Is there such a thing as "good tournament coach"?

[BoiledSteel]

First you say CMP has performed to the seedings, then you pick 4 of his 9 tourneys and blame it on being under seeded? Pretty strange interpretation. For example, it is very hard for me to see why our 2011 3-seed should have been a 2-seed, we got handled by VCU in the round of 32 (they had a great tourney that year but a 3 seed should translate to a sweet 16 appearance). The evaluation of "coaching performance" based on seed may be unfair or even utter nonsense, but it is a stastically derived estimate of performance compared to all other NCAA teams. It's impossible to meaningfully talk about performance without numbers to back it up.

That is the problem with posters like 35, nag, 4Purdue, et al. They simply don't post with any sort of logic. And when presented with factual data, they pick and choose what they want to admit to or ignore.

They have buckets of egg on their faces right now since they have been shown to be wrong about everything they touted about all season long, so really they simply cannot be taken seriously when they make comments anymore.

 

[tjreese]

How so?

If a model had little error ...accounting for the sources of variation making up almost the entire sum of squares., people would bet the he'll out of it against people that didn't understand it. Obviously it isn't precise (concerning which team should win) and in any form of multiple regression or secondary relationships, the correlation goes to he'll quickly. However, in my original post I said it would be interesting to see the model and assumptions..
And perhaps some day for my curiosity I'll investigate it.

 

[BoilerStation]

I would argue that Painter is getting close to everything he can from his current roster of guys. That makes him a good to great coach. The next level is just stepping up the recruiting and ceiling for the players a little bit. Cal, Coach K, Self and Roy Williams are getting nowhere near the max level of potential out of their players. We will take the next step in the tourney when we increase the overall potential of our players such as getting more athletes who can turn it on come tourney time and take the game to another level. We are slowly trending in this direction. Too many people are worrying about the final score of the Kansas game and how will we ever beat them which is silly. You don't assemble a team to beat one other team at a specific game style. A couple of losses in a row in previous years and we act like Painter is the worst tourney coach of all time. The VCU loss stunk but losing RH like we did and dismissing Barlow right before the tourney started had a huge impact on the team

 

[Dryfly88]

If a model had little error ...accounting for the sources of variation making up almost the entire sum of squares., people would bet the he'll out of it against people that didn't understand it. Obviously it isn't precise (concerning which team should win) and in any form of multiple regression or secondary relationships, the correlation goes to he'll quickly. However, in my original post I said it would be interesting to see the model and assumptions..
And perhaps some day for my curiosity I'll investigate it.

I'm just curious how that would apply to a point spread? That's why I asked how this model would help you predict margin of victory? Unless I'm just misunderstanding (completely possible).

I've been known to bet a penny or two on college b-ball, so when you inferred this might help someone win bets on the game I was curious.....

 

[tjreese]

I'm just curious how that would apply to a point spread? That's why I asked how this model would help you predict margin of victory? Unless I'm just misunderstanding (completely possible).

I've been known to bet a penny or two on college b-ball, so when you inferred this might help someone win bets on the game I was curious.....

I'm speaking in generalities since I do NOT know the assumptions and model used to predict who should win and by how much...and I have no intention to insult anyone with anything. If the simplist correlation between two variables for example were .7 then the explained variation would be .49 (r**2)or like flipping a coin. When you take a correlation of a correlation your ability to predict obviously loses some power.

Back in the mid 80s I used to calculate a lot by hand, but when I got into multiple regression I had to learn how to use SAS and could study things using forward and backward stepwise. FWIW in studies on humans...psychological studies a correlation of .7 would probably be considered great...think about that . I was pretty good at this stuff over 30 years ago, but have lost a lot of my edge. Today, I just have a little knowledge of what I did when I used this stuff a lot. Google will find my two most enjoyable teachers..both with Purdue backgrounds...Charlie Hicks http://www.purdue.edu/uns/html3month/2001/011023.Nichols.Hicks.html and Robert McLean http://www.stat.purdue.edu/giving/bob_and_marjorie_mclean_scholarship.php My work has been industrial and the statistics catered to that area. However, in the early days Bob (McLean) worked with Bill Sanders in developing TVAAS (I have the original 1983? study that Bob gave me). TVAAS was a statistical method to more properly evaluate "student gain" whereas most statistical methods would be biased if a teacher always had better students most likely due to classroom requests. "TVAAS is primarily" in Tennessee ( http://www.mosteffectiveschools.org/sch/about.htm )

I used to be involved in education and had the IDOE remove data after I pointed out their bias not supported by their own data on reduced lunches and such. Anyway, I'm not very good anymore because I've slept a lot of nights since 1985

 

[mathboy]

I have often wondered if the age consistency of the team has more to do with their performance in the tournament than raw talent. Yes, we have seen freshman stars ignite a team, but their surrounding cast is often upperclassmen. My speculation has to do with the stabilizing influence of much of the team being seniors or juniors.

It certainly seems to fit our pattern in some respects. We stubbled badly with freshmen and sophomores being the core of our team. Yet this year, as that class matured, we did much better, Kansas game not withstanding. I am taking a thousand foot view of this situation and generalizing, which is always prone to perception errors. However, I thought it might be interesting to chew on this idea.

 

[tjreese]

I have often wondered if the age consistency of the team has more to do with their performance in the tournament than raw talent. Yes, we have seen freshman stars ignite a team, but their surrounding cast is often upperclassmen. My speculation has to do with the stabilizing influence of much of the team being seniors or juniors.

It certainly seems to fit our pattern in some respects. We stubbled badly with freshmen and sophomores being the core of our team. Yet this year, as that class matured, we did much better, Kansas game not withstanding. I am taking a thousand foot view of this situation and generalizing, which is always prone to perception errors. However, I thought it might be interesting to chew on this idea.

***************
It sounds reasonable to me. I think both talent (if it could be accurately measured in a team setting) and experience (if it could be accurately measured since number of games played may not be an indicator) would both be factors in determining the success. I wonder if the relationships between "talent" and "experience" might not be linear...kinda like a diminishing returns type thing (maybe requiring three levels of study). It would seem to me though that the "interaction" between the two would be a better predictor than the main effects of each separately. In other words, I think the most talented measure combined with the most experience possible (is there any data due to those players leaving before full experience?) would provide one response and the least of those two would be another response. The differences between those responses (interaction) would be greater than the difference between most talent and least or most experience and least by themselves .

 

[cgnops]

...VCU loss stunk but losing RH like we did and dismissing Barlow right before the tourney started had a huge impact on the team

This is a good point and suggests that an argument could be made that we were over seeded that year as a result of these circumstances.

I'm speaking in generalities since I do NOT know the assumptions and model used to predict who should win and by how much...

The 538 piece just took the results of NCAA tourney games from 1985-2015. From those distributions the average win for a seed and variance can be calculated. They then add up a coaches wins and compare with the sum of the "expected (average) wins". In this way a coach is reduced to a number labeled WAE, "wins above expectation". it would be nice to see the full distribution of WAE for coaches with X number of appearances but this was not provided by the 538 articles I found, just tabulated list of best and worst performers.

Nevertheless its a nice result in that it shows Painter has been much better than the people arguing he isn't good enough to coach here. And as we have noted countless, who knows what would have happened if Robbie stayed healthy. If anything it suggests we should still be optimistic about Painters ceiling.