2014 wrap-up: maddest dance, most upsets in 30 years

2014_madometer_FINAL

UConn’s victory over Kentucky on Monday made official what we already knew: this year’s tourney broke 30-year records for the highest level of unpredictability and most upsets.

Before this year, the record Madometer reading was 20.8% deviation from perfect high-seed dominance, set in 2013. One dance later, the record now stands at 21.4%. And before the Shabazz Shocker, there were five tournaments with 13 upsets (seed gap between opponents of four or more). 2014 saw 15 upsets. It takes some serious craziness to reach that level, particularly since the later rounds usually feature so few games with upset potential.

We’ve now seen the Madometer reach record levels in three of the last four seasons. In fact, the last five years have easily been the craziest of the 30-year 64-team era. In the first 25 years of the modern bracket, the average dance deviated from higher-seed perfection by 13.6%. In the last five years, that deviation is 19.2%, more than 40% wilder. Coincidence? I don’t think so. There are lots of reasons to explain it: one-and-done, rise of the Mid-Majors, the three-point equalizer, the vanishing big man, mis-seeding mayhem, the perils of parity,  havoc of shifting rules. It reads like a list Hardy Boys titles.

I’m not quite sure what the real reason is for the Madder March Madness, but I do know this: it’s going to make filling out your bracket that much more difficult. About the only thing I got right this year was my December prediction that the tourney would be crazier than ever. That and my assessment, after comparing the 2014 quality curves with past tourneys, that this year most resembled 2011. I just didn’t know how much it would resemble that year. Kemba Chaos morphed into Shabazz Shock.

The bracket models took a beating. Even the best one, “Coaching + Efficiency” would not have won a pool. Heck, you would’ve done about as well as Nate Silver by just picking the higher seed in every match-up. And you would’ve trounced KenPom and ESPN BPI. Maybe when every single ESPN analyst picked MSU to cut down the nets, we should’ve known that something strange was up. Here are the final results of the models:

  • 85.7% – 4 Coach+Efficiency
  • 80.7% – 13 Nate Silver
  • 77.6% – 8 Baseline
  • 69.4% – 16 Keeper Small
  • 69.4% – 5 Pulse Check
  • 69.4% – 7 Upset/Toss-up
  • 62.1% – 2 Final Four/Champ
  • 62.1% – 0 From the Gut
  • 46.9% – 1 KenPom
  • 43.2% – 11 ESPN BPI
  • 39.7% – 9 Seed Match-ups
  • 33.3% – 3 Outcome Match
  • 27.4% – 6 Factor PASE
  • 24.8% – 15 Billion Dollar
  • 22.3% – 12 Contrarian
  • 16.0% – 14 Brackomatic

All in all, 2014 was one nutty year. We could pass it off as an outlier and go back to our conventional ways of predicting the tourney. But given the last five seasons of madness, I don’t think that’s the right approach. I have seven months now to ponder the results and come out with a new way of looking at the tournament. As rugged as this year was, I’m looking forward to trying to make sense of it–knowing that I might never get there.

Thanks, everyone, for thinking alongside me.

Posted in General News, Measuring Madness | 59 Comments

Coaching fortunes will change after the 2014 tourney

When the dust settles on the 2014 dance, we’re likely to see dramatic changes in the performance rankings of the active coaches. Some coaches, like perennial overachiever Tom Izzo and newcomers Archie Miller and Kevin Ollie, will post big PASE numbers. Others, like Coach K, Bill Self and Jay Wright, will suffer significant declines in their performance against seed expectations. Here’s a quick list of where the coaches stand heading into the Elite Eight.

Changing Jobs?

