PDA

View Full Version : NCAA Tournament Success by Coach Since 2000 vs KenPom and Seed Expectation



DukeTrinity11
03-05-2019, 12:19 PM
I ran into this interesting table compiled by Bart Torvik that evaluates the success of all Division 1 Tournament coaches in the NCAAT compared to their KenPom and NCAA Seed Expectation going into the tournament.

http://barttorvik.com/cgi-bin/ncaat.cgi?type=coach&sort=1&yrlow=2000&yrhigh=2018

Ole' Roy is #1 in both followed by Izzo and Cal while K is #340 in the KenPom ranking and #358 vs Seed Expectation.

This is really eye opening to me because it shows how many missed opportunities we've had over the past 2 decades to add to Coach K and Duke Basketball's legacies.

Are there any takeaways from this as to why our brand of basketball isn't translating to the tournament given expectations while our rival is outperforming expectations constantly despite also getting a very high seed each year?

For what its worth, UVA's Bennett sits dead last amongst all Division 1 coaches in both rankings. I guess losing to a 16 seed does that to you...

left_hook_lacey
03-05-2019, 12:23 PM
I ran into this interesting table compiled by Bart Torvik that evaluates the success of all Division 1 Tournament coaches in the NCAAT compared to their KenPom and NCAA Seed Expectation going into the tournament.

http://barttorvik.com/cgi-bin/ncaat.cgi?type=coach&sort=1&yrlow=2000&yrhigh=2018

Ole' Roy is #1 in both followed by Izzo and Cal while K is #340 in the KenPom ranking and #358 vs Seed Expectation.

This is really eye opening to me because it shows how many missed opportunities we've had over the past 2 decades to add to Coach K and Duke Basketball's legacies.

Are there any takeaways from this as to why our brand of basketball isn't translating to the tournament given expectations while our rival is outperforming expectations constantly despite also getting a very high seed each year?

For what its worth, UVA's Bennett sits dead last amongst all Division 1 coaches in both rankings. I guess losing to a 16 seed does that to you...

May the odds forever be in your favor....

AGDukesky
03-05-2019, 12:25 PM
Thank you for sharing- I was just thinking of a way to try to capture this type of comparison. Now I will have to try to breakdown the scoring..

CDu
03-05-2019, 12:44 PM
Well, several things. We've been a 1, 2, or 3 seed in nearly every season over that span. Seriously. We've been a 1 or a 2 seed in 15 of 18 seasons, and a 3 seed once, a 4 seed once, and a 6 seed once. It's hard to overachieve in the tournament when "overachieving" requires making the Final Four. On the flipside, an early exit as a high seed has substantial negative impact on your standing.

Conversely, if you are a 7 seed or lower, it's easier to accumulate a big gain and hard to really underperform to any meaningful degree.

So a big part of it is just that: we haven't had much opportunity for upward mobility but have had lots of chances for downward mobility.

Beyond that, "the program" has not been some monolithic thing, so there isn't any single reason innate to the program. Up through 2010, we were one type of program (generally the 3-4 year guys). From 2011-2014, we were another type (3-4 year guys with a one-and-done mixed in). From 2015 onward, we've been largely driven by the one-and-dones with a couple of 3-4 year guys mixed in. So the reasons for earlier-than-hoped losses is different for those different eras.

From 2000 through 2004 (and actually spanning back to 1998), all of our tourney losses were to a team that won the title or made the title game. The 2005 and 2006 teams were overachieving veteran squads with no scoring depth that eventually ran into a team that did enough to limit Redick and Williams to pull upsets, and both of those losses were to Final Four teams. The 2007-2009 teams were younger, less athletic teams that got bounced by by athletic and physical opponents (in 2009, that opponent made the Final Four). The 2011 team dealt with the reintroduction of Irving and ran into a hot opponent. The 2012 team was beaten in the first round by an underseeded opponent who had more experience and the best individual player on the court (future NBA star CJ McCollum). The 2013 team lost in the Elite 8 to the #1 team who eventually won the title now vacated. The 2014 team was a very inexperienced and awful defensive team that was picked apart by a veteran squad. The 2016 team lost to the #1 seed who made the Final Four. The 2017 team lost essentially a true road game against a team that got hot and made the Final Four. And the 2018 team lost in the Elite 8 to the 1 seed.

Some of it was just bad luck. As you can see, 11 of our 16 losses came to teams that made the Final Four or better. The reasons for the other losses are widespread. One of the other losses came to a team with the best player on the floor (2012). Two others were because we were just a woefully inexperienced team (2007, 2014). Another was a combination of "hot hand" and us trying to regain our chemistry with the return of a ball-dominant PG (2011). The other was to a team that was just way more athletic and physical than us (2008).

gam7
03-05-2019, 12:48 PM
I ran into this interesting table compiled by Bart Torvik that evaluates the success of all Division 1 Tournament coaches in the NCAAT compared to their KenPom and NCAA Seed Expectation going into the tournament.

http://barttorvik.com/cgi-bin/ncaat.cgi?type=coach&sort=1&yrlow=2000&yrhigh=2018

Ole' Roy is #1 in both followed by Izzo and Cal while K is #340 in the KenPom ranking and #358 vs Seed Expectation.

This is really eye opening to me because it shows how many missed opportunities we've had over the past 2 decades to add to Coach K and Duke Basketball's legacies.

Are there any takeaways from this as to why our brand of basketball isn't translating to the tournament given expectations while our rival is outperforming expectations constantly despite also getting a very high seed each year?

For what its worth, UVA's Bennett sits dead last amongst all Division 1 coaches in both rankings. I guess losing to a 16 seed does that to you...

