• Welcome to OGBoards 10.0, keep in mind that we will be making LOTS of changes to smooth out the experience here and make it as close as possible functionally to the old software, but feel free to drop suggestions or requests in the Tech Support subforum!

A Look at Team Recruiting vs. On-Court Success in Last Decade of College Basketball

Some pretty clear tiers in those graphs and we seem to be in the seventh tier.
 
pourdeac said:
Don't you mean the top AND the bottom teams? The correlation is basically a line between the two extremes, which makes sense. The best players are the best players leading to wins, and the worst ones are bad leading to lot of losses. I'm not sure there's a correlation out side of the extremes though.
Click to expand...

No, it looks like there is less correlation as you move down and left.

My opinion is that recruiting is the most important factor in determining success in college sports. However, I think recruiting rankings, which are compiled by 20 year-olds without enough skill to make it in coaching, are quite flawed (especially in football). I also think there is a self-fulfilling prophecy for some recruits (again, more so in football) where they are ranked highly because they were signed by a top team. There are so many examples of the recruiting "experts" all missing on incredible, all-star level talent. That's why I'm interested in the strength of correlation, regression line, etc.
 
This thread is an example of people using stats to support whatever argument that they want to make.

My take is that Gregg Marshall is a f-ing awesome coach. Huggy Bear, Few and Bo Ryan and Greg Gard are similarly fantastic.
 
Our coaching has sucked since Skip (and his final two years). That's what this confirms for me.
 
OldGoldBeard said:
I don't know whom else he brought in, but Gregory at least landed an All-ACC player in Lammers. Bzdoofus was here for four seasons, and the best we got were a couple of All-ACC Honorable Mention campaigns from CMM and Devin Thomas
Click to expand...

Not disagreeing with your point at all, but I think CJ Harris made an all ACC team and Tyler Cavanaugh turned out to be a very good player.
 
Rafi said:
Not disagreeing with your point at all, but I think CJ Harris made an all ACC team and Tyler Cavanaugh turned out to be a very good player.
Click to expand...

CJ was a Dino/Skip recruit. CJ started as a frosh on the last WF team (coached by Dino) to make the NCAA tourney before last year.
 
Pilchard said:
This thread is an example of people using stats to support whatever argument that they want to make.

My take is that Gregg Marshall is a f-ing awesome coach. Huggy Bear, Few and Bo Ryan and Greg Gard are similarly fantastic.
Click to expand...

Agreed. Do you think that is because they are good at identifying talent, developing players, or Xs and Os? I would guess Wisc has put 5 or so players in the NBA during that time and Wichita State probably 3 or so, so I think a good part of it is identifying talent and/or developing talent.

Pilchard said:
CJ was a Dino/Skip recruit. CJ started as a frosh on the last WF team (coached by Dino) to make the NCAA tourney before last year.
Click to expand...

Good point. He played for Bz but was not a Bz recruit.
 
Rafi said:
Not disagreeing with your point at all, but I think CJ Harris made an all ACC team and Tyler Cavanaugh turned out to be a very good player.
Click to expand...

CJ Harris was recruited and played for Gaudio

Cavanaugh turned out to be a very good player...elsewhere

According to his player profile, Cavanaugh was part of the 21st-best recruiting class in the country, which again supports the theory that we've underachieved relative to recruiting rankings
 
Pilchard said:
This thread is an example of people using stats to support whatever argument that they want to make.

My take is that Gregg Marshall is a f-ing awesome coach. Huggy Bear, Few and Bo Ryan and Greg Gard are similarly fantastic.
Click to expand...

BINGO!!!
 
Rafi said:
No, it looks like there is less correlation as you move down and left.

My opinion is that recruiting is the most important factor in determining success in college sports. However, I think recruiting rankings, which are compiled by 20 year-olds without enough skill to make it in coaching, are quite flawed (especially in football). I also think there is a self-fulfilling prophecy for some recruits (again, more so in football) where they are ranked highly because they were signed by a top team. There are so many examples of the recruiting "experts" all missing on incredible, all-star level talent. That's why I'm interested in the strength of correlation, regression line, etc.
Click to expand...
Oh, there is undoubtedly heteroskedasticity in the data. The recruiting rankings were definitely less reliable for lower quality recruits - the difference in a player ranked 350 vs. 400 is likely not the same as the difference in a player ranked 1 vs. 51. There was likely more information that went into the higher rankings being assigned than the lower rankings. There were also more NAs for lower quality teams. That is likely why you see more variability as recruiting rank increases. I can get a correlation for you if you'd like, though.

There is also the issue of taking the average of an ordinal variable and then transforming this average back to an ordinal variable. For instance (an extreme case to make the point), if the average player ranks were:

Wake: 136.6
UVA: 136.7
Clemson: 160

Then, the rank would be:

Wake: 1
UVA: 2
Clemson: 3

Obviously you lose information by doing that. But like I said, this was purely supposed to be a very rough estimate.
 
Last edited:
Rafi, to answer your question, the correlation between average player ranking (after dropping the bottom 15%) for all the teams I collected data on and sum of AdjEM is -0.773.

Running a simple OLS regression, the adjusted R-squared is 0.5953.

