Talent Comparison Game by Game 2014

[ImportedVoL]

Please this kind of analysis is so silly. Do you think Vegas cares? Good grief, these are the worst.....

WTF.. what does it matter what vegas cares about?

good grief, these troll replies are the worst...

its ok for someone to say we are going to lose becuase of talent level, but not that we will win because of talent level.

Rapidly gaining my vote for worst poster on volnation, everything you say is either smart arse/rude or just plain stupid.

And from the replies you get to every post you make.. I am not the only one who thinks this.

 

[sjt18]

Beattie gonna be Beattie.

What will he do when we start winning?

If he's a troll fan of another team like I once suspected then he'll just disappear. If he's really a Vol fan... just deceived by his own vanity... then he'll probably either disappear or change screen names.

He doesn't come across as the type to say, "I'm sorry. You were right and I was wrong."

 

[Quaint88]

I would say every unbiased critic would agree.....you think Johnny football is worth .2 wins?

That is a long term average, not the ceiling that an individual QB can achieve in a season. The average is 0.2 games per year despite occasional extreme outliers such as Manziel, who probably accounted for 10-15 times that (or 2-3 wins) last year.

 

[Quaint88]

Absolutely and he does not deny it. In his analysis, Carolina was not an upset and we should have beaten missouri. In addition, playing an inexperienced qb matters very little. I'm not attacking him because he is far,far,far smarter than me. I'm just not sure his analysis has any practical application. Being a predictor of 70 percent of all games is not saying much at all really. We know everybody has at least three built in wins. We know that the vols have more talent than chattanooga and south alabama without looking.

The recruiting analysis is far from perfect, but it at least gives an easy formula to get an objective starting point for predictions that has a roughly 70% success rate. Then, you can use the eyeball test to try to eat into the other 30%, with the caveat that you could also be adding to it instead if you're wrong. I guess I just don't see why statistics and the eyeball test have to be at odds when they can compliment each other. Also, this type of analysis is geared more towards making smart bets in individual games, not telling how happy of a fan you will be at the end of an entire season. I agree that +/- 2 wins in a 12-14 game season is huge, but I see that as one of the less useful things you can do with this data because of the cumulative variance.

 

[led0321]

The only thing is that 15 of 27 players in the #13 rated class of 2011 have either left, or never see the field.

And 13 of 22 players in the #17 class of 2012 have either left, or never see the field.

So....no matter what your analysis says....the stats don't mean squat here bro.

UT has been bitten hard by the bug of attrition and Dooley. Those were good classes by the ratings, but they don't do us much good if they're not on the field.

UT wins...but not because of it's Rival rankings.

These stats are a dose of reality.

 

[daj2576]

The recruiting analysis is far from perfect, but it at least gives an easy formula to get an objective starting point for predictions that has a roughly 70% success rate. Then, you can use the eyeball test to try to eat into the other 30%, with the caveat that you could also be adding to it instead if you're wrong. I guess I just don't see why statistics and the eyeball test have to be at odds when they can compliment each other. Also, this type of analysis is geared more towards making smart bets in individual games, not telling how happy of a fan you will be at the end of an entire season. I agree that +/- 2 wins in a 12-14 game season is huge, but I see that as one of the less useful things you can do with this data because of the cumulative variance.

This is what I am having trouble communicating. It isn't a plus or minus two game prediction, the vector of the uncertainty doesn't matter. It is simply that 80% of the teams will finish within 2 games of their predictions. It only matters, at least in my view, if you are trying to pick how an individual team will do, then you do have a four game window. Using other factors, such as coach effect, can give you a stronger indication of which way the vector for an individual team's over or under performance will go.

Looking at it broadly, even if you insist on viewing it as a four game predictive window, when in reality it is only a 2 game window (doesn't matter which side of the prediction you are on), looking at the numbers like this helps pin point coaches who are under or over-performing.

Under performing: Muschamp, Dooley, Kiff, Fulmer (later years), Saban (a bit misleading due to his talent, as he can never over-perform, although I can't find that he did when he had rosters that weren't the best).

Over performing: Spurrier, Jones (except last year), Petrino (at Arkansas), Franklin, Pinkel (except for last year) and Mullins.