For teams consistently ranked highly in KenPom or with consistently high seeds, it is extremely difficult to exceed the seed expectation. In fact, if you are #1 ranked in KenPom and have a #1 seed expectation, it would be impossible to exceed expectations. Duke is in the tier of schools that have consistently been ranked near the top in both categories, so this would clearly be a factor. I'd put UVA in that category too (although it's really been the last 5-10 years where they have been in the stratosphere).

I suspected Kevin Ollie would be high on the list, and sure enough he's top 10 - UConn won the NCAA tourney under his watch as a 7-seed I believe. But he only coached 6 years. This list has some flaws...

fidel
03-05-2019, 12:50 PM
There's K...right below Sendek.

whoa. :confused:

There are a lot of one-hit wonders (and two-hits), but that's still pretty far down. One could argue that everyone brings their best game, but still.

Ian
03-05-2019, 12:58 PM
Let's not sugarcoat this, K has underachieved in the tourament since 2000. We can debate why this is without ignoring facts. Maybe it's just bad luck, maybe it's we are getting screwed by the commitee and put in more difficult brackets/matchups. Maybe because we overachieve in the regular season. Maybe it's because we have a target on our chest and teams get up more for us. Maybe K's short bench leads to players worn out by the end of the season. All plausible explanantions.

brlftz
03-05-2019, 12:58 PM
I expected to hear about the "we're always highly ranked so it's hard to over achieve" factor, but then how do you explain the over achievement of Ol' Roy, Izzo, and Cal?

Acymetric
03-05-2019, 12:59 PM
There's K...right below Sendek.

whoa. :confused:

There are a lot of one-hit wonders (and two-hits), but that's still pretty far down. One could argue that everyone brings their best game, but still.

I wonder what the numbers would look like if you could remove the Lehigh and Mercer years. I'm sure Bennett took a huge hit because of UMBC last year (literally last place).

Troublemaker
03-05-2019, 01:07 PM
Many of the coaches towards the bottom are just coaches who overachieve during the regular season. Some of the high seeds that Mike Montgomery achieved with his unathletic Stanford teams, for example, or Duke getting 2-seeds in 2008 and 2009 despite playing Singler and Thomas at center. Those teams are ripe for upset come tournament time.

But yeah, part of it is that Coach K hasn't been a great tournament coach since the mid-90s.

Troublemaker
03-05-2019, 01:34 PM
Let's keep this in perspective, too.

In the timeframe mentioned, Coach K is tied for the most national championships with Roy Williams at 3. (If you go back one more year, then Jim Calhoun joins the club).

So, essentially, one could argue that only UNC fans would not trade their coach for Coach K during that timeframe. (But then you get into that pesky issue that Duke and UNC are rivals, and Coach K has the head-to-head win-loss edge on Roy).

CDu
03-05-2019, 01:42 PM
In fact, if you are #1 ranked in KenPom and have a #1 seed expectation, it would be impossible to exceed expectations.

This isn't quite true. Even a team ranked #1 with a 1 seed would not be expected to win it all mathematically. That team would be expected to win, probably about 4 or 4.5 games. So you can indeed overachieve even as a 1 seed and even as the #1 overall. You just can't overachieve by much.

Where the model falls apart is that one good run as a 7-seed or worse basically blows up the model. For example, Calipari's numbers are largely inflated by making it to the title game as an 8 seed back in 2014. That gave him an extra ~4-4.5 wins above expectation. Boeheim's numbers are inflated by two Final Four runs as a lower seed.

You'd almost have to do this as a logistical model rather than a model of win counts relative to expectation. Because it is much harder to accrue extra wins against expectation when you are a high seed than it is when you are a lower seed.

That being said, I'd absolutely agree with those who note that Calipari and Williams have been better tournament coaches over the past 19 years than Coach K. Those coaches have generally not had the early exits in the tournament that Duke has had. I just don't think the difference is nearly as much of a difference as is being suggested by this type of analysis.

And, as Troublemaker notes, we do still have 3 titles, a Final Four, and two Elite 8s in that time frame.

Troublemaker
03-05-2019, 01:48 PM
Let's keep this in perspective, too.

In the timeframe mentioned, Coach K is tied for the most national championships with Roy Williams at 3. (If you go back one more year, then Jim Calhoun joins the club).

So, essentially, one could argue that only UNC fans would not trade their coach for Coach K during that timeframe. (But then you get into that pesky issue that Duke and UNC are rivals, and Coach K has the head-to-head win-loss edge on Roy).

In case anyone wonders how both the bolded above and Coach K's ranking of #340 can exist at the same time, it's basically just that this statistical analysis (arguably) doesn't weigh national titles enough. (I haven't dug into it, but I doubt it does ANY weighing of specific outcomes.)

Also, essentially, Duke has been an "all or nothing" type tournament team in the past two decades. We either win it all (3 times) or we get upset early. The difference between post-2000 Coach K and early career Coach K is that we used to make Final Fours even in non-title years. Since 2000, we've "only" made 4 Final Fours, but thankfully our conversion rate of Final Fours into National Championships has been extremely high (75%).

freshmanjs
03-05-2019, 01:51 PM
There are other ways to look at how we have performed as a 1 seed. Overall, in that data set, 1 seeds made the final 4 29 out of 76 times, for 38%.

Duke in that time frame was a #1 seed 9 times (i think) and made the final4 4 times. So that's a bit more than normal. Of course, we lost in the sweet 16 several times as a #1 also.

