Here is the answer you have been looking for (with proof-LOOONG).

#51
#51
daj2576 said:
Yes. Ole Miss recruits surprisingly well for as bad as they have performed recently (notwithstanding 2012).
Click to expand...

Yes I noticed that, same thing about UT too, ugh.

Now, looking at Vandy, and it's not because I hate Vandy or Franklin...If you just go by recruiting rankings it looks like to me, Wake Forest would have been predicted to beat Vandy in 2011. starting in '08...WF was #58, 64, 69, 69 and 69 in '12...Vandy has some huge swings from'08 #90 to '12 #29.

In '08 and '10 Wake Forest beat Vandy, with Vandy winning in 2011, the two teams did not play in '09.

This might be of interest, while I'm not much of a math guy, I am a stat guy. I appreciate your efforts here, it's really interesting once one starts digging into it. Thanks!

2011 Vanderbilt football schedule, picks and results
 
Last edited:
#52
#52
Vercingetorix said:
Very interesting. If you're looking for comment on your methodology, the thing that jumps out immediately is this: it looks like you're putting too much on the raw recruiting rankings, which imply a granularity of distinction that simply does not exist. I.e., an 18th ranked class is not really better in any meaningful sense than a 19th or 20th ranked class. This ends up meaning that, in your system, two coaches can recruit almost identical classes, but only one can get full credit for beating the other. (If the coached of the 18th ranked team beats the 19th, that's a game he's supposed to win, but if the 19th beats the 18th, that's an "overperformance." Even though the talent levels are essentially identical.)

I'd like to see this kind of analysis run with some sort of groupings to blur out that false granularity -- letter grades for recruiting classes rather than raw ordinal rankings, perhaps?

Interesting stuff.
Click to expand...

All excellent points actually.

At some point, unless one is getting paid to do these sorts of evaluations, you reach a point where the initial data is good enough to create the baseline strata.

Interestingly enough, one of my best friends who is a stock analyst gave the same critique. As we started digging into the numbers together (on a previous run of similar calculations) he ultimately concluded, as did I, that once you get to a point of about 80% accuracy you realize that you could spend so much time and effort trying to get out that last 20% (with no guarantee of different results) that you just leave it.

I readily admit that there are many weaknesses with this evaluation, like the fact that it is only 1 year. To do it effectively you should guage a coach over his full tenure at a single school. Those spreadsheets would get VERY long and complicated indeed.

I actually did that for the SEC a few months back. The results for each coach when reviewed over the timeline of their tenure, were very close to the 1 year measurements.
 
#53
#53
98stage2 said:
I read your post because I'm a numbers geek as well. Two questions:

1)Why did you only go back 4 years?
2) Were you able to workaround the fact your method punishes schools like Alabama and LSU because they should always win?

Good work on a difficult project. *tips hat*
Click to expand...

1) As I was using recruiting as a baseline (arbitrary? perhaps, but I had to start somewhere) I began with the presumption that any team is roughly an amalgamation of 4 years of players (fresh, soph, juniors, seniors).

2) When I reviewed the data I actually went back and gave a cursory review to those teams that had what I am loosely calling "the coefficient" of anywhere from -2 on up. I did that to at least give those coaches who are supreme recruiters the benefit of the doubt. This list would seem to punish guys like Miles and Saban, but it doesn't (and here is the real jump) you probably aren't going to be pulling away any coach from a team whose recruiting rankings average top 5 and is winning 10 plus games a year. So, at that point it becomes moot. What I was looking for were reasonably "gettable" coaches who have shown an ability to over perform with their talent. One could surmise that most of those coaches could probably recruit much better at a school like UT, but those factors deserve some deeper evaluation at the point of decision.
 
#54
#54
So the deck is stacked against Saban, who can never achieve a positive ranking without playing the Dolphins.
I would assume that recruiting has value and should be ranked, perhaps as an additional variable. That said, interesting analysis and a darn good start.

** partially answered with your latest post...
 
#55
#55
daj2576 said:
All excellent points actually.

At some point, unless one is getting paid to do these sorts of evaluations, you reach a point where the initial data is good enough to create the baseline strata.

Interestingly enough, one of my best friends who is a stock analyst gave the same critique. As we started digging into the numbers together (on a previous run of similar calculations) he ultimately concluded, as did I, that once you get to a point of about 80% accuracy you realize that you could spend so much time and effort trying to get out that last 20% (with no guarantee of different results) that you just leave it.