But yeah, as I said yesterday, it was more challenging collecting recruiting data on some of the lower-level teams included in these numbers.
 
Dali Llama said:
Sorry if those plots are a little big. I can fix them later.

And I should've added Wake was 35th in recruiting and 93rd in on-court success (-58 differential).
Click to expand...

If you have the dataset handy and find the time, could you post the same representation with the Bz years excluded?
 
CharlotteDeac1 said:
If you have the dataset handy and find the time, could you post the same representation with the Bz years excluded?
Click to expand...

I'd be interested to see/compare the Dino years, Buzz years and Manning years.
 
Did you weight the fact of having multiple years of recruits on the same team?

Did you delete players who were injured and missed the season? Or were kicked off the team? Suspended?
 
Last edited:
These are just raw numbers for Manning years:

Recruiting ranking average: 61 (w/o 2018), 51.6 (w/ 2018)
Kenpom AdjEm average: 85.5

BUT, these numbers are incredibly misleading. The most important stat I see that I just gathered is that the recruiting average for Manning leading into the 2016-17 season was 68.66 (2014-2016). The Kenpom ranking for that team with the average 68.66 recruiting ranking was 36 to end the year.

So, last year Manning's coaching ability, and likely talent evaluation, led us to be just over 30 spots better than our recruiting average.

IMPORTANT NOTE: The 2014 class on 247 does not include Dinos Mitoglou so I got creative with this alternative scneario. I decided that Dinos would be ranked around where Olivier is on 247 (193, 85 rating - which I think is pretty generous out of high school for Dinos). This changed the overall recruiting ranking average for manning to be 59.33. Therefore, even in this scenario Manning coached our guys to be 23.33 spots better on KP than our recruiting rankings indicate we should be.

ETA: I did this really quickly so if someone wants to spot check to see if I made a mistake then please feel free. This would seem to indicate that Manning is indeed good at coaching basketball though so I'm sorry to some of the h8ers:cool:
 
Dali Llama said:
Oh, there is undoubtedly heteroskedasticity in the data. The recruiting rankings were definitely less reliable for lower quality recruits - the difference in a player ranked 350 vs. 400 is likely not the same as the difference in a player ranked 1 vs. 51. There was likely more information that went into the higher rankings being assigned than the lower rankings. There were also more NAs for lower quality teams. That is likely why you see more variability as recruiting rank increases. I can get a correlation for you if you'd like, though.

There is also the issue of taking the average of an ordinal variable and then transforming this average back to an ordinal variable. For instance (an extreme case to make the point), if the average player ranks were:

Wake: 136.6
UVA: 136.7
Clemson: 160

Then, the rank would be:

Wake: 1
UVA: 2
Clemson: 3

Obviously you lose information by doing that. But like I said, this was purely supposed to be a very rough estimate.
Click to expand...


Did you just assume my skedasticity? Not going to lie, I had to look it up. #publicschool
 
RJKarl said:
Did you weight the fact of having multiple years of recruits on the same team?

Did you delete players who were injured and missed the season? Or were kicked off the team? Suspended?
Click to expand...
I don't fully understand your first question.

As to your second question, no, I did not. Tracking players' activities (suspended, leave early, etc.) across 11 years for roughly 6000 players on roughly 150 teams and then assigning respective weights is just not something I'm going to do for a little project I did strictly out of curiosity. This isn't my full-time job or academic research. As I emphasized in the first post and have stated many times since, this was supposed to be a very rough estimate with many limitations. I don't really know how else to say that.


Charlotte, at some point I can post data that includes the recruiting class of 2005 through the class of 2009 vs. KenPom AdjEMs of the 06-07 seasons through the 09-10 season if that's what you're looking for. Unfortunately, I didn't label class years when I was compiling the data, but it shouldn't be that difficult to go back in and cut it off after the 09 class for each team.
 
Dali Llama said:
I don't fully understand your first question.

As to your second question, no, I did not. Tracking players' activities (suspended, leave early, etc.) across 11 years for roughly 6000 players on roughly 150 teams and then assigning respective weights is just not something I'm going to do for a little project I did strictly out of curiosity. This isn't my full-time job or academic research. As I emphasized in the first post and have stated many times since, this was supposed to be a very rough estimate with many limitations. I don't really know how else to say that.


Charlotte, at some point I can post data that includes the recruiting class of 2005 through the class of 2009 vs. KenPom AdjEMs of the 06-07 seasons through the 09-10 season if that's what you're looking for. Unfortunately, I didn't label class years when I was compiling the data, but it shouldn't be that difficult to go back in and cut it off after the 09 class for each team.
Click to expand...

If you have time that would be interesting. Of course, this is on your own time so no rush. I would be really interested to see all the data you have from 06-07 to - 16-17 excluding the Bz years ('10-'14).
 
OK, I could do that at some point. I would just have to make two cuts instead of one. Which classes do you suggest be excluded from the recruiting data? Just the 2010 class (that only played in the 2011 through 2014 seasons, which will be excluded from the on-court success measure)? Or the 2009/2011 classes as well (which would have played for only one year of the on-court performance captured)?

ETA: Obviously this doesn't apply to transfers/redshirts.
 
Last edited:
You must log in or register to reply here.
Back
Top