The tourney has always served as a kind of interview for young coaches to jump to more prominent programs. Look back no farther than last year, when Andy Enfield parlayed his 15 seed Florida Gulf Coast squad’s surprise Sweet 16 run into a new job at USC. So who’s in line for a raise after this year’s dance? Here are the most likely advancers, in order of age:

  • Archie Miller (Dayton), 35 years old
  • Saul Phillips (ND State), 41 years old
  • Tommy Amaker (Harvard), 48 years old
  • Johnny Dawkins (Stanford), 50 years old
  • Brad Underwood (Stephen F Austin), 50 years old
  • Bob Hoffman (Mercer), 57 years old

Dayton’s Archie Miller is the hottest commodity. Not only have his 11 seed Flyers made it to the Elite Eight, but he’s also the youngest coach on this list. Saul Phillips helped his stock as well, with both a 5v12 upset and a stirring tear-filled post-game interview. (“I love these guys. Love them. Absolutely Love them.”). Coach K disciples Tommy Amaker and Johnny Dawkins could also get looks from bigger programs. Brad Underwood and Bob Hoffman might also change jobs, but their age may be limiting factors.

Big Increase

  • Tom Izzo
  • John Calipari
  • Kevin Ollie
  • Cuonzo Martin
  • Scott Drew
  • Steve Alford

Depending on the fates of Michigan and Michigan State, Tom Izzo could leapfrog John Beilein for the PASE lead among veteran active coaches. Izzo probably has the inside track. His Spartans are favored over UConn—and as a four seed, he’s got a higher ceiling for seed overachievement.

John Calipari will also see a sizable jump in his +.336 PASE value. Even if he didn’t win another game, Calipari is about 2.3 games above the projected wins for an eight seed. (Whether Kentucky should’ve gotten an eight seed is another question.)

Kevin Ollie and Cuonzo Martin will start their tourney careers with big, fat overachieving PASE numbers. Ollie’s will be at least +2.200 and Martin’s will be around +1.500.

Scott Drew should see a modest bump in his already-impressive +.580 PASE, since the average six seed wins only 1.16 games per appearance. And Steve Alford should climb several rungs out of the PASE basement with his Bruins’ Sweet 16 run. He’s currently a -.531 PASE underachiever…but should wind up around -.400 when the final numbers are calculated.

Slight Increase

  • Steve Fisher
  • Mark Few
  • Josh Pastner
  • Jamie Dixon
  • Jim Crews

Steve Fisher’s solid +442 PASE is due for a slight uptick since he beat expectations by about .475 games. Mark Few, Josh Pastner and Jamie Dixon will see modest improvements with their opening round wins as eight/nine seeds. And Jim Crews will yank up his sagging -.290 PASE, despite underachieving as a five seed. He fell only about .129 games short of expectations.

Slight Drop

  • Doug McDermott
  • Roy Williams
  • Fred Hoiberg
  • Rick Pitino

Off these four coaches, McDermott fell furthest away from seed expectations in the 2014 tourney. However, his PASE is already a weak -.171, so missing the projected wins for a three seed by .862 has less of an impact. Roy Williams came just .164 games from meeting expectations, so his PASE will drop, but not by much. Hoiberg and Pitino actually overachieved slightly for three and fours. But since their PASE values were already lofty (Hoiberg +.336, Pitino +.733), they’ll suffer small decreases.

Given the obstacles Hoiberg’s Cyclones faced with the Niang injury, his coaching performance is pretty darned impressive. That won’t go unnoticed by bigger schools. In fact, Hoiberg is probably the most likely candidate to get an offer from a prominent program—or even a pro team like the Timberwolves. Whether the Mayor leaves his town, however, is a matter of debate.

Big Drop

  • Gregg Marshall
  • Tony Bennett
  • Mike Krzyzewski
  • Jay Wright
  • Bill Self
  • Thad Matta
  • Mick Cronin
  • Shaka Smart
  • Lon Kruger
  • Jim Boeheim

Plenty of coaches are going to take big tumbles down the PASE rankings. Wichita State’s Gregg Marshall missed seed-projected win totals by the most of any coach in the 2014. Sure, it’s a little unfair that Kentucky was a four seed in disguise. But other top seeds have had to face the same adversity.

Tony Bennett also fell short of expectations for a one seed, but his two wins mitigated the damage. Coach K’s failure as a three seed was actually a more dramatic underachievement than Bennett’s. We’ve already estimated that Coach K’s PASE will drop to about +.300. That’s solid—but nothing like the invincible +.920 PASE he owned at the turn of the century.