I readily admit that there are many weaknesses with this evaluation, like the fact that it is only 1 year. To do it effectively you should guage a coach over his full tenure at a single school. Those spreadsheets would get VERY long and complicated indeed.

I actually did that for the SEC a few months back. The results for each coach when reviewed over the timeline of their tenure, were very close to the 1 year measurements.
Click to expand...

Ideally it would be possible for, say, Nick Saban to get credit for beating Les Miles. They both have five stars backing up five stars all over the field. They both have A+ talent. But if Saban wins, he's supposed to, while if Miles wins, he's overperforming. I'd like to see a system that recognized games that are close to toss-ups based on talent, where either coach gets a win out of it.

That's if you wanted to flesh it out into a full fledged system for evaluating recruiting-independent coaching, if you will. And this is potentially over halfway there. Good stuff as is. Thanks for sharing it.
 
#56
#56
globadoc said:
So the deck is stacked against Saban, who can never achieve a positive ranking without playing the Dolphins.
I would assume that recruiting has value and should be ranked, perhaps as an additional variable. That said, interesting analysis and a darn good start.

** partially answered with your latest post...
Click to expand...

Well thank you for the compliment.

What I initially set out to determine a few months ago when I began this hobby was to determine if coaching or recruiting was a bigger factor. I did a pretty lengthy evaluation of the SEC (certainly a microcosm of the larger NCAA subgroup) and determined that over the past 8 seasons, within 1 game (+/-) average, the team with the better trailing 4 year recruiting average would win 80% of the time. That seemed to indicate that recruiting was, in fact, king (at least within that very limited subgroup). Using that as a baseline assumption, I wanted to see if any coaches tended to push that "normal" away from the trend line. Sure enough, guys like Petrino, Spurrier, Dooley (negatively) and others really jumped off of the page. Sure, Saban and Miles recruit fantastically and they have bumped up against the ceiling of "perfection". In essence, they have the perfect combination of coaching and recruiting (but who didn't know that?).

Without making this conversation much more complicated than it should be, Saban's time at MSU could be evaluated similarly and you would probably see that his performances are tied directly to his stocked talent. My conclusion is that at a school like UT, where recruiting is incredibly handicapped from the beginning, that a coach who can get more out of slightly less talent is what is needed presently.
 
#57
#57
Vercingetorix said:
Ideally it would be possible for, say, Nick Saban to get credit for beating Les Miles. They both have five stars backing up five stars all over the field. They both have A+ talent. But if Saban wins, he's supposed to, while if Miles wins, he's overperforming. I'd like to see a system that recognized games that are close to toss-ups based on talent, where either coach gets a win out of it.

That's if you wanted to flesh it out into a full fledged system for evaluating recruiting-independent coaching, if you will. And this is potentially over halfway there. Good stuff as is. Thanks for sharing it.
Click to expand...

I think we are in total agreement about the weaknesses of this evaluation. I certainly appreciate your thought and your compliments!

But to be clear, or maybe redundant, I wasn't necessarily looking for coaches like Saban and Miles who have great talent and stay within a few wins and losses of the trend. I was looking for people who spike the trend, meaning they can get more done with less talent. Perhaps I incorrectly framed my initial point, and that is, I wasn't looking to prove that guys like Miles and Saban were worth their money, I was looking to find a coach who has proven himself as a tactician (a coach). The recruiting will come in the SEC, with the wins.
 
Last edited:
#58
#58
daj2576 said:
I think we are in total agreement about the weaknesses of this evaluation. I certainly appreciate your thought and your compliments!

But to be clear, or maybe redundant, I wasn't necessarily looking for coaches like Saban and Miles who have great talent and stay within a few wins and losses of the trend. I was looking for people who spike the trend, meaning they can get more done with less talent. Perhaps I incorrectly framed my initial point, and that is, I wasn't looking to prove that guys like Miles and Saban were worth their money, I was looking to find a coach who has proven himself as a tactician (a coach). The recruiting will come in the SEC, with the wins.
Click to expand...

I am more ambitious for you. I want to see you build this into a system. Keep the algorithm private, publish your findings, eventually you make a lot of money either as a consultant or by Jimmy Sexton paying you to keep quiet.

Curious: what does your system say about Mark Richt?
 
#59
#59
Vercingetorix said:
I am more ambitious for you. I want to see you build this into a system. Keep the algorithm private, publish your findings, eventually you make a lot of money either as a consultant or by Jimmy Sexton paying you to keep quiet.