Our performance as a #2 hasn't been great.

gam7
03-05-2019, 02:04 PM
This isn't quite true. Even a team ranked #1 with a 1 seed would not be expected to win it all mathematically. That team would be expected to win, probably about 4 or 4.5 games. So you can indeed overachieve even as a 1 seed and even as the #1 overall. You just can't overachieve by much.




What you say is right. My thinking was a bit more rudimentary. I'm not sure exactly how these rankings are calculated, but - the way I am thinking about it, if the analysis is whether you meet the expectation of your tourney seed, then it is irrelevant whether you win the championship or make the finals. The analysis on that element would end after the regional finals. The one seeds can only meet their seed expectation.

And if you are a 1-seed that is also #1 in kenpom, then the expectation is that you will finish as the #1-ranked team in the country.
So, if you win the championship, you can only reach the level of #1; there's no higher you can go. So, on those two elements, you can only meet expectations...

Tripping William
03-05-2019, 02:17 PM
In case anyone wonders how both the bolded above and Coach K's ranking of #340 can exist at the same time, it's basically just that this statistical analysis (arguably) doesn't weigh national titles enough. (I haven't dug into it, but I doubt it does ANY weighing of specific outcomes.)

Also, essentially, Duke has been an "all or nothing" type tournament team in the past two decades. We either win it all (3 times) or we get upset early. The difference between post-2000 Coach K and early career Coach K is that we used to make Final Fours even in non-title years. Since 2000, we've "only" made 4 Final Fours, but thankfully our conversion rate of Final Fours into National Championships has been extremely high (75%).

This prompted a thought-experiment (at least for me): As a Duke fan, if forced to choose, would you take K's first 19 years (1981 to 1999) or his second 19 years (2000 to 2018)?

Recall that if you take the former, you not only get seven Final Fours in nine years and back-to-back natties (plus two other title game losses), but you also get Concerned Iron Dukes, the 43-point loss to UVa, and the Back Surgery Season, among other things. If you take the latter, well, this thread has described that fairly well thus far.

Acymetric
03-05-2019, 02:19 PM
This isn't quite true. Even a team ranked #1 with a 1 seed would not be expected to win it all mathematically. That team would be expected to win, probably about 4 or 4.5 games. So you can indeed overachieve even as a 1 seed and even as the #1 overall. You just can't overachieve by much.

Where the model falls apart is that one good run as a 7-seed or worse basically blows up the model. For example, Calipari's numbers are largely inflated by making it to the title game as an 8 seed back in 2014. That gave him an extra ~4-4.5 wins above expectation. Boeheim's numbers are inflated by two Final Four runs as a lower seed.

You'd almost have to do this as a logistical model rather than a model of win counts relative to expectation. Because it is much harder to accrue extra wins against expectation when you are a high seed than it is when you are a lower seed.

That being said, I'd absolutely agree with those who note that Calipari and Williams have been better tournament coaches over the past 19 years than Coach K. Those coaches have generally not had the early exits in the tournament that Duke has had. I just don't think the difference is nearly as much of a difference as is being suggested by this type of analysis.

And, as Troublemaker notes, we do still have 3 titles, a Final Four, and two Elite 8s in that time frame.

Actually, by my count (which spans 33 years but I'm not sure exactly which 33 years, it definitely doesn't include 2018 so not the freshest data but still a good sample) it is closer to 3.4 wins expected for a #1 (or at least that is the average number of wins for a #1 seed over that 33 year span), a number we narrowly outperform.

We have outperformed the average wins for our seed 8 times since 2000: 2001, 2003, 2004, 2010, 2013, 2015, 2016, 2018 and underperformed in the other 11.


Here is a chart showing what I found (obviously 3, 4, and 6 don't mean much since they comprise only 4 years between the three seedings)



Seed
Duke Avg. Wins
National Avg. Wins


1
3.6
3.4


2
1.4
2.4


3
1
1.8


4
2
1.6


6
0
1



Overall, we would be expected to average 2.7 wins, and average about 2.4 (6 more wins over the past 19 years would put us above the average).

CDu
03-05-2019, 02:20 PM
I expected to hear about the "we're always highly ranked so it's hard to over achieve" factor, but then how do you explain the over achievement of Ol' Roy, Izzo, and Cal?

Izzo has a number of appearances as a low seed. He made the Final Four twice as a 5 seed and once as a 7 seed. That accounts for most of his wins above expectation. Izzo has had a slightly worse record with worse seeding. If this was done on a logistical model, I'd imagine that they would be pretty close.

Cal and Williams have been better tournament coaches during this span, no question. They've both achieved better winning percentages with similar seeding (slightly worse).

I think it's entirely appropriate to say that Coach K was a poor tournament coach from 2002-2009, and again from 2011-2014. Since we have committed fully to the one-and-done model, though, his results are pretty good: 1 title, 1 elite 8, 1 sweet sixteen as a 4 seed, all of which slightly or substantially exceeded win expectations. Just the one early exit in that span. But it's hard to make up for about a decade's worth of tourney struggles from when your team constantly overachieved all season.

CDu
03-05-2019, 02:22 PM
Actually, by my count (which spans 33 years but I'm not sure exactly which 33 years, it definitely doesn't include 2018 so not the freshest data but still a good sample) it is closer to 3.4 wins expected for a #1 (or at least that is the average number of wins for a #1 seed over that 33 year span), a number we narrowly outperform.