The one-win runs of Jay Wright (+.053) and Bill Self (+.082) as two seeds will likely drop their treading-water PASE values into the depths of underachievement. Thad Matta and Mick Cronin will also pay for their -1.164 game deficits as six seeds. And Kruger and Boeheim will experience modest tumbles as well.

The coach that will surely see his PASE fall the hardest is Shaka Smart. With that golden run to the 2011 Final Four and a follow-up overachieving dance in 2012, Smart owned a sterling +1.267 PASE heading into this dance. But his 5v12 upset at the hands of Stephen F. Austin will knock his PASE down to about +.670. That would still be seventh best among active coaches with at least three tourney appearances…but a big slide from his current position atop that list. He’s in danger of becoming a one-dance wonder like Jim Larranaga, the former George Mason coach. Yo Shaka: better grab that better coaching offer now.

Jury Still Out

  • Billy Donovan
  • Sean Miller
  • John Beilein
  • Bo Ryan

Given the high win expectations for one and two seeds, it’s too early to tell where these four coaches will wind up. Both Donovan and Miller would have to get to the finals to improve their PASE values. Due to his #1-ranked +.775 PASE among veteran coaches, Beilein needs one more win to improve on his numbers. Bo Ryan is a different story. Because his career PASE is just +.063 and he’s already +.594 games above expectations, his numbers will surely increase. Whether it’s a big or small increase hinges on winning at least one more game.

Posted in Coach Ratings | 1 Comment

Basic champ check could fail for first time in 14 years

With just three teams who meet basic champ criteria left in the dance—and all of them slated to play each other before the finals—the streak of 13 straight predictions is in jeopardy. Heading into the dance, there were six squads that met these criteria:

  • A one, two or three seed
  • Member of a Power conference: ACC, Big East, Big Ten, Big 12, Pac-10 or SEC
  • Either went to the previous year’s dance or have an All-American
  • Led by a coach with more than five tourney trips and at least one Elite Eight run
  • Averaging more than 73 points per game
  • Allowing fewer than 73 points per game
  • An average scoring margin of at least seven points per game
  • A schedule among the 75 strongest in the country

The teams were Arizona, Villanova, Kansas, Wisconsin, Michigan and Duke. Only the Wildcats, Badgers and Wolverines remain (sounds like National Geographics death match). The odds of the basic champ check working again are diminished by the fact that Arizona plays Wisconsin tonight…and the winner would play Michigan (if it got there) in the Final Four. One thing is for sure: at least one team with basic champ cred will be alive next week.

Next year, I’ll probably do a combination of basic and advanced champ criteria. The fact is, all of the remaining teams have schedules among the 75 strongest in the country. Seven of the eight teams have an average scoring margin better than seven points a game (only Dayton falls short). Six of eight also have coaches with the requisite experience and success (only Archie Miller and Kevin Ollie are lacking). And only Kentucky, UConn and Dayton didn’t go to the previous tourney. If Florida or UConn don’t win the dance, the point-scoring threshold also remains intact. I will probably ditch the seeding requirement and replace it with KenPom offensive and defensive efficiency rankings.

Speaking of the advanced champ tests, three teams pass at least one of them. Remember: since 2003, the worst a champion has performed on offense in terms of points per 100 possessions is 112.2. That number belongs to last year’s champ Louisville. And the worst a champ has been ranked on offense is 18th (both Louisville and UConn in 2013). On defense, the most points a champ has allowed per 100 possessions is 95.4; that would be North Carolina in 2009—and they were ranked 49th in the country (of course, there offensive efficiency was through the roof).

Arizona and Florida meet the requirements for OE above 112.2 and DE below 95.4. And when it comes to the ranking comparison (top 18 offensive, top 49 defense), Florida and Michigan State make the grade.

To summarize, five teams meet at least one of the champ checks. Arizona and Florida meet two of them, and Wisconsin, Michigan and Michigan State meet one. That means if Dayton, Kentucky or UConn win the tourney, all bets are off… and the champ criteria will need a major overhaul.

Posted in Champ Credentials | 47 Comments

No clear pool winner among 16 models

After the Sweet 16, here’s where the models stand.