Curious: what does your system say about Mark Richt?
Click to expand...

Richt recruits in the top 3-4 of the SEC. The interesting thing about his schedules in the past few years is that they have been comparatively very easy, and he tends to lose games he should win. Overall he stays within a game or two of his trend line, but I haven't gone back and reexamined his data specifically.

I would like to have the time/effort and ability to create exactly what you are describing and make a ton of money off of it. Sadly, I work full-time and I am a full-time graduate student (not math) and I don't see that in the cards. Plus, there is nothing proprietary in these figures, just simple math. But the vote of confidence is nice to hear! thank you.
 
#60
#60
daj2576 said:
Well thank you for the compliment.

What I initially set out to determine a few months ago when I began this hobby was to determine if coaching or recruiting was a bigger factor. I did a pretty lengthy evaluation of the SEC (certainly a microcosm of the larger NCAA subgroup) and determined that over the past 8 seasons, within 1 game (+/-) average, the team with the better trailing 4 year recruiting average would win 80% of the time. That seemed to indicate that recruiting was, in fact, king (at least within that very limited subgroup). Using that as a baseline assumption, I wanted to see if any coaches tended to push that "normal" away from the trend line. Sure enough, guys like Petrino, Spurrier, Dooley (negatively) and others really jumped off of the page. Sure, Saban and Miles recruit fantastically and they have bumped up against the ceiling of "perfection". In essence, they have the perfect combination of coaching and recruiting (but who didn't know that?).

Without making this conversation much more complicated than it should be, Saban's time at MSU could be evaluated similarly and you would probably see that his performances are tied directly to his stocked talent. My conclusion is that at a school like UT, where recruiting is incredibly handicapped from the beginning, that a coach who can get more out of slightly less talent is what is needed presently.
Click to expand...

I agree that we have a natural recruiting disadvantage and that we need a coach that can do more with less, making your analysis reasonable at face value. And I respect your desire for parsimoniousness, but the complications make it more interesting! :)

Here are a few complications: Do teams get credit for beating Auburn's better recruiting class this year, even though their record is awful? What if the players comprising a better class leave (a la Penn State or Kiffin's class here)? Does the point spread or home/away matter? Shouldn't we underweight freshman classes? Won't more years be a fairer evaluation? Shouldn't we discount everything Sir Lies-A-Lot (Petrino) does? Should teams get more credit if the recruiting class is waaay lower (I.e. Appy State over Michigan)? Should a coach's record follow him, as would be the case with Sumlin?

Interesting stuff, indeed. Btw, don't answer these questions. I do respect your analysis, just pointing out other issues to consider in a more robust measure.
 
#61
#61
globadoc said:
I agree that we have a natural recruiting disadvantage and that we need a coach that can do more with less, making your analysis reasonable at face value. And I respect your desire for parsimoniousness, but the complications make it more interesting! :)

Here are a few complications: Do teams get credit for beating Auburn's better recruiting class this year, even though their record is awful? What if the players comprising a better class leave (a la Penn State or Kiffin's class here)? Does the point spread or home/away matter? Shouldn't we underweight freshman classes? Won't more years be a fairer evaluation? Shouldn't we discount everything Sir Lies-A-Lot (Petrino) does? Should teams get more credit if the recruiting class is waaay lower (I.e. Appy State over Michigan)? Should a coach's record follow him, as would be the case with Sumlin?

Interesting stuff, indeed. Btw, don't answer these questions. I do respect your analysis, just pointing out other issues to consider in a more robust measure.
Click to expand...

Some day, perhaps over a beer if that is your style, we should discuss those if we are ever in near proximity. Some of your questions I have actual answers for, some I had not considered in this evaluation and others I just made blanket assumptions. Good points, great conversation.
 
Last edited:
#62
#62
One other thing, I really do appreciate you bringing this to the board, the only other team that had a bigger swing than Vandy, from a 90+ ranking to borderline top 25, is TCU. In 2011, their Avg. rank was 53.5 and they were expected to win 10 games. Of all the teams predicted to win 10 games or more only Pittsburgh at 41.5 comes close. So, seems like Gary Patterson would be at the top of the list as he went 11-2 in 2011. In other words, he exceeded extremely high expectations with a mediocre at best recruiting rank.
 
#63
#63
Globadoc addressed several of my methodological considerations, most notably the discrepancy between recruiting rankings as of National Signing Day and longterm retention. I too was thinking specifically of Kiffin's class, which, on paper, was a top-10 class but consisted of so many high risk players that it was whittled by attrition to virtually nothing.

