Comparing AP Top 25 and 247 Team Talent Composite Rankings
We're five weeks into the season. Most teams have played at least one 'real' game by now (or at least moved on to conference foes) which means that the Top-25 rankings are starting to have some meaning. So I thought it might be fun to check in to see which teams are under/over-performing in the rankings based on their roster talent.
The first table shows the AP Top-25 compared to the Team Talent Composite rankings from 247 sports. The 'Difference' column is simply their talent ranking minus their AP ranking, so positive/negative values indicate which teams are outperforming/underperforming based on their talent rankings and by how much.
As the table shows, Michigan is over performing based on its talent rankings by about 10 spots on the AP poll. The biggest overperformer is BYU who, while ranked #108 in the 247 talent composite, is ranked #16 in the AP poll. The biggest underperformer would be Crawfish Kelly's LSU Tigahs, with a -17 point difference in the team talent composite and AP polls.
Even more fun is the next table, which shows the top 25 teams in the 247 sports talent composite compared to their AP ranking. The 'Difference' column is again calculated by subtracting the AP ranking from the talent composite ranking.
In case you're wondering whether I went through this whole exercise to make fun of TAMU and Texas, you'd be partly correct. I mean, holy shit how do you drive the 4th and 6th most talented rosters into the ground like that? Of course, the other reason I'm doing this is to point and laugh at Florida and Miami.
Overall, 13 of the top 25 teams in the talent composite are currently unranked in the AP poll.
Out of curiosity, I also looked at the Talent Composite Rank and the Talent Composite Score of Michigan's past and upcoming opponents. The 'Rank Diff.' column shows how many points UM is ranked above/below their opponent, while the 'Points Diff.' column gives the number of points that UM is above/below their opponent in the talent composite. (For reference, UM is #14 in the Talent Composite with 847.24 points).
Of course, OSU is the only team in the conference with a more talented roster than Michigan. But it is interesting that, while 'only' ranked 11 points ahead of UM, the difference in composite points between OSU and UM is actually pretty close to the difference between UM and Sparty, who is ranked 20 spots lower than UM. We are actually closer to Nebraska in talent composite points than we are to OSU. The accumulation of talent at the top is real, and it is unreal.
Any thoughts? Other comparisons you might like to see in the future? Am I going to have to take back everything I said about Tennessee before the season began?
October 5th, 2022 at 4:48 PM ^
This is awesome
October 5th, 2022 at 5:42 PM ^
Happy to see that you are happy. Weekend is almost here.
October 5th, 2022 at 7:19 PM ^
These kinds of analyses are bordering on a ridiculous abuse of statistics and are just reasons for losers like me to not do my work. After all, they are based on talent assessments when kids were 17 years old, by a bunch of man-children who get paid to sit and watch high school games and express unchallenged thoughts. To do this reasonably you'd need to do it in retrospect and take into account players who make it to the NFL, earn college honors, and even take into account college statistics, because "talent" isn't reducible to recruiting rankings, as Derrick Green can tell you. Ronnie Bell is a great example of the depths of this mystery and the many facets- identification, development, luck w/r to injuries, good coaching- that lead to a conclusion of talent. There's even luck w/r to who's ahead of you on the depth chart and what your coach thinks of you. Remember Barry Sanders sitting beneath Thurman Thomas until his Jr. year, then immediately making obvious that he was the greatest running back in history. Look at Tom Brady. This kind of reductionism of comparing "talent" to ranking (also subjective) is such an oversimplification that I wish I could just take a shit on this thread in the nicest possible way.
October 5th, 2022 at 7:21 PM ^
How do you explain that there's a .78 correlation between FPI rankings and these talent rankings then? (I was a loser who calculated this after work)
October 5th, 2022 at 8:19 PM ^
I was told there’d be no math …
October 5th, 2022 at 9:12 PM ^
Of course there's a correlation or no one would do it, and recruiting would be revamped to a system that worked and had predictive value. My point is more that, saying UM "overachieved" by being ranked 4 when our talent is 14 is pretty much within in the margin of error of all this hand waving. A team at 14 being ranked by alleged humans as 4 at a random point in time isn't really very interesting from a statistical point of view.
October 5th, 2022 at 9:48 PM ^
I really don't get what your problem is. It's completely obvious to anyone looking at it that this is simply a shapshot of one point in time. I wasn't trying to generalize to a larger population of cases or predict who would be successful at the end of the year. Hell, my point wasn't even to say that Michigan is overachieving (though I did say that Michigan is outperforming its "talent rankings"). It was to have some, you know, fun at the expense of TAMU, Texas, Miami, etc. Like it or not, stars are fairly strong predictors of individual success, but these teams haven't been able to get out of their own way to find team success for years.
October 6th, 2022 at 2:30 PM ^
If you insist that I have "a problem", it would be drawing a qualitative conclusion that we have "overacheived" our level of talent because we are ranked in polls 10 spots above our alleged talent level. But if you were to ask me, I wouldn't call it a problem, I'd call it expressing a thought process on a Football forum.
October 6th, 2022 at 3:09 PM ^
This is some pretty interesting language for someone just expressing their thoughts.
