Recruiting Ranking and Results in 3-4 Years

#77
#77
I did another analysis using recruit rankings and experience. I think I can use this to create more accurate pre-season predictions.

sites.google.com/a/email.arizona.edu/d-shane-miller/blog/2-9-12-grading-papers-and-college-football-rosters

Here's what if would have predicted 2011 compared to what actually happened.

I read your blog post. A few things to think about.
1. You may try averaging rankings over 5 years since so many kids redshirt. Not sure if it will affect your results.
2. Honestly, I have no experience with ordinal data, but you may want to look at running some sort of maximum likelihood regression. It would be easier to control for additional variables with some kind of regression format.
3. Alternatively, it might be easier to predict number of wins. You could use a count model (Poisson or Negative Binomial Regression)
4. count off points for grammar/punctuation. That's what I do with my students.
 
#78
#78
I read your blog post. A few things to think about.
1. You may try averaging rankings over 5 years since so many kids redshirt. Not sure if it will affect your results.
2. Honestly, I have no experience with ordinal data, but you may want to look at running some sort of maximum likelihood regression. It would be easier to control for additional variables with some kind of regression format.
3. Alternatively, it might be easier to predict number of wins. You could use a count model (Poisson or Negative Binomial Regression)
4. count off points for grammar/punctuation. That's what I do with my students.

Now that I think about it more, you don't really have many observations to work with, so I guess you couldn't really do much with regression, but a small n (12) also means that your analysis is probably somewhat suspect.
 
#79
#79
A very simple chart. It just takes the final recruiting rank and shows the average of how that rank performs junior and senior years.

Recruiting Rank.jpg

Edit: The title on the y axis got cut off. That is the final regular season ranking.
 
#81
#81
I read your blog post. A few things to think about.
1. You may try averaging rankings over 5 years since so many kids redshirt. Not sure if it will affect your results.
2. Honestly, I have no experience with ordinal data, but you may want to look at running some sort of maximum likelihood regression. It would be easier to control for additional variables with some kind of regression format.
3. Alternatively, it might be easier to predict number of wins. You could use a count model (Poisson or Negative Binomial Regression)
4. count off points for grammar/punctuation. That's what I do with my students.

Just saw this...

1. Why? Not that many that redshirt will be around 5 years later, whereas most of the highly ranked athletes that drive the rankings will be gone.

2. Using a scalpel when a sledgehammer seems to be doing the trick.

3. Might do that, but I'm not sure "number of wins" is any more sound than the ratio. Better teams play more games (conf championships, bowls). That's just another factor distorting the data.

4. Seriously? What do you teach? Surely not something where you require your students to have critical thinking skills.

And your post below regarding the number of observation points - 12 data points, a significant p-value, and a strong pattern all indicate that the pattern is there. We're talking about college football, not chemistry or building bridges.
 
#82
#82
Just saw this...

1. Why? Not that many that redshirt will be around 5 years later, whereas most of the highly ranked athletes that drive the rankings will be gone.

2. Using a scalpel when a sledgehammer seems to be doing the trick.

3. Might do that, but I'm not sure "number of wins" is any more sound than the ratio. Better teams play more games (conf championships, bowls). That's just another factor distorting the data.

4. Seriously? What do you teach? Surely not something where you require your students to have critical thinking skills.

And your post below regarding the number of observation points - 12 data points, a significant p-value, and a strong pattern all indicate that the pattern is there. We're talking about college football, not chemistry or building bridges.

Perhaps your small sample size is not problematic for spearman's rho. With some methods, small sample sizes can increase the likelihood of Type I error when using significance tests (for instance, anything with asymptotic assumptions). That's why you could not use Poisson or Negative Binomial.

I was of course referring to regular season games, not championship games when I wrote about using count models.

As for counting off for grammar and punctuation, it increases the readability of papers, and it makes grading easier.
 
#83
#83
Perhaps your small sample size is not problematic for spearman's rho. With some methods, small sample sizes can increase the likelihood of Type I error when using significance tests (for instance, anything with asymptotic assumptions). That's why you could not use Poisson or Negative Binomial.

I was of course referring to regular season games, not championship games when I wrote about using count models.

As for counting off for grammar and punctuation, it increases the readability of papers, and it makes grading easier.

Ahhh...seems as though we just had a communication failure there.

And regarding punctuation, I count off for that, but then the problem comes with actually evaluating the arguments.
 
#84
#84
Looking at that graph, the one common thread among the teams which severely underperformed, TN and Auburn, is a serious loss of key talent.

I'm no statistician, but there has got to be a way to adjust for losing starters which would build a better model for what happened. TN didn't just lose Hunter last season, we lost Moore, Stocker, and Jackson, all integral parts of the 2010 team.

When you get a chance, plug in next years teams and lets see what comes out!
 
#85
#85
Looking at that graph, the one common thread among the teams which severely underperformed, TN and Auburn, is a serious loss of key talent.

I'm no statistician, but there has got to be a way to adjust for losing starters which would build a better model for what happened. TN didn't just lose Hunter last season, we lost Moore, Stocker, and Jackson, all integral parts of the 2010 team.

When you get a chance, plug in next years teams and lets see what comes out!

Good point. Might be able to do that. As soon as the teams release their media guides, should be able to plug in the data and use that to put them in some kind of order.
 
#86
#86
A very simple chart. It just takes the final recruiting rank and shows the average of how that rank performs junior and senior years.

View attachment 44818

Edit: The title on the y axis got cut off. That is the final regular season ranking.

Interesting chart. Care to walk us through what you see?
 
#87
#87
Interesting chart. Care to walk us through what you see?

That your top 6 recruiting ranked teams perform better jr year and not sr year. I think this is because these top class have so many go pro early.

After that teams tend to perform better as sr's
 

VN Store



Back
Top