Thanks for putting this together. Since you already did all this work, I hope you don't have to go into hibernation again to come up with an update next week after National Signing Day.
The Race for the Number One Recruiting Class
I’ve been in hibernation for a couple weeks working on getting all of my recruiting data in order and wanted to open a first post on the 2013 class by looking at how the overall picture of the top classes looks.
A quick refresher on the methods I use to rate recruits. Each recruit is given an overall ranking at each of the four major recruiting sites. For recruits in the top 250-300 the site gives that rating themselves. For recruits outside of the top threshold, I use an implied value based on position rating and player grade (available for everyone but Scout) to produce a final player ranking for all players. This ranking is then applied to a log scale so that the very top players are given an extra “bonus.” A unanimous #1 like Robert Nkemdiche is this year will finish with a rating of 99 points. Michigan’s top recruit Derrick Green, is the 21st highest rated player overall and is rated at 80 points. David Dawson is #101 at 58 points and a player on the fringes of the national rating like #305 Maurice Hurst are worth about 40 points.
Michigan’s top rivals are all having outstanding recruiting seasons as well. To gauge the classes, I have plotted each of the teams' commitments alongside each other, ranked from highest to lowest.
Where’s the threat?
Notre Dame’s class features the best #1 with the class (Jaylon Smith), #2 (Max Redfield) and #3 (Greg Bryant) before falling back in line with Michigan’s class.
Despite a marquee name at the top, Ohio State features the strongest overall top ten before the depth falls below Notre Dame and Michigan.
The strength of this Michigan class is in the quality depth. All 26 of the Wolverine position player commits rank in the top 750 nationally.
Meanwhile in East Lansing there is a clear talent gap as the peak is significantly lower and the decline is even faster. Any thoughts that the Spartans had of closing the recruiting gap seem laughable at this point.
The National Elite
In addition to Notre Dame and Ohio State, four teams from the SEC along with USC are making runs at the nation’s top 2013 recruiting class.
Michigan, Florida and LSU all have nearly identical classes with only slight deviations in player rating at each level.
Alabama is very strong through the top 10 but features a serious decline from there on.
Texas A&M is this year’s packed house with over 30 commits. There is a definite separation through the bulk of their class and the rest of the national elite, but without a sharp dropoff at the tail, the class is more than just a collection of also-rans.
USC’s class is small due to the NCAA sanctions but is absolutely loaded. The Trojans feature only 14 commitments but every single one of them are in top 200 players nationally.
Picking A Winner
Splitting hairs over which class is slightly better at this point in time is a mostly absurd. As you can see, the margins between the top classes are very slim and although I am a firm believer in recruiting ratings at a high level, there are a lot of classes within the margin for error for top class.
With that in mind, it’s seven months until the next meaningful college football game and so let’s assess the contenders using various methodologies.
Add ‘Em Up
Probably the simplest method, take each recruit for each team and add up their points and see who has the most. Using this method we currently have a top ten that looks like this:
Certainly not a bad way to look at it but the huge class at Texas A&M certainly seems to be overrated in this method. Add to that the opportunity cost by loading up a single class in terms of ability to offer the future, and this look is insightful but a bit incomplete.
Average ‘Em Up
An average versus a sum takes away the issue of opportunity cost lost by over-offering the current season and looks at where each team ranks on players committed, taking away the class size bonus.
USC small class size becomes irrelevant in this rating as the difference between their class and #2 Alabama is the same as #2 and #9. Michigan still finishes a solid 7th.
This method also has its drawbacks as now the opportunity cost is reversed. It values teams holding their offers for future classes, potentially costing the team opportunities in the present if they can’t keep a full scholarshipped roster.
A Player Rank Approach
One way I have been looking at classes this year is similar to the graphs above. Comparing each team’s Nth recruit versus all of the other classes to see how they stack up. With a limit of 25 scholarships per class, I gave the best 1st in class player 25 points, the second best player who was the best in his team’s class 24 etc. For each subsequent team spot, I dropped the points and players evaluated by 1 so for each team’s 25th best player, only the top one received a single point.
|Player||Pts||Mich Rank||Nth best rank||Team Pts|
Based on this method each of Michigan’s top 15 commitments garnered at least 10. Jake Butt at #13 and Ben Gedeon at #15 where each the top players are their respective position within the class (No other team had a 13th or 15th best player rated as high as these two). Shane Morris, despite his senior year slide, earned Michigan’s highest point total with 18 points as the 5th best #4 prospect in any current class.
The hybrid approach puts Michigan at #5, behind top rated Notre Dame but just ahead of Ohio State.
In the end it really is splitting hairs with high degree of variability. Michigan’s class is probably not the #1 class but it is certainly a top 5 class with lots of quality depth. With back to back elite classes under its belt, Michigan should return to national elite roster levels within the next 2-3 seasons, a position it hasn’t been in since 2007.
Great work, Mathlete. I like the reasoning behind each approach you provide, helps out a lot. I'm curious if I'm missing something about Texas A&M...33 commits? Wow, didn't take them long to assimilate to the SEC. Are they really losing that much to graduation and/or declaring early? The current roster shows less than 20 seniors.
And finally: Re: Sparty...."the peak is significantly lower and the decline is even faster"
From your fingers to God's ears, my friend.
Really love the recruit ratings vs Nth best recruit plots...was wondering how someone could come up with a quick visual to show something like this, and this perfectly shows it off. Nice work!