These kinds of analyses are bordering on a ridiculous abuse of statistics...
...is such an oversimplification that I wish I could just take a shit on this thread in the nicest possible way
...isn't really very interesting from a statistical point of view
To me it sounds like you're taking the whole thing a bit too seriously.
October 6th, 2022 at 4:39 PM ^
You would be right, if one could reliably infer affect from online posting. Which one cannot.
October 5th, 2022 at 7:39 PM ^
Sure, there are always exceptions. Recruiting is imperfect and there are some misses. But, though imperfect, stars are highly correlated with college success. Just look at the draft rate across the different star-rating categories:
From On3 analyses of NFL draft rates:
- 5* (59% get drafted)
- 4* (22%)
- 3* (8.2%)
- 2* (6.2%)
It seems to me that they do a pretty good job of predicting future success.
October 5th, 2022 at 7:39 PM ^
No worries, people do re-rankings as well!
However we wouldn't be able to make those posts for 4+ years later (so, hold your horses until then?), so for now we'll have to settle for these.
And you're right about false reductivism; combining both a quantitative AND qualitative mindset always shows results in optimal performance. And, actually, the the highest experts in their respective fields tend to do just that (it's been studied by a crazy russian guy who did som research here at UM!), even more interesting they somehow show *less* brain activity than novices? Paradox? From the scans, it seems that they don't even *consider* the "bad moves" that a novice would spend brain-cycles (a metaphor) calculating, i.e. they only evaluate the best moves--experience beats theory every time (until it doesn't, and then you are revolutionary--> exceedingly rare, literally more lottery winners than field-revolutionaries: better odds)
October 5th, 2022 at 9:13 PM ^
Sounds like you're talking about chess, a sport that now dopes with vibrating anal beads.
October 5th, 2022 at 4:51 PM ^
Does this factor in transfers as well?
October 5th, 2022 at 4:56 PM ^
Yes. If you go to the 247 Team Talent Composite page (and click on the hyperlink for # of commits), it has the full roster and the *s for each player that it is based on. I noticed that Michigan's recently changed within the past 2-3 weeks which I assume was because of Okie.
October 5th, 2022 at 4:52 PM ^
Really well-done. Beyond my math skillz, but would be interested in the correlation between recruiting and ranking. More power if done for a few years and based on end of year, but this snapshot suggests that "r" is not as strong as I'd have thought, at least not once you get past the top few.
October 5th, 2022 at 6:54 PM ^
Great idea. I think I'll do that but, like you said, at the end of the season with the final rankings. But it would have to be with something like SP+ that gives a ranking for all teams so I can calculate the correlation for teams that aren't in the top 25 of the final rankings.
October 5th, 2022 at 7:23 PM ^
I just did it using ESPN's College FPI to include all 131 FBS programs. The correlation between FPI ranking and 247's Talent ranking is .78
October 5th, 2022 at 8:18 PM ^
That’s a pretty strong “r” from where I sit. Suppose that’s why the math guys actually do the calculations instead of eye-balling it!
October 5th, 2022 at 9:29 PM ^
It's a pretty impressively high r. Kind of makes me wonder whether 247 goes back in and re-rates guys after they matriculate. No idea if this happens.
October 5th, 2022 at 4:57 PM ^
I somewhat feel for Texas. They would be much better with their first-string QB that totally conned OSU.
October 5th, 2022 at 5:13 PM ^
Much better is a big statement. Card has played well. And all we saw of Ewers was (less than) one half. He looked good but Spencer Rattler looked amazing right out of the gate before struggling (and later transferring). I LOVE watching Texas disappoint but that is not solely due to Ewers injury.
October 5th, 2022 at 5:55 PM ^
I don’t think it’s (mainly) about the QB either. Maybe you need a star QB to win a NC but not to be a top 25 team. Michigan in 2015-2019 and 2021 is exhibit A. Texas seemingly does not have a great HC — no team he’s coached has ever lost fewer than 4 games. But I question whether UT is just bad at hiring coaches or institutional forces are preventing coaches from succeeding (and/or good coaches from taking a job there).
October 5th, 2022 at 5:09 PM ^
Really interesting. I don’t think Michigan really has only the 14th most talented roster. Since the rankings are based on stars, I think it means Michigan is better at assessing talent than the people at 247.
October 6th, 2022 at 9:52 AM ^
I also think Michigan has done better at finding the "right" players to fit into the program and scheme - with the "second part" being player development. I am hoping that Michigan has crossed the key hurdles it faced and has the foundation for a long and consistent future.
October 5th, 2022 at 5:22 PM ^
Mormonism = Football overachievement.
Fun Fact: I remember reading (years ago) that the CIA has a disproportionately large number of people (in both Intelligence and Operations) from within the Mormon community. The reasons? They don't drink or do drugs (i.e., no loose lips); they don't cheat on their wives (or at least are less inclined to do so, making them tougher targets for "honey pots" and double-agent status); and they have an incredible faith-based work ethic. I guess that translates to football!
October 5th, 2022 at 5:44 PM ^
I have several co-worker/friends that are Mormon ex-CIA folks. You are correct on all counts.