About right: Miles, Freeze, Malzahn, Richt (although CMR is trending a bit negative of late), Stoops, Sumlin.

 

[Quaint88]

This is what I am having trouble communicating. It isn't a plus or minus two game prediction, the vector of the uncertainty doesn't matter. It is simply that 80% of the teams will finish within 2 games of their predictions. It only matters, at least in my view, if you are trying to pick how an individual team will do, then you do have a four game window. Using other factors, such as coach effect, can give you a stronger indication of which way the vector for an individual team's over or under performance will go.

Looking at it broadly, even if you insist on viewing it as a four game predictive window, when in reality it is only a 2 game window (doesn't matter which side of the prediction you are on), looking at the numbers like this helps pin point coaches who are under or over-performing.

Under performing: Muschamp, Dooley, Kiff, Fulmer (later years), Saban (a bit misleading due to his talent, as he can never over-perform, although I can't find that he did when he had rosters that weren't the best).

Over performing: Spurrier, Jones (except last year), Petrino (at Arkansas), Franklin, Pinkel (except for last year) and Mullins.

About right: Miles, Freeze, Malzahn, Richt (although CMR is trending a bit negative of late), Stoops, Sumlin.

I get that. I was just trying not to muddy the waters any more than necessary when my point is that you don't have to use the data to predict an entire season at a time.

 

[daj2576]

I get that. I was just trying not to muddy the waters any more than necessary when my point is that you don't have to use the data to predict an entire season at a time.

Very true. The most helpful, at least to me, is the 70/30 distribution of wins due to talent. One game at a time.

 

[Quaint88]

Very true. The most helpful, at least to me, is the 70/30 distribution of wins due to talent. One game at a time.

Yes, but it would be nice if there were some simple way of including attrition in the short-term numbers. We were hit pretty hard in the classes that should be upperclassmen now. I'm sure we're not even with OK. I would guess we'd be down somewhere around Ole Miss (but certainly above Missouri), but I wish I had an objective way to see if that's the case.

 

[daj2576]

Yes, but it would be nice if there were some simple way of including attrition in the short-term numbers. We were hit pretty hard in the classes that should be upperclassmen now. I'm sure we're not even with OK. I would guess we'd be down somewhere around Ole Miss (but certainly above Missouri), but I wish I had an objective way to see if that's the case.

This has been a relatively long term complaint of these numbers. After the spring games last year I did a comparison of attrition across the SEC. I started with the recruiting averages, then went through each roster and assigned points for stars, dividing by the number of players. In other words, I accounted for attrition. When I compared the lists, that is the recruiting averages versus the recruiting averages corrected for attrition up to last year's spring game, the order of the list was largely unchanged. Importantly, UT did not move in relation to our opponents.

That told me a few things. First, we tend to pay very close attention to UT and our biased by that. That closeness magnifies the feeling of loss when player(s) leave. If we paid as close attention to all other teams, we would find that, in general, all teams suffer from a typical range of attrition. Second, in a roster of 85, even if a team has 2 or 3 more players leave than other schools, it doesn't really move the needle when examining latent talent. Finally, if one can get to 70% prediction without accounting for attrition, it stands to reason that the majority of teams suffer from similar attrition rates, or that the differences of attrition rates are accounted for in the remaining 30%.

What I tend to find is that when you look at the remaining 30% of unpredictibility, you tend to isolate coaches who consistently over or under perform, not a random distribution of teams who might be explained by attrition.

 

[Quaint88]

This has been a relatively long term complaint of these numbers. After the spring games last year I did a comparison of attrition across the SEC. I started with the recruiting averages, then went through each roster and assigned points for stars, dividing by the number of players. In other words, I accounted for attrition. When I compared the lists, that is the recruiting averages versus the recruiting averages corrected for attrition up to last year's spring game, the order of the list was largely unchanged. Importantly, UT did not move in relation to our opponents.

That told me a few things. First, we tend to pay very close attention to UT and our biased by that. That closeness magnifies the feeling of loss when player(s) leave. If we paid as close attention to all other teams, we would find that, in general, all teams suffer from a typical range of attrition. Second, in a roster of 85, even if a team has 2 or 3 more players leave than other schools, it doesn't really move the needle when examining latent talent. Finally, if one can get to 70% prediction without accounting for attrition, it stands to reason that the majority of teams suffer from similar attrition rates, or that the differences of attrition rates are accounted for in the remaining 30%.