My 4-4.5 was just a guesstimate for the #1 overall team, not just any 1 seed.

dukelifer
03-05-2019, 02:23 PM
I ran into this interesting table compiled by Bart Torvik that evaluates the success of all Division 1 Tournament coaches in the NCAAT compared to their KenPom and NCAA Seed Expectation going into the tournament.

http://barttorvik.com/cgi-bin/ncaat.cgi?type=coach&sort=1&yrlow=2000&yrhigh=2018

Ole' Roy is #1 in both followed by Izzo and Cal while K is #340 in the KenPom ranking and #358 vs Seed Expectation.

This is really eye opening to me because it shows how many missed opportunities we've had over the past 2 decades to add to Coach K and Duke Basketball's legacies.

Are there any takeaways from this as to why our brand of basketball isn't translating to the tournament given expectations while our rival is outperforming expectations constantly despite also getting a very high seed each year?

For what its worth, UVA's Bennett sits dead last amongst all Division 1 coaches in both rankings. I guess losing to a 16 seed does that to you...
Despite what many on this board think- Roy is arguably among the best coaches of all time. This UNC team with Luke Maye as the leader has a legit chance at a National Championship. Roy does not have Ks presence and overplays his good ol boy persona- but he is as good as they come. If he wins another this year- it will be very hard to deny his greatness even by the most ardent critic.

BandAlum83
03-05-2019, 02:25 PM
Despite what many on this board think- Roy is arguably among the best coaches of all time. This UNC team with Luke Maye as the leader has a legit chance at a National Championship. Roy does not have Ks presence and overplays his good ol boy persona- but he is as good as they come. If he wins another this year- it will be very hard to deny his greatness even by the most ardent critic.

^^This^^ Plus he has all those excess timeouts that he can redeem after retirement. (If he isn't using them to buy hamburgers)

Acymetric
03-05-2019, 02:27 PM
My 4-4.5 was just a guesstimate for the #1 overall team, not just any 1 seed.

Sure, I knew you were just throwing a rough number out (when I actually ran the numbers I expected the average to be higher than it turned out to be for #1 seeds generally, was surprised at the result). Still, I would bet the average for #1 overall is no more than 4 and probably slightly less.


Izzo has a number of appearances as a low seed. He made the Final Four twice as a 5 seed and once as a 7 seed. That accounts for most of his wins above expectation. Izzo has had a slightly worse record with worse seeding. If this was done on a logistical model, I'd imagine that they would be pretty close.

Cal and Williams have been better tournament coaches during this span, no question. They've both achieved better winning percentages with similar seeding (slightly worse).

I think it's entirely appropriate to say that Coach K was a poor tournament coach from 2002-2009, and again from 2011-2014. Since we have committed fully to the one-and-done model, though, his results are pretty good: 1 title, 1 elite 8, 1 sweet sixteen as a 4 seed, all of which slightly or substantially exceeded win expectations. Just the one early exit in that span. But it's hard to make up for about a decade's worth of tourney struggles from when your team constantly overachieved all season.

I do think you have to admit that our path to the Sweet 16 in 2016 (as a 4 seed) was pretty easy, going through UNCW and 12 seed Yale (who upset Baylor). Essentially, we are slightly better than average as a #1 seed (although its a bit of all-or-nothing where we either win the tourney or bow out early), but significantly worse than average as any other seed (caveats about small sample sizes aside).

barjwr
03-05-2019, 02:33 PM
Despite what many on this board think- Roy is arguably among the best coaches of all time. This UNC team with Luke Maye as the leader has a legit chance at a National Championship. Roy does not have Ks presence and overplays his good ol boy persona- but he is as good as they come. If he wins another this year- it will be very hard to deny his greatness even by the most ardent critic.

I don't want this interesting thread to degenerate into a rehashing of the nonsense that occurred in the Dump on the Hump for all those years. That said, I have a difficult time including Roy Williams on a Mt. Rushmore of the NCAA's best when much of his success has been linked to institutional academic fraud for athletic gain, no matter what the objective data report on his tournament success.

Nugget
03-05-2019, 02:42 PM
Where the model falls apart is that one good run as a 7-seed or worse basically blows up the model. For example, Calipari's numbers are largely inflated by making it to the title game as an 8 seed back in 2014. That gave him an extra ~4-4.5 wins above expectation. Boeheim's numbers are inflated by two Final Four runs as a lower seed.

You'd almost have to do this as a logistical model rather than a model of win counts relative to expectation. Because it is much harder to accrue extra wins against expectation when you are a high seed than it is when you are a lower seed.

That being said, I'd absolutely agree with those who note that Calipari and Williams have been better tournament coaches over the past 19 years than Coach K. Those coaches have generally not had the early exits in the tournament that Duke has had. I just don't think the difference is nearly as much of a difference as is being suggested by this type of analysis.

And, as Troublemaker notes, we do still have 3 titles, a Final Four, and two Elite 8s in that time frame.

It's also worth noting that Kentucky's 2013-2014 team grossly underachieved in the regular season -- it was pre-season ranked #3. So, it's hard to say who did a "better" job that year between K and Cal: we were pre-season #4, had a solid, consistent year, finishing at #8 before the first round upset, whereas UK played itself all the way out of the rankings before going on a run in the tournament to end up where they were projected to have been in the first place.

Having said that, there's no denying that Coach K has suffered a much larger number of unexpected tournament losses than have Cal or Roy in the past 20 years.

mkirsh
03-05-2019, 02:45 PM
Isn't the converse of the thread the coaches who under-performed most in the regular season? Ie if Cal was such a great coach how did he manage to only get an 8 seed with a final four caliber team?