October 5th, 2022 at 5:52 PM ^
Also translating to football: older athletes. People lose some fitness and skills when on mission, but 23 and 24 year olds will generally be better players than "more talented" 20 year olds.
October 5th, 2022 at 6:02 PM ^
You just described Catholic Priests as well (no wives) for the most part. Kind of faith-based highly ethical people you would want to trust your kids with.
October 5th, 2022 at 8:00 PM ^
Not really. He described Mormons as having wives and a high fidelity to their marriage vows due to their faith.
It’s reprehensible that a small percentage of Catholic priests have abused children but as a Catholic myself, I take offense at the tongue-in-cheek implication that as a group, all Priests are not to be trusted with children. I personally know many, many good and holy priests who are great men and would never even think of harming a kid.
October 6th, 2022 at 1:01 PM ^
No tongue in cheek. It happened. It can happen again.
October 6th, 2022 at 11:38 AM ^
Either that or because they want to infilatrate the deep state and take over the country! Isn't that what all minority religious groups do? :D
October 5th, 2022 at 5:32 PM ^
Very interesting, and thanks for sharing. Nebraska has the 2nd highest talent points of our opponents but is talent-ranked behind 8 of them... what am I missing?
October 5th, 2022 at 5:35 PM ^
Shit. Nebraska is #24. I don't know how I got that wrong in the last table.
October 5th, 2022 at 5:39 PM ^
It doesn’t make sense that msu has 694 pts but is ranked 34 while Nebraska has 754 and is ranked 59 in the talent composite. Maybe that’s a typo and Nebraska should be 29?
October 5th, 2022 at 5:43 PM ^
Yep. It was a typo. Should be #24 (like the second table). BuddhaBlue caught it just a few minutes ago.
October 5th, 2022 at 6:02 PM ^
Very interesting nonetheless, thx for putting together. I was taking my sweet time typing up my answer and only saw the previous response afterwards!
October 5th, 2022 at 6:25 PM ^
Thanks. I hope I didn't sound snarky for saying another poster caught the mistake. I appreciate people catching my errors, but always want to kick myself for making them in the first place. 😊
October 5th, 2022 at 5:42 PM ^
Thanks for putting this together!
I think that in the table of Michigan opponents, Nebraska's talent rank should be 24 not 59 (and the rank difference 10).
October 5th, 2022 at 5:54 PM ^
I have a theory about this year being slightly more even than most years bc of the 6th year senior eligibility that would probably help the non-OSU/Bama/UGA teams more (because those teams always have tons of talent crammed into the scholarship limit, whereas other teams can make up for the recruiting deficits by having older players, transfers, etc). No clue if it holds up or not, but in my head it made some sense.
October 5th, 2022 at 6:05 PM ^
I had a similar thought as well. I really thought that a team like Wisconsin would benefit more than others from having a 6th year for players. Another year learning the same system for their big OL, etc. Man was I wrong about that one.
October 5th, 2022 at 5:59 PM ^
Next years top 25 NR list will look close to the same. Recruiting dominance matters for exactly 4 teams. If you want to be #7-25 recruiting is negotiable.
UM is THE ONLY team trying to buck this trend. I say that because Okie State tends to fall of by seasons end.
October 5th, 2022 at 6:18 PM ^
I think you mean "negligible," but with NIL you're also not wrong.
October 5th, 2022 at 7:17 PM ^
Yeah it kind of works even if it wasnt attended.
;-)
October 5th, 2022 at 6:14 PM ^
What I hate is when pundits find excuses for teams like Miami and TAMU... Jimbo is in his 5th year... 5th year!!! He doesn't need any rebuilding years.
October 5th, 2022 at 6:30 PM ^
Roster Management
Coaching
Player Development
October 6th, 2022 at 9:55 AM ^
Kewaga - and if I may add a fourth dimension - it would be in-game execution.
Let's say you're running a passing play. The blocking is great. The receiver runs a great route. The QB delivers the pass accurately. The pass hits the receiver on the hands and in front of him (to allow YAC). And, the receiver drops the ball.
October 5th, 2022 at 6:31 PM ^
I think your last point is the most salient. The talent gap is biggest at the top. The difference between the third most talented team and the 10th most talented team is bigger than the difference from 20th to 50th. And you see that in who is ranked and not.
in general the elite talent teams are the elite teams with a real shot at a national championship whereas being a talented but not not elite team puts more onus on the coaching and the right roster construction to have a good year.
October 5th, 2022 at 6:57 PM ^
Some notes:
When Texas/Oklahoma join SEC, SEC has 6/9 most talented teams.
If USC/Oregon/Washington/Notre Dame/Stanford join B10, that would mean 21/25 most talented teams are in the B10 or SEC. The other 4 teams (Clemson, UNC, Miami, FSU) are all ACC.
Seems like Big 12 and Pac 12 are toast. Hard to see ACC surviving past their current contract. I wonder if in 10 years, when the ACC blows up finally, we end up with a pro baseball situation in which teams play exclusively within their league, and their respective champions (hopefully determined by playoff) play each other at the end.
October 5th, 2022 at 9:17 PM ^
With relegation systems!