Friday, February 23, 2007

2007 Season Predictions

In my last couple of posts on this subject, we have been able to show a strong and positive correlation between success at recruiting as measured by Scout.com, and eventual success on the football field for the 2006 year.

I have gone back now and compiled total scores using the data at Scout.com for the four year recruiting period of 2004-2007. As my co-blogger Henry points out, Scout.com used a new rating system for 2007 that resulted in substantially higher scores. However, the new scores were applied evenly across all schools, thus we can still combine all the scores without compromising the effectiveness.

In making the predictions, we are using essentially a bell curve type analysis. We know that in 2006, teams averaged about 7.39 wins a year. Thus, the majority of teams can be expected to win 7 games, with the number of wins narrowing towards the higher and lower ends of the spectrum. Since you cannot win “partial” games (.39), we will round off our predictions.

To make our predictions, we determined what percentage of the overall recruiting scores for 2006 equaled X wins, and apply that same percentage to the scores for the 2004-2007 period.

Since we also determined there was a difference between the scores to success for major and mid-major conference teams, we split them apart. First, the major conference teams, broken down in thirds –

Major Conference – Top Third



That’s right ladies and gents – the Florida Gators have the best total 4 year score per Scout.com in recruiting. Three of the four classes are now Meyer’s, so this speaks very well to the job he has done in this area.

Now, it admittedly makes me cringe that the numbers crunch out to the Gators going undefeated in this exercise. I, for one, do not think it will happen, especially with our losses on the defensive side of the ball, and with our tough SEC schedule. Likewise, the Gators play numbers 4, 5, 6, 10, 12 and 17 on this list. And, many of those teams also play each other. So I think it impossible that Florida can win them all, while LSU wins 12, UT wins 12, and UGA wins 11.

I will address the SEC in a moment, but the point of this exercise is not to make “perfect” predictions next year. It is to show a number of wins that teams should be able to come within a game or two of. Applying the predictor to last year’s teams there was one instance where it was off by as many as 8 games (Rutgers), however the overall the average miss was less than 1 game.

Again looking at the top third, I think it is very reasonable to say USC will come very close to 13 wins. Texas at 12 seems reasonable, as does Michigan at 11, and Oklahoma at 10. Ohio State ought to win around 10, and Notre Dame is a lock at about 9. Follow this list, and the others, and see if any of the predictions look outrageous.

A final note on the top third – see North Carolina at 20th? I would think that if UNC wins 8 games next year it will be a big story. If so, you heard it here first.

Now, the 2nd group.


Major Conference – Middle Third


And, the final group –

Major Conference – Bottom Third

Once again, any given team here is open to debate. But if our model holds true, on average teams will be within a single game of the prediction.

Now turning our attention to the mid-majors. Among the mid-major group, Boise State was one of the top recruiting scores for the 4 years ended 2006. Thus, perhaps it was not so much of a surprise they won so many games, given as they played a schedule of teams with mostly worse talent (Oklahoma obviously excepted).

The top 5 recruiting scores for mid-majors are –



Once again, you can say you heard it here first (providing it proves true), one of these teams will be the “surprise” mid-major that is 2007’s BCS buster.

Finally, to the unique issue of the SEC. With so many teams highly ranked, and with the situation that some must inevitably lose, the following is a modified predictor for the SEC only –



And there you have it - no need to play the games next year. It’s Florida against LSU in Atlanta. But the Gator’s one loss will keep them out of the title game, regardless of the results of the SEC Championship, as USC faces Texas or Michigan for all the mythical marbles.



19 comments:

SMQ said...

I don't think it's possible for that many SEC teams to be that good in terms of wins - you have four SEC teams winning 11 games or more, and three winning 12 or more. I know there are 13 games these days, but that would be a stunning development. Although I guess three SEC teams did win 11 last year, and a fourth (Arkansas) came close. But it won't field four of the top six.

I have a bigger problem with Louisville (5 wins), West Virginia (5) and Rutgers (4) and with Duke (4) winning more than Wake Forest (3), anyway, though I don't think big improvement from North Carolina is crazy at all. Wake probably will revert to the mean, but that's not sub-Duke.