I don't really believe this, nor do I believe the initial thesis of some coaches being substantially better tournament coaches. In reality, I think this shows that college basketball is just a small sample size game, with only 30 regular season games and a single elimination 6 game tournament it is very easy to "over" and "under" perform seeding. If Grayson Allen's shot goes down last year Coach K moves up the list; if UNC is called for the charge against Arkansas in 2016 or Luke Maye misses the buzzer beater against UK in 2017 Coach Williams goes down the list - single elimination format produces highly variable results.

CDu
03-05-2019, 02:59 PM
Sure, I knew you were just throwing a rough number out (when I actually ran the numbers I expected the average to be higher than it turned out to be for #1 seeds generally, was surprised at the result). Still, I would bet the average for #1 overall is no more than 4 and probably slightly less.



I do think you have to admit that our path to the Sweet 16 in 2016 (as a 4 seed) was pretty easy, going through UNCW and 12 seed Yale (who upset Baylor). Essentially, we are slightly better than average as a #1 seed (although its a bit of all-or-nothing where we either win the tourney or bow out early), but significantly worse than average as any other seed (caveats about small sample sizes aside).

Well, the whole things small sample size driven, yes. Basically, this all boils down to 3 seasons over which we amassed a total of two tourney wins as a 2 seed. In the 13 years as 1, 3, 4, or 6 seed, we have played roughly to seed average overall. Slightly outperforming as a 1 and 4, slightly underperforming as a 3 and a 6. And in 3 of our 6 tries as a 2 seed we made the second weekend (twice exceeding expectations). But in those other 3 tries as a 2 seed, we got bounced early. And that crushed us, because it is really hard to make up those ~6 additional wins against expectation when you are almost always a 1 or 2 seed.

Acymetric
03-05-2019, 03:07 PM
Well, the whole things small sample size driven, yes. Basically, this all boils down to 3 seasons over which we amassed a total of two tourney wins as a 2 seed. In the 13 years as 1, 3, 4, or 6 seed, we have played roughly to seed average overall. Slightly outperforming as a 1 and 4, slightly underperforming as a 3 and a 6. And in 3 of our 6 tries as a 2 seed we made the second weekend (twice exceeding expectations). But in those other 3 tries as a 2 seed, we got bounced early. And that crushed us, because it is really hard to make up those ~6 additional wins against expectation when you are almost always a 1 or 2 seed.

Agreed. Of course the worse was 2012...in addition to the statistical impact it had, that loss had a big impact on the "gut feel" evaluation of our tournament performance in that time span (probably along with 2014). Losing in the first round in 2007 was disappointing but not exactly a gut punch given where that team was at (and the fact that VCU was probably underseeded).

CDu
03-05-2019, 03:09 PM
Isn't the converse of the thread the coaches who under-performed most in the regular season? Ie if Cal was such a great coach how did he manage to only get an 8 seed with a final four caliber team?

I don't really believe this, nor do I believe the initial thesis of some coaches being substantially better tournament coaches. In reality, I think this shows that college basketball is just a small sample size game, with only 30 regular season games and a single elimination 6 game tournament it is very easy to "over" and "under" perform seeding. If Grayson Allen's shot goes down last year Coach K moves up the list; if UNC is called for the charge against Arkansas in 2016 or Luke Maye misses the buzzer beater against UK in 2017 Coach Williams goes down the list - single elimination format produces highly variable results.

Right, also, it's important to note the arbitrary time point here. If we did this analysis from, say, 1985, Coach K would look MUCH better than this. And in fact, when this sort of analysis was done around 2001 or so, Coach K was blowing the model up. Did he suddenly forget how to coach in the tournament, or did the small sample size of overperformance just fluctuate the other way for about a decade or so?

The reality is that Coach K has a 94-29 record in the NCAA tournament, or a 76.4 winning %. He has a record of 1125-342 overall, or 76.7%. So, over that sample, he's performed about as well in the tournament as he has overall. I imagine if more coaches had a similarly long resume, their records would be no better than his. Roy Williams, for example, has an 866-232 overall record (78.9%) and a 77-25 tourney record (75.4%).

OldPhiKap
03-05-2019, 03:20 PM
I don't want this interesting thread to degenerate into a rehashing of the nonsense that occurred in the Dump on the Hump for all those years. That said, I have a difficult time including Roy Williams on a Mt. Rushmore of the NCAA's best when much of his success has been linked to institutional academic fraud for athletic gain, no matter what the objective data report on his tournament success.

This. Unless you take out the games they played with ineligible players, it's comparing apples and oranges.

Roy is a good coach, and has recruited excellent players (some of whom were not burdened with, you know, going to class and all). I forget when the poll of coaches came out that had him pegged "most overrated" by his peers, but there's something to that too.

As for K. he's been to 12 final fours and won 5 National Championships. I think he's okay. I disagree strongly with the assertion that "our brand of basketball isn't translating to the tournament given expectations." CDu, gam7 and others have expressed the mechanics of the fallacy better than I can do so just sign me up to those posts too.

If there is any "problem" here, it's not with the team -- it's with the expectations and misunderstanding of the gauntlet that is presented by the Tournament.

Ian
03-05-2019, 03:42 PM
This. Unless you take out the games they played with ineligible players, it's comparing apples and oranges.

Roy is a good coach, and has recruited excellent players (some of whom were not burdened with, you know, going to class and all). I forget when the poll of coaches came out that had him pegged "most overrated" by his peers, but there's something to that too.

As for K. he's been to 12 final fours and won 5 National Championships. I think he's okay. I disagree strongly with the assertion that "our brand of basketball isn't translating to the tournament given expectations." CDu, gam7 and others have expressed the mechanics of the fallacy better than I can do so just sign me up to those posts too.

