that higher seeds have posted in every seed matchup for all six rounds of the tourney.
|
|
0 Comments...
|
|
|
|