Henry Gomez said...

SMQ,

I agree with you about the number of wins possible for SEC teams.

On your second point I think this is where I believe coaching (the non-recruiting aspects of it) comes into play. If there were a direct relationship between recruiting and winning then Zook would have had a top 5 team probably in his last season at Florida. But good coaches can get good performance from mediocre players and poor coaches can get mediocre performance from good players.

Mergz analysis doesn't take that into account but if a team radically underachieves it's predicted outcome you may start looking at coaching decisions as a reason.

Same with teams that overachieve their predictions.

Henry Gomez said...

And again I would say that the new scoring system by scout.com (damn them) skews things toward teams that cleaned up this last recruiting season. And the thing is that those recruits will have the least amount of input in most teams good fortune or misfortune in the 2007 season.

Henry Gomez said...

I wonder if we contacted scout.com it they would take this year's scoring system and apply it to the 3 or 4 previous years to do an apples to apples comparison. It would make sense for them to have the information.

Mergz said...

SMQ-

I couldn't agree more with your comment on the SEC, and you see I said the same in my post, as well as making a modified SEC schedule only prediction below.

That said, if you look at the outcome from last year (as you point out), 3 SEC teams won 11, 1 won 10, and UGA won 9. So if they get those totals again, they will be within the margin of error I have assigned to the predictions (about 1 game).

I think Wake Forest is going to absolutely tank next year. Look at their roster last year - almost all seniors. I see them winning 5 games max.

Duke winning 4? Unlikely. But once again, it is a target.

As for the other teams you mention (Louisville, Rutgers, WVU), perhaps the Big East belongs grouped with mid-majors. They play a pretty easy schedule, so they get a lot of wins. It was this group that was most "off" in the 2006 work. I will look into it.

miguel said...

I put this in your post below, but why not rejigger your formulas to weight by seniority?

Determine 2004 points at 100% (seniors), 2005 (juniors) at 80% and so on. Tweak until appropriate.

Also, how the heck did you get comments as a small pop-up window? Would appreciate an email to ideateller of the hotmail kind.

Thanks for the analysis.

Nate said...

Great stuff. Fortunately, football has enough variables in terms of coaching, health, and emotion that this sort of wonkishness won't take the excitement and romance out of the game. Reading all the Bill James' books on sabermetrics ruined baseball for me.

I know somebody said this already in response to your analysis yesterday, and it probably presents an insuperable task, but there needs to be some way to account for class attitrion. Zook's 2004 class and Meyer's 2005 class have both been decimated by early graduation and players with grade/disciplinary problems. The entire back seven of Florida's defense next year will have only two upperclassmen on scholarship: seniors Tony Joiner and Kyle Jackson at safety. Every other LB, S, or CB will be either a sophomore of freshman. (I'm assuming that Jon Demps is gone, but even he, the most experienced of the entire LB corps, would be only a redshirt sophomore.)

It will take an extraordinary coaching job by Meyer and the defensive staff next year for Florida to win 12+ games, let alone win the SEC East.

Mergz said...

Nate-

Agreed about the attrition. However, virtually every high ranked school suffers attrition on some level (can you imagine USC last year if Reggie Bush had stayed?), so it probably is not much different across the high ranked schools.

On the other side, a school like Wake gets a year where they suddenly have a senior laden roster, and they overachieve. Schools like that tend to fall back into the pack pretty quickly (like Northwestern the year it won the Big 10).

nate said...

You are right, but I think the attrition in UF's 2004 and 2005 classes has been unusually high. Their junior and senior classes are very shallow, even in this age of early exits to the NFL.

It will be interesting to see. I see Tennessee, Georgia, and South Carolina moving up and Florida falling back to 5-3 next season. If Florida actually manages to go 7-1 in the SEC and play for the SEC Championship next December, not only will I be elated, I will come back here and acknowledge that you alone saw it coming. It will prove a lot not only about the accuracy of the scout.com ratings but about the value of raw talent compared to experience.

SDGator said...