  • Coaching+Efficiency 93.3% – ARIZONA
  • Nate Silver 87.8%
  • Baseline 84%
  • Keeper 73.5% – FLORIDA
  • Pulse Check 73.5%
  • Upset/Toss-Up 73.5% – FLORIDA
  • Seed Match-ups 73.5%
  • From the Gut 63.3% – FLORIDA
  • Final Four/Champ 63.3% – ARIZONA
  • Contrarian 41.9%
  • KenPom 41.9%
  • ESPN BPI 37.0% – ARIZONA
  • Brackomatic 28.3% – MICHIGAN STATE
  • Outcome Match 24.5%
  • Factor PASE 18.1%
  • Billion Dollar 15.5% – FLORIDA

The top model heading into the Elite Eight is Coaching+Efficiency, which uses a combination of coaching performance and KenPom numbers to arrive at bracket picks. The model retains three of its four Final Four teams, (Florida, Arizona and Michigan State), losing a shot only at the Midwest, where Louisville fell.

The Keeper bracket is the next closest model with a champion still in the running. It could leap into the top spot with a Gator championship. Eight of the models still have their champ in the running, as noted by the CAPS above. Four of them like Florida, three Arizona and one MSU.

Interestingly, three of the four best models from last year are among the worst this year. ESPN BPI, Outcome Matching and KenPom ranked second through fourth in 2013. This year, not so good. Nate Silver fans will be happy to know that he had the best model of 2013 and has the second best model of 2014. Of course, Nate picked Louisville as his champion, so his chance for a big percentile jump is limited.

If I focus on the brackets done by a system, the average performance is 52.7%. That ties for the second worst tourney in the ten years I’ve posted models. Only the craziness of 2011’s UConn run was worse. I suppose Coaching+Efficiency and the Keeper bracket might have you in the running for your pool, depending on who cuts down the nets. But odds are, none of these models are currently “pool-win worthy.”

Posted in General News | Leave a comment

Stephenson on Coach K and late-career slumps

This email from Insider Tom Stephenson is worth posting in full. Great observations on how many coaches tend to underachieve as they near retirement. Tom mentioned some legends who retired at the top of their tourney game. I’ll add Rollie Jim Calhoun and Rollie Massimino.

Tom followed up this email with another great observation: “There will now be at least four Rookie coaches in the Sweet 16: Archie Miller, Kevin Ollie, Johnny Dawkins, and  Cuonzo Martin. Brad Underwood could make it five, though it doesn’t look likely. At least one coach (either Miller or Dawkins) will make the Elite Eight.”

Thanks Tom!

Coach K certainly isn’t the first legendary head coach to endure a late-career slump.  Look at the following career arcs (obviously, I’m using some pre-1985 tournaments here, so there’s no way to calculate PASE): 

  • Bob Knight, through 1987: 11 trips, 27-8 record, 6 E8, 3 CH
  • Bob Knight, 1988-93: 6 trips, 11-6 record, 2 E8, 0 CH
  • Bob Knight, since 1994: 11 trips, 7-11 record, 0 E8

Knight basically averaged an Elite Eight every other trip up through 1987, and a championship roughly every four trips; that’s about as good a tournament record as any coach not named Wooden.  Starting in 1988, the program began to slip (one of the Elite Eights was a loss as a 1 seed and was technically underachieving), then he really started to struggle in the tournament after 1993.  Knight was 53 in 1993, or four years younger than Coach K when he began slipping.

  • Denny Crum, through 1986: 12 trips, 27-12 record, 6 E8, 2 CH
  • Denny Crum, after 1986: 11 trips, 15-11 record, 1 E8

Like Knight, Crum was averaging an Elite Eight every other year up through his second championship season.  But he only made one more Elite Eight after that.  The overall record isn’t awful, but Louisville was never really a championship threat after 1986 and toward the end of Crum’s career started struggling to make the tournament (much less make deep run.)  Crum was 49 in 1986.