I don't like that there are lines connecting the discrete points, I'd rather see a scatter plot which each player be a seperate dot. /nitpick
Always good to get a few different cuts of the data with different analytical approaches. Insightful as always.
I really like those graphs you used to compare class depth and quality. Very cool.
Jaylon Smith is going to be nasty. At least we'll only see true freshman/sophomore versions.
So your saying theres a chance?
This was amazing, nice work & thanks for putting it together. I loved the Nth-rank way of looking at things.
The nth best approach reiterates the concept of this class being a deep collection of very good players. Every player from spots 3-26 is top 5, so lots of depth.
Clearly reinforces the quality of OL with 4 of top 8 and the concerns with our WR talent given all 3 in the bottom 7 of position players.
Kudos to Hoke for a top 5 class. Well done.
I love that Mike Valenti just said the other day that Ohio State's recruiting is "on a different level" than Michigan's.
What else did you expect?
he only does this to get a rise out of you? That part of their schtick is so obviously contrived that I find the rest of the show isn't worth a listen. It insults my intelligence.
On a related note, it's important and meaningful that Michigan fans have regular opportunities to lol at Sparty. With that in mind Mathlete, is there any chance you could do that last bit with MSU's class?
Believe it or not thats their way of getting people to call in and michigan fans care about their team so they always call in when they talk shit about michigan... When they talk about msu no fans call in because they just done care lol
Oh shit, I love this. I've tried some similar rankings in the past, so I know how difficult and time-consuming it is, and I absolutely love some of the decisions you made here (logging, comparing Nth best recruits, etc.). These are better than anything I did and better than anything I've seen others do.
It's interesting to me that according to Scout, Michigan's class is far and away #1, while the others have M between 5 and 7. Scout heavily weights recruits according to their position ranking, and Michigan has a ton of top 5 players at their position. I'm not sure it's a good way to do it, but Michigan looks nice at #1.
I wonder if finding the area under the curves on those graphs wouldn't be a good way to rank teams. It presumably would be a lot less work to calculate it on an ongoing basis. You could automate it and release it every week. I'd do it, but my calculus is beyond rusty.
Negligible differences between our class and Ohio's aside, this is another fantastic effort by Hoke & Co. I cannot wait until our overall roster values reach elite levels.
...that is both informative and impressive. You my friend are ranked #1 in recruiting class analysis! Thanks for info.
Hoke has really built this team from the inside out which is by far the best way to do it. If you go back thru the years the O and D line are your money makers.When other teams D players are on their backs and the QB is under the gun, thats where you win.Would not trade this class for anyone elses.
Just another idea for how to rank these teams. In Cross Country (CC) you get points for your place, and the LOW score wins. Your top 5 runners place counts toward your score and your next 2 runners can push runners from other teams back. For example in a dual meet, a shut out occurs when you place your top 7 runners before the other team's #1 runner. Your placements are 1,2,3,4,5,6,7 and your score is 1+2+3+4+5 = 15. The other team's placements are 8,9,10,11,12,13,14 and their score is 8+9+10+11+12 = 50. If you have more than 7 athletes on your team, they don't count in the scoring.
An advantage of this system is that it would take advantage of the depth of a class, as well as the top. Highly rated players get few points, but a lower ranked player can really scuttle the ranking. (Imagine if your CC team went 1,2,3,4,150. That would be a score of 160. Despite having the 4 best runners there, you --almost certainly-- wouldn't win. I've seen a score as high as ~130 win the state meet, but that was a very close meet...#2 had only a point or 2 more.)
Adapting this to recruiting rankings is (relatively) simple. We'd first need to determine what teams you care about and rank ONLY THOSE players. Now there are two parameters that you'd have to set. First, how many players contribute to a team's score. (I'd probably start with 20.) Next you'd have to determine how many players you want to cap the total class size at. (25 seems reasonable to me). Once you set this second threshhold, you can eliminate any players above that threshhold from your "scoresheet".
Lastly, you'll have to complete the teams who have less than 20 recruits. Suppose there are 240 players on your scoresheet. Because USC only has 14 players, they need 6 more to get to 20. So you'd have to add in player # 241, 242, 243,...246 to their tally. You'd also repeat this for any other team with less than 20 recruits (starting at #241 again). From there it would be as simple as adding up places.
Are you using R to analyze this data? The graphs look like R outputs. Just curious because I'm taking an R Data Analysis class on Coursera right now.
Looks like Excel to me.
But R IS awesome. Especially when coupled with Sweave or knitr.
However, I think the real measure of a class is how it stacks against the needs of the team. For example, if Bama already has 6 five star running back recruits, and they pass on one this year for a four star lineman where they need depth, is that not a better recruiting class for Bama.
The question is how to measure this: perhaps by extending the third method back three or four classes, and then getting a relative need ranking by position. This of course is further compicated by adding in base defensive and offensive scheme and associated desired depth levels by position.
So, for now I'll be happy just knowing we have a top 10 class, and we're really strong in the areas we have seen we need depth in (OL, RB, and QB).
Are people in Ohio starting to believe this?
Thanks for the hard work, Mathlete. I wonder - how hard would it be to input past class rankings into these formulas? I'd like to see how classes from four years ago and beyond would score in these formats. We could score schools on running four year segments and see of one of these would be a better predictor of a football programs success. I know there are other variables (coach's x's & o's ability, injuries, transfers, etc.), but I'm curious.
Thanks for doing this, it is great! Please update after NSD, if you find the time.
I'm sure the Tigers or Wings would pay you good $$$ for a stat guy....