What I tend to find is that when you look at the remaining 30% of unpredictibility, you tend to isolate coaches who consistently over or under perform, not a random distribution of teams who might be explained by attrition.

Thanks for that excellent information. Last year's numbers would've included the 2010 (Kiffin/Dooley) class that had a huge amount of attrition of highly rated players. If the attrition in that class couldn't impact our relative position even in the short-term, we really are seeing our attrition through a haze of Orange Kool-Aid.

I still wonder about giving the older classes more weight than the new ones, though. Obviously, the #5 class would have boosted our average just as much no matter where it fell, but it would help us more on the field if it had been a year earlier.

 

[BeattieFeathers]

How accurate do you think Vegas is, if you're not impressed with 70% using only recruiting averages?

My cursory numbers indicate that when picking straight winners and losers, ignoring the spread, Vegas is right about 80% of the time. Have you ever considered how complicated the algorithms are that Vegas uses to get that extra 10%?

Do you know that to make money betting in Vegas, you only have to be right (beat the spread) about 54% of the time? These numbers dont beat the spread enough to be lucrative, but they are strong predictors.


From a standpoint of pure predictive ability, any number that gives you a better than a 50/50 shot at being right is a good start. 70/30 is considerably better than that.

I bet if someone told you that you had a 70% chance to be injured tomorrow, you'd think that was a strong enough indicator to change your behavior, but here your resisting the same forewarning with a passion.

Vegas doesn't pick winners per se, but if you are saying favorites win 80 percent of the time straight up, I hear you. However, keep in mind that a fairly large percentage of games aren't handicapped. I'm taking about games against FCS schools. Since most sec schools play 1 fcs school per season, that can be 14 games that aren't predicted. Since there are only 112 total games scheduled that involve SEC members, that's over 10 percent of the games that they don't bother with that are no brainers that enter into your 70 percent figure. Much easier to predict winners against fcs schools than conference games using your method.

 

[BeattieFeathers]

This has been a relatively long term complaint of these numbers. After the spring games last year I did a comparison of attrition across the SEC. I started with the recruiting averages, then went through each roster and assigned points for stars, dividing by the number of players. In other words, I accounted for attrition. When I compared the lists, that is the recruiting averages versus the recruiting averages corrected for attrition up to last year's spring game, the order of the list was largely unchanged. Importantly, UT did not move in relation to our opponents.

That told me a few things. First, we tend to pay very close attention to UT and our biased by that. That closeness magnifies the feeling of loss when player(s) leave. If we paid as close attention to all other teams, we would find that, in general, all teams suffer from a typical range of attrition. Second, in a roster of 85, even if a team has 2 or 3 more players leave than other schools, it doesn't really move the needle when examining latent talent. Finally, if one can get to 70% prediction without accounting for attrition, it stands to reason that the majority of teams suffer from similar attrition rates, or that the differences of attrition rates are accounted for in the remaining 30%.

What I tend to find is that when you look at the remaining 30% of unpredictibility, you tend to isolate coaches who consistently over or under perform, not a random distribution of teams who might be explained by attrition.

Even great programs like Bama suffer an attrition rate of 45 percent. That is why I can't get excited about anyone until I've actually seen them play because I know that about 50 percent will be busts,flunk out, not adjust, etc. I do like the fact that we have more bullets than most since we signed 32 players and they signed 25 or so. I would like to have that ratio every year.

 

[BeattieFeathers]

Define "better players". If you are talking about things that Rivals can accurately assess in a HS player then you are wrong. If you are talking about the skills they developed and the coaching they received... then you are probably right.

The USCe game in spite of your refusal to accept it was no fluke. UT matched up well on both LOS's. USCe likes to run downhill whereas the teams that hurt UT most ran outside the tackles. USCe's WR's were good but not great players who played in Spurrier's system. They made themselves easier for UT's DB's to cover.

On D, their strength was in their DL. They made up for some weaknesses in the back 7. UT's O strength was STILL their OL. Except for a better day by Clowney, UT was able to contain their DL.