I wonder if you can get stats on the actual makeup of each team, at least for the top 10 (or the SEC), and then weight them individually based on the number of freshmen, sophs, juniors and seniors? I'm sure it would be tedious to put in the spreadsheet, but then the weightings would be pretty danged close.

Anonymous said...

Outrageously your 40th something ranked Wisconsin Badgers may play USC for all the marbles as they have a solid returning class and the top freshman running back. Texas won't see a another mythical championship game anytime soon, they barely beat Iowa. Michigan and Florida doesn't have the defense to get very far in their conferences. Tennessee and Georgia will be better but not top ten material yet.

Henry Gomez said...

Anonymous,

Feel free to come back in about 11 months to tell us how your predictions did.

BTW, how did you come up with them? Do you use the Kirk Herbstreit method of picking teams?

machete said...

Gators will stay very near the top next year, but they won't win it all (not sure who yet). The 'Canes will be back crackin' the Top 10 again. The 'Noles will be lucky if they're ranked.

ps - wisconsin? really?? I'd give them some love as a Top 25 team, but BCS championship game? 'tas loco, acere...

algebraic!

Mergz said...

Personally, I would be shocked if Wisconsin had only the 5 wins the model predicts. The Big 10 is probably too easy a conference for it to happen.

Likewise, I would be totally astounded it Wisconsin was anywhere near a BCS game.

The problem with ranking these teams as I did it that it hurts teams conferences like the Big East who play mostly easy opponents. Louisville almost cannot do as bad as predicted because they play to many creampuffs.

In the case of Wisconisn, they are ranked behind Michigan, Ohio State, Penn State, Iowa, and Illinois in the Big 10. Do I think they lose to everyone? Probably not. Could I see them losing 3 or 4 of those. Yeah, easily.

I am going to revise this list next week on a by conference basis (like I did with the SEC at the end here). That could be a more accurate way of predicting.

Plus remember - these are targets, albeit targets that were pretty accurate last year (within one game on average). You will always have a team like a Wake ranked low that will do pretty well. However, what you didn't have was many of the highest ranked teams doing poorly.

Anonymous said...

mergz ...don't count the Badgers out on hoops either, bunch of freshman Buckeyes comin' on but they'll get knocked off early. Buckeye football is increasingly irritating cause they think beating the Wolverines ends a season, but as we speak the conference is in talks to extend the Big Ten season into December so that could put an end to that crap. henry's right on about Meyer, Florida won't retain him past a couple more seasons, he wants to coach at Ohio State cause of the tradition thing, the stadium, the band, all that crap the Gators don't have. The question is when does Tressel leave for the NFL..not sure about that.
Anyhow enjoy your posts but remember those star ratings are bias, no way those Tennessee, Georgia classes rank better than say a Wisconsin or Ohio States.

Henry Gomez said...

LMAO

There was "no way" that Florida was going to beat OSU either.

I never said Meyer wouldn't stay at Florida. I actually said that it's our rivals' wishful thinking that he'll leave

jimcaserta said...

I don't like a recruiting only model, but I think it will be UF v LSU in Atlanta. You know the holes in your model - trouble accounting for mid-majors and teams like WVU or Wisc.

I wonder how different this, or how much more accurate it is than just a rehash of prior polls? You could try 2005 final season to 2006.

Gator Duck said...

I've been looking at the recruiting issue from a different perspective. I normalized the total points assigned at all 119 teams by scout.com and then looked at the SEC from that perspective.

Go to http://gatorduckspeaks.blogspot.com/2007/02/different-look-at-recruiting-picture.html

Anonymous said...

The Noles will be lucky to be ranked??
You are going to learn alot about football this year! Many would disagree with you. This spring has already seen off the bat a 65 yard TD run by Smith and other big gainers by the the backs plus a pinpoint 31 yard TD strike to Carr by Lee. Tricketts blocking schemes are opening holes.

A scuffle has already broken out over aggressive blocking by the recievers on the cornerbacks who the coaches say look intimidated by the new blocking schemes.

All is well in Tallyho. Don't worry to much about us. Soon the the Seminole Vengance Tour will be underway.

http://seminoles.cstv.com/sports/m-footbl/spec-rel/032407aab.html