If there is any "problem" here, it's not with the team -- it's with the expectations and misunderstanding of the gauntlet that is presented by the Tournament.

This doesn't make sense, the expectations that is being measured against here is the performance of your average NCAA head coach, not some arbitrarily high standard applying only to Duke.

Acymetric
03-05-2019, 03:45 PM
Right, also, it's important to note the arbitrary time point here. If we did this analysis from, say, 1985, Coach K would look MUCH better than this. And in fact, when this sort of analysis was done around 2001 or so, Coach K was blowing the model up. Did he suddenly forget how to coach in the tournament, or did the small sample size of overperformance just fluctuate the other way for about a decade or so?

The reality is that Coach K has a 94-29 record in the NCAA tournament, or a 76.4 winning %. He has a record of 1125-342 overall, or 76.7%. So, over that sample, he's performed about as well in the tournament as he has overall. I imagine if more coaches had a similarly long resume, their records would be no better than his. Roy Williams, for example, has an 866-232 overall record (78.9%) and a 77-25 tourney record (75.4%).

Starting from '84 (to present):



Seed
Duke Avg.
National Avg.


1
3.9
3.4


2
2.7
2.4


3
1.7
1.8


4
2
1.6


5
2
1.2


6
0
1.1


8
0
.7


Overall
2.8
2.5



As CDu suggested, the numbers look much better in this span.

Edit: Math error...updated our average as a 3 seed to be 1.7 (originally listed at 2, "accidentally" omitted 2014 in the original calc", no change to overall which was done correctly)

Ian
03-05-2019, 03:50 PM
Another explanation is that he's mellowed out in his older years, and no longer has the energy and intensity in his 60s and 70s as he had in 40s and 50s.

OldPhiKap
03-05-2019, 03:54 PM
This doesn't make sense, the expectations that is being measured against here is the performance of your average NCAA head coach, not some arbitrarily high standard applying only to Duke.

I will quote CDu:


Well, several things. We've been a 1, 2, or 3 seed in nearly every season over that span. Seriously. We've been a 1 or a 2 seed in 15 of 18 seasons, and a 3 seed once, a 4 seed once, and a 6 seed once. It's hard to overachieve in the tournament when "overachieving" requires making the Final Four. On the flipside, an early exit as a high seed has substantial negative impact on your standing.

Conversely, if you are a 7 seed or lower, it's easier to accumulate a big gain and hard to really underperform to any meaningful degree.

So a big part of it is just that: we haven't had much opportunity for upward mobility but have had lots of chances for downward mobility.

The "average head coach" has not been a 1, 2 or 3 seed nearly every season.

Wouldn't everyone including the haters agree that a system ranking Coach K #358 in anything is -- um -- a bit off?

Acymetric
03-05-2019, 03:58 PM
I will quote CDu:



The "average head coach" has not been a 1, 2 or 3 seed nearly every season.

Wouldn't everyone including the haters agree that a system ranking Coach K #358 in anything is -- um -- a bit off?

It depends on what it is trying to measure, which in this case turns out essentially to be "who is the best underdog". People who aren't underdogs will be rated low, but there is some insight to be gained from the data. I don't think it was ever purporting to actually evaluate who was the best tournament coach.

tbyers11
03-05-2019, 04:03 PM
It's also worth noting that Kentucky's 2013-2014 team grossly underachieved in the regular season -- it was pre-season ranked #3. So, it's hard to say who did a "better" job that year between K and Cal: we were pre-season #4, had a solid, consistent year, finishing at #8 before the first round upset, whereas UK played itself all the way out of the rankings before going on a run in the tournament to end up where they were projected to have been in the first place.

Having said that, there's no denying that Coach K has suffered a much larger number of unexpected tournament losses than have Cal or Roy in the past 20 years.

Cal has been a very good tournament coach by this metric over this time frame. However, he doesn't get dinged at all for not making the tournament in 2013. I know that you need an NCAA seed to determine how well you performed against that seed. But it seems that grading coaches over a time period should account negatively for not making the tournament at all.

I wonder what PASE is for NIT 1 seeds. Because losing that first round NIT game to Robert Morris that year certainly wasn't up to par. ;)

Ian
03-05-2019, 04:04 PM
I will quote CDu:



The "average head coach" has not been a 1, 2 or 3 seed nearly every season.

Wouldn't everyone including the haters agree that a system ranking Coach K #358 in anything is -- um -- a bit off?

The average performance of a head coach of a 1 seed, the average performance of a head coach of a 2 seed,The average performance of a head coach of a 3 seed, etc.

Acymetric
03-05-2019, 04:13 PM
This. Unless you take out the games they played with ineligible players, it's comparing apples and oranges.

Roy is a good coach, and has recruited excellent players (some of whom were not burdened with, you know, going to class and all). I forget when the poll of coaches came out that had him pegged "most overrated" by his peers, but there's something to that too.

As for K. he's been to 12 final fours and won 5 National Championships. I think he's okay. I disagree strongly with the assertion that "our brand of basketball isn't translating to the tournament given expectations." CDu, gam7 and others have expressed the mechanics of the fallacy better than I can do so just sign me up to those posts too.

If there is any "problem" here, it's not with the team -- it's with the expectations and misunderstanding of the gauntlet that is presented by the Tournament.

One totally cherry-picked comparison between a few coaches...first round losses. I'll note that unc has only lost once in the opening round since 1978 under any coach, in 1998 under Guthridge (obviously they missed the tourney some of those years as well).