To some of us, it wasn't a surprise at all.

Failing to recognize that USCe is NOT a top team in terms of talent is to deny what Spurrier has done with the talent he's gotten. The guy can flat coach.

I think the nfl draft tells us that Carolina puts its fair share of players in the pros. It's not like he is coaching a bunch of pop warner players. To say that we had better material than Carolina is illogical.

 

[Volru]

This has been a relatively long term complaint of these numbers. After the spring games last year I did a comparison of attrition across the SEC. I started with the recruiting averages, then went through each roster and assigned points for stars, dividing by the number of players. In other words, I accounted for attrition. When I compared the lists, that is the recruiting averages versus the recruiting averages corrected for attrition up to last year's spring game, the order of the list was largely unchanged. Importantly, UT did not move in relation to our opponents.

That told me a few things. First, we tend to pay very close attention to UT and our biased by that. That closeness magnifies the feeling of loss when player(s) leave. If we paid as close attention to all other teams, we would find that, in general, all teams suffer from a typical range of attrition. Second, in a roster of 85, even if a team has 2 or 3 more players leave than other schools, it doesn't really move the needle when examining latent talent. Finally, if one can get to 70% prediction without accounting for attrition, it stands to reason that the majority of teams suffer from similar attrition rates, or that the differences of attrition rates are accounted for in the remaining 30%.

What I tend to find is that when you look at the remaining 30% of unpredictibility, you tend to isolate coaches who consistently over or under perform, not a random distribution of teams who might be explained by attrition.

Great analysis. Last I saw we were about in the middle of the conference based on class rankings. On the other hand, I think high school recruits in the mountain west region may tend be a little underrated (or overlooked). The Southeast and West Coast do get the hype. Utah State obviously has some talent.

 

[daj2576]

Vegas doesn't pick winners per se, but if you are saying favorites win 80 percent of the time straight up, I hear you. However, keep in mind that a fairly large percentage of games aren't handicapped. I'm taking about games against FCS schools. Since most sec schools play 1 fcs school per season, that can be 14 games that aren't predicted. Since there are only 112 total games scheduled that involve SEC members, that's over 10 percent of the games that they don't bother with that are no brainers that enter into your 70 percent figure. Much easier to predict winners against fcs schools than conference games using your method.

You make many incorrect assumptions that I don't have time to correct. Let's just leave this where it is. I'm fairly certain more data won't change your perception.

 

[sjt18]

I think the nfl draft tells us that Carolina puts its fair share of players in the pros. It's not like he is coaching a bunch of pop warner players. To say that we had better material than Carolina is illogical.

Nope. I didn't actually say that... but it still isn't "illogical". You can explain last fall by pointing to transition from one scheme to a very complex one dependent on understanding and precise execution on O. You could even say that Jones was playing for the future while letting the Srs and Tiny have their season. You could say that one off season isn't long enough for the players to be re-developed into the type Jones wants... Or you could just say Jones can't coach and UT is in big trouble. But the talent on UT's roster was underutilized last fall.

And... Spurrier is STILL an elite coach that does more with less.

 

[SPRINGHILLVOL]

Utah State game: Slow down the Utah State qb...get the ball to wide receivers.. limited turnovers.. Result: Tennessee 35 Utah State 24.....P.S. Hurd and Lane should have great days..

 

[BeattieFeathers]

Nope. I didn't actually say that... but it still isn't "illogical". You can explain last fall by pointing to transition from one scheme to a very complex one dependent on understanding and precise execution on O. You could even say that Jones was playing for the future while letting the Srs and Tiny have their season. You could say that one off season isn't long enough for the players to be re-developed into the type Jones wants... Or you could just say Jones can't coach and UT is in big trouble. But the talent on UT's roster was underutilized last fall.

And... Spurrier is STILL an elite coach that does more with less.

We had no receivers....until North emerged. We had very little qb experience and once he started to emerge he got injured. What was jones supposed to do with a freshman qb who was rushed into service, who by all indications was going to be redshirted? Worley/north beat vandy, IMO. They play aubie and missouri better, but we lose both. Defensively, we were slow as molasses. Look at our anemic sack stats over the past few years. Pitiful....