  • Adolph Rupp, through 1958: 10 trips, 20-6 record, 4 CH
  • Adolph Rupp, after 1958: 10 trips, 10-12 record, 0 CH

I’m not going to bother with Rupp’s Elite Eights since the tournament was so small back then (in fact, in Rupp’s first four tournaments there were only eight teams.)  And if you exclude the famous 1966 championship game run, Rupp only went 7-11 after his final national championship.  What’s more, Rupp’s teams started struggling to win the SEC; after winning or sharing 13 of 14 SEC championships from 1943-58 (in 1944 no champion was crowned, and Kentucky didn’t field a team in 1953), the Wildcats only won three from 1959-67; they made two more tournament appearances in 1959 and 1961 as a result of SEC champ Mississippi State being barred by state law from playing in the tournament.  They did recover to win the SEC in Rupp’s last five years but didn’t make the Final Four in that stretch.  Rupp was 56 when he won his last championship.

Those aren’t the only examples; there are probably plenty more, and there are also probably some who struggled to even make the tournament late in their career.  And of course some legends (like Dean Smith, for example) don’t go through a late career slump, or like Jim Boeheim they’re as up and down as they’ve always been.  But what Duke is currently going through is hardly unusual for a program coached by an all-time great nearing retirement.

Posted in Coach Ratings | 8 Comments

Round 2, Day 2 Forum

Credit Stanford for a great game plan–and blame Wiggins all you want for a poor game. But Bill Self did not do his star any favors by making him hang out on the wing, out of the action. He could’ve made Wiggins the post-flash player. And shame on Self for not putting Frankamp back into the game until late in the second half. Here’s a kid who sparked the Jayhawks in the first half…then rode the bench to the last two minutes of the game–when he hit two key threes to keep Kansas in the hunt. Needless to say, Self’s PASE will go down. And that’s fitting.

Posted in Bracket Forums | 18 Comments

The Five Stages of Bracket Grief

  1. Denial and Isolation: “What? Duke lost? No way. That’s got to be wrong.”
  2. Anger: “Damnit! Why the Hell did I pick them in the Final Four anyway!? Coach K did this on purpose to mess with me!”
  3. Bargaining: If only I’d factored in Duke’s ten-year PASE and watched a few more of their games.”
  4. Depression: “Aw, what’s the use? I’ll never figure out this stupid bracket thing.”
  5. Acceptance: “Oh well, crazy sh*t happens. Where’s that link to the round-by-round challenge?”
Posted in General News | 19 Comments

A decade of Duke woes…

With its startling loss to Mercer, Duke has now fallen short of seed-projected win totals in eight of 10 tourneys. And they’ve added to their record as upset victims, with 11 crushing losses. The next most shocked team is Oklahoma, with nine upsets.

I did a quick calculation, and Coach K’s PASE will go down to around +.304. That’s still overachieving…but it’s just 21st best of the 99 active coaches with three or more appearances.

More alarming: in the last ten years, Coach K’s PASE is a woeful -.810–nearly a game per dance below expectations. That would rank 97 out of 99, better than only Frank Haith and Oliver Purnell. Ouch! Blue Devil nations is hurting…

Posted in Measuring Madness, Upset Spotting | 4 Comments

Round 1 Forum, Friday games: “POP” go the brackets!

Pop! That was the sound of millions of brackets getting busted across the land–including my Keeper bracket. The is the eighth year in 10 tourneys that Duke has failed to meet seed expectations. And Duke is the all-time victimized school, with Cinderellas knocking them off in 29 years. Recent crushing losses include:

  • 2014 – 3v14 loss to Mercer
  • 2012 – 2v15 loss to Lehigh
  • 2008 2v7 loss to West Virginia
  • 2007 6v11 loss to VCU

For anyone thinking that All-American Jabari Parker would be the difference in this game, he was. Only in reverse. Duke played better when Parker wasn’t on the floor.

Posted in Bracket Forums | 47 Comments

Round 1 Forum, Thursday Games

Member BD suggested that I open a forum to discuss the results of each round. Good idea. So far, three models have pegged the two upsets: F4/Champ, Seed Match-ups and Upset/Toss-up. The Billion Dollar bracket is done. That was fun. Right now, the Madometer reads 46.4%. Crazy mad. But it’s early.

Posted in Bracket Forums | 21 Comments