A special case in point pertains to Kansas State. I don't know if he is practicing it as extensively during the current tenure, but Snyder made a virtual science out of recruiting junior college players. He recruited them far more heavily than just about anyone I can remember, in some years totaling almost 50% of class signees. Since juco players do not typically figure into class rankings, KSU usually ranks quite low in recruiting. How do you massage the numbers in his case?
 
#64
#64
vol66 said:
One other thing, I really do appreciate you bringing this to the board, the only other team that had a bigger swing than Vandy, from a 90+ ranking to borderline top 25, is TCU. In 2011, their Avg. rank was 53.5 and they were expected to win 10 games. Of all the teams predicted to win 10 games or more only Pittsburgh at 41.5 comes close. So, seems like Gary Patterson would be at the top of the list as he went 11-2 in 2011. In other words, he exceeded extremely high expectations with a mediocre at best recruiting rank.
Click to expand...

If memory serves, TCU (which I did review) had an insanely weak schedule in 2011, even with their relatively lack recruiting. They were predicted to win 12 games and only won 10. I am recounting this from memory, so forgive me if I am off a bit. The averaged mid 50's in recruiting in 2011 and made a huge jump to low 40's in the one year average.

I wanted Patterson to have a stronger showing, which is why I included him.

EDIT: I will recheck TCU's numbers in the morning. I think I transposed a number on my spreadsheet for him.

EDIT2: I rechecked. They should have won 12 games last year and only won 10.
 
Last edited:
#65
#65
daj2576 said:
If memory serves, TCU (which I did review) had an insanely weak schedule in 2011, even with their relatively lack recruiting. They were predicted to win 12 games and only won 10. I am recounting this from memory, so forgive me if I am off a bit. The averaged mid 50's in recruiting in 2011 and made a huge jump to low 40's in the one year average.

I wanted Patterson to have a stronger showing, which is why I included him.
Click to expand...

I took my numbers off of your pdf.

predicted to win 10...53.5 avg....went 11-2

weak schedule but did beat Boise St.
 
#66
#66
#67
#67
Ahh, I had written a post wondering who did the Rivals talent rating methodology.

If you ever do any correlation co-efficient with these, the amount of correlation between recruiting and winning, at least in this league, jumps off of the page with how strong it is.

The one thing I would like to do in the future is to go even further with this and see who recruits the most guys who are actually able to stay in and complete a 3 or 4 year program.
 
#68
#68
daj2576 said:
which two teams are you showing on their schedule for 2011 that had better recruiting than them?

I also only include regular season games. Conference championships and bowl games are not included.
Click to expand...

I didn't mean to put Vandy's schedule up there, I edited it out of the post.
 
#69
#69
KingNick865 said:
Ahh, I had written a post wondering who did the Rivals talent rating methodology.

If you ever do any correlation co-efficient with these, the amount of correlation between recruiting and winning, at least in this league, jumps off of the page with how strong it is.

The one thing I would like to do in the future is to go even further with this and see who recruits the most guys who are actually able to stay in and complete a 3 or 4 year program.
Click to expand...

I did the same thing in the SEC and created a thread a few months back. Something like 80% of the time, the team with the higher four year trailing average will win the predicted amount of games or more during the regular season.
 
#71
#71
daj2576 said:
I did the same thing in the SEC and created a thread a few months back. Something like 80% of the time, the team with the higher four year trailing average will win the predicted amount of games or more during the regular season.
Click to expand...

Yep. For my research, I took the conference records of the 12 "original" SEC teams from the years 2002-2011 and compared them with the average star rating of those schools between 2002-2012, and got a correlation co-efficent of 0.9239893852124414.

I've attached the data I used if you, or anyone else for that matter wants to take a look at it.
 

Attachments

  • secrecruitingtrends.txt
    1.5 KB · Views: 2
  • Like
Reactions: 1 person
#73
#73
Tldr, but I think your list is terrible. If that's UT's actual list, particularly if it's in that order, God help us.
 
#74
#74
So based on this criteria, Dooley would actually get points over James Franklin for similar results, simply because Dooley gets out-recruited by Franklin. Nevermind that the head coach at Tennessee should never get out-recruited by anyone at Vanderbilt.
 
#75
#75
Man you sure do post some interesting stuff.

What do you do for a living?
 
New posts
You must log in or register to reply here.

VN Store



Back
Top