First round losses as a head coach:

K: 4 (3 in the last 7 years - Seeds 2, 3, 6, 8)
Roy: 0 (Kansas or unc)
Calipari: 0 at Kentucky (1 at Memphis as a 7 seed to ASU, his first NCAA appearance)
Tom Izzo: 4 (2002 - 10 seed loss to #7 NC State, 2004 7 seed loss to #10 Nevada, 2006 6 seed loss to #11 George Mason, 2016 2 seed loss to Middle Tennessee)

I feel like that is kind of the traditional "big 4" of active coaches, anyone else I should look at? Needs to be someone with longevity, someone who is successful but only been with a big time program for a few years doesn't count.

OldPhiKap
03-05-2019, 04:35 PM
The average performance of a head coach of a 1 seed, the average performance of a head coach of a 2 seed,The average performance of a head coach of a 3 seed, etc.

So -- you agree that there are more than three hundred NCAA tournament coaches better than K since 2000? Has he been that much of a disappointment to you relative to expectations?

wobatus
03-05-2019, 04:44 PM
One totally cherry-picked comparison between a few coaches...first round losses. I'll note that unc has only lost once in the opening round since 1978 under any coach, in 1998 under Guthridge (obviously they missed the tourney some of those years as well).

First round losses as a head coach:

K: 4 (3 in the last 7 years - Seeds 2, 3, 6, 8)
Roy: 0 (Kansas or unc)
Calipari: 0 at Kentucky (1 at Memphis as a 7 seed to ASU, his first NCAA appearance)
Tom Izzo: 4 (2002 - 10 seed loss to #7 NC State, 2004 7 seed loss to #10 Nevada, 2006 6 seed loss to #11 George Mason, 2016 2 seed loss to Middle Tennessee)

I feel like that is kind of the traditional "big 4" of active coaches, anyone else I should look at? Needs to be someone with longevity, someone who is successful but only been with a big time program for a few years doesn't count.

Maybe not all opening round losses, because UNC had some byes those years, but Dean lost UNC's opening NCAA Tournament games 3 years in a row. 1978 (1st round to Bill Cartwright led San Francisco), 1979 (Black Sunday loss as a 1 seed in 2nd round to Final 4 bound UPenn and Tony Price; Duke lost to a Wayne McCoy led St. John's - McCoy being yet another great 1977 class recruit- that same day as a 2 seed, missing Bob Bender; both games in Raleigh, hence Black Sunday), and 1980 loss to Texas A&M, led by Rudy Woods, in OT.

Acymetric
03-05-2019, 04:44 PM
So -- you agree that there are more than three hundred NCAA tournament coaches better than K since 2000? Has he been that much of a disappointment to you relative to expectations?

I don't think anyone is claiming that, including Torvic. This board is cranky the last couple days. You're trying to make it out that this is measuring something that it isn't.


Maybe not all opening round losses, because UNC had some byes those years, but Dean lost UNC's opening NCAA Tournament games 3 years in a row. 1978 (1st round to Bill Cartwright led San Francisco), 1979 (Black Sunday loss as a 1 seed in 2nd round to Final 4 bound UPenn and Tony Price; Duke lost to a Wayne McCoy led St. John's - McCoy being yet another great 1977 class recruit- that same day as a 2 seed, missing Bob Bender; both games in Raleigh, hence Black Sunday), and 1980 loss to Texas A&M, led by Rudy Woods, in OT.

Fair enough, I should have stuck with the 64+ team era. I'll restate as: unc has only lost in the opening round once in the 64+ team era (1985).

Ian
03-05-2019, 04:58 PM
So -- you agree that there are more than three hundred NCAA tournament coaches better than K since 2000? Has he been that much of a disappointment to you relative to expectations?

No, because of the 340, most have too small a sample size for their numbers to matter. I also don't think Mason Plumlee is a great 3 point shooter just because he's 1-1 from the 3 point line.

But among the coaches who have had, say a minimum of 10 NCAA appearances in that span, Coach K has not been among the best in terms of meeting seed expectations. (I emphasize it's seed expectations, not *my expectations*. ) Now as I said before, there could be numerous reasons for this apart of his coaching. But let's first agree on the facts.

Ian
03-05-2019, 04:59 PM
I don't think anyone is claiming that, including Torvic. This board is cranky the last couple days. You're trying to make it out that this is measuring something that it isn't.



Fair enough, I should have stuck with the 64+ team era. I'll restate as: unc has only lost in the opening round once in the 64+ team era (1985).

Weber St?

CDu
03-05-2019, 05:04 PM
But among the coaches who have had, say a minimum of 10 NCAA appearances in that span, Coach K has not been among the best in terms of meeting seed expectations. (I emphasize it's seed expectations, not *my expectations*. ) Now as I said before, there could be numerous reasons for this apart of his coaching. But let's first agree on the facts.

The "in that span" is a key part of the story. That, and the model being used isn't really designed to answer the question very well. It's a model designed to better assess "who is the best underdog coach".

Acymetric
03-05-2019, 05:05 PM
No, because of the 340, most have too small a sample size for their numbers to matter. I also don't think Mason Plumlee is a great 3 point shooter just because he's 1-1 from the 3 point line.

But among the coaches who have had, say a minimum of 10 NCAA appearances in that span, Coach K has not been among the best in terms of meeting seed expectations. (I emphasize it's seed expectations, not *my expectations*. ) Now as I said before, there could be numerous reasons for this apart of his coaching. But let's first agree on the facts.

I wonder if it helps to look at it this way:

In that time span, we have underperformed against our seed expectations 11 times (10 by more than one game). We have overperformed 8 times (3 by more than one game, I'll let you all guess which three years those are ;)). In other words, in terms of tournament performance, we've essentially had 3 great years, 6 average years, 8 bad years, and 2 very bad years. Of course, most of those years were "good" years looking at the season as a whole, albeit with disappointing endings in several cases, but we aren't looking at the season, only at the tournament.

wobatus
03-05-2019, 05:29 PM
I don't think anyone is claiming that, including Torvic. This board is cranky the last couple days. You're trying to make it out that this is measuring something that it isn't.



Fair enough, I should have stuck with the 64+ team era. I'll restate as: unc has only lost in the opening round once in the 64+ team era (1985).

Just a bit of trivia. Lots of coaches, even the best, go through stretches like that. It took Wooden 14 years to win a meaningful NCAA tournament game (he won a west regional consolation game once). Whiffed in the tourney 3 times and missed it altogether many times even though ranked, before he broke through.

Kedsy
03-05-2019, 05:34 PM
Fair enough, I should have stuck with the 64+ team era. I'll restate as: unc has only lost in the opening round once in the 64+ team era (1985).

Restate again to say 1999, and you're cooking with gas.

Though UNC also lost as a #3 seed to a #11 seed in 2006. Not sure why losing to a double-digit seed in the second round would be less embarrassing than losing to a double-digit seed in the first round.

Acymetric
03-05-2019, 05:39 PM
Restate again to say 1999, and you're cooking with gas.

Though UNC also lost as a #3 seed to a #11 seed in 2006. Not sure why losing to a double-digit seed in the second round would be less embarrassing than losing to a double-digit seed in the first round.

1985 is the year we went to 64 teams. I cited 1999 in my original post. 😉

To your second point: meeting them in the second round is better because a top seed losing in the first round is just a really bad look. Losing in the second round (which we have also done) is also a bad look, but not AS bad.

Kedsy
03-05-2019, 05:45 PM
The other factor that this analysis doesn't take into account is "schedule strength." I'm pretty sure this analysis is based on who the team is supposed to play, but sometimes it doesn't work out that way. For example, in 2016, UNC as a #1 seed didn't play a team ranked better than #5 until it reached the championship game (which it lost). That was one of four years in the 2000 to 2018 timeframe that a Roy-coached team played two or more games against teams seeded two or more spots worse than "chalk." In those four such years, Roy won a natty, lost in the championship game twice, and made an Elite Eight, undoubtedly helping his overall profile. In the same time period, Coach K only had one season in which he played two or more games against teams seeded two or more spots worse than chalk (he won the natty).

Reverse that luck, which had nothing to do with coaching, and maybe we're not talking about this right now.

Acymetric
03-05-2019, 05:46 PM
Just a bit of trivia. Lots of coaches, even the best, go through stretches like that. It took Wooden 14 years to win a meaningful NCAA tournament game (he won a west regional consolation game once). Whiffed in the tourney 3 times and missed it altogether many times even though ranked, before he broke through.

Wooden/the old tournament format just isn't comparable (which is why I should have stuck to the modern 64 team era).

I'd argue that it is a little more than just trivia because it's a thing that happened which in part shapes the narrative of the last 15-20 years.

Skydog
03-05-2019, 07:35 PM
The other factor that this analysis doesn't take into account is "schedule strength." I'm pretty sure this analysis is based on who the team is supposed to play, but sometimes it doesn't work out that way. For example, in 2016, UNC as a #1 seed didn't play a team ranked better than #5 until it reached the championship game (which it lost). That was one of four years in the 2000 to 2018 timeframe that a Roy-coached team played two or more games against teams seeded two or more spots worse than "chalk." In those four such years, Roy won a natty, lost in the championship game twice, and made an Elite Eight, undoubtedly helping his overall profile. In the same time period, Coach K only had one season in which he played two or more games against teams seeded two or more spots worse than chalk (he won the natty).

Reverse that luck, which had nothing to do with coaching, and maybe we're not talking about this right now.
True dat. A more accurate analysis would be based on actual NCAA games played. So calculate Duke's expected wins based on our seed and our actual opponent seed (or on our KP rating vs opp rating). Repeat for all teams with 10 or more appearances. This would get rid of the lucky or unlucky path effects.

Acymetric
03-05-2019, 07:40 PM
True dat. A more accurate analysis would be based on actual NCAA games played. So calculate Duke's expected wins based on our seed and our actual opponent seed (or on our KP rating vs opp rating). Repeat for all teams with 10 or more appearances. This would get rid of the lucky or unlucky path effects.

Unlikely that will help Duke, as what drags us down is our first, second, and third round losses where we would be favored regardless of which seed we played. Those three first round losses are going to make us look bad no matter what method you use if you're looking at expected performance by seed.

In fact, this can pretty much only hurt because currently it more or less assumes the most difficult path. Factoring in that sometimes our path is easier (ie #12 Yale in the second round) is going to make things worse.

HereBeforeCoachK
03-05-2019, 10:08 PM
But among the coaches who have had, say a minimum of 10 NCAA appearances in that span, Coach K has not been among the best in terms of meeting seed expectations. (I emphasize it's seed expectations, not *my expectations*. ) Now as I said before, there could be numerous reasons for this apart of his coaching. But let's first agree on the facts.

The reason for this is that he has had so many fantastic regular seasons that it would be very tough for him to be among the best by this metric. This kind of metric is going to be won by people who have moderate to low expectations due to mediocre regulars seasons and then exceed those in the NCAAT. When mediocre teams meet expectation, say winning one round, it's nothing like when a great team merely meets expectations....by winning it all.