Football Interludes, Part Two*: B1G Performance and Recruiting Classes Considered in Overlapping Four-Year Periods
February 15th, 2016 by 89 Chemistry

* See the end of the Post for a B1G Basketball addendum.

While not a prerequisite, Part One should interest those Spartan Football junkies who missed it earlier.

Five years ago, I began tracking and sharing four-year win totals of Big Ten teams on LSJ’s Hey Joe! blog.   Since I did not publish a graph updated for the 2014 season, one is now overdue.

Four years turns out to be an especially appropriate duration in assessing program strength.   In thinking about what makes a program a program last summer, I looked at coaching tenures.   To my surprise, the average number of seasons on the job at the beginning of the 2015 season for the FBS’s 128 head coaches was only four.   The average for B1G teams was also only four.

Four years is also the standard duration we associate with student-athlete university careers.   (It is not entirely accurate—red-shirted players do usually contribute to team success (both inside and beyond practices).   But generally, players have four consecutive years to do their part in intercollegiate contests.)   This both leads us to “graduating” (out-going) classes and returns us to recruiting (incoming) classes.   With the former coinciding with a valid program-strength sample size (of games), and the latter not yet mold-encrusted in the minds of college football junkies, I think it interesting to compare the two somehow.

And what better how than the obvious? Here is how things were basically supposed to be,

B1G recruiting 2015

and here, how they actually were:

B1G results 2015


That phrase “Big Two, little twelve” certainly applies to recent recruiting classes!   (Generally, that is; there are several exceptions in individual years.)   At least Urban Meyer and his players are doing their part in coming close to expectations; the best thing I can say for the Gulos is their underperformance was even worse five years ago.   Michigan—for now—looks to be recruiting like an 11-2 program and playing like an 8-5 program.

The average quality of recognized (or projected) talent for B1G teams has improved slightly over the last six years to 2.99 Stars/signee/class—with MSU, Indiana, Northwestern, and Purdue exhibiting significant gains.   Of these four programs, MSU is the only one to show improved results along with its upgrades in anticipated talent.   In contrast, Purdue is doing so much less with more that one might think their locker room is plastered with “Be Like Michigan” signs.   Apart from saying that MSU’s Stars/signee/class average should improve to ~3.3 for 2013-2016—while Michigan’s should stabilize—I refuse to speculate on what these curves might look like in the future.

Does the recruitment of broadly recognized talent matter?   I think it generally does, but I am not sure how best to test causation.   Correlation, however, is strong.

It was not always so: Five years ago, there was reportedly a 30% correlation between recruiting-class strength and team performance.   But after the 2014 season, analyst Bill Connelly found a 67% correlation between five-year recruiting-class strength and team performance (defined by an advanced rating called F+).   The data in my B1G graphs show a 74% correlation between four-year Stars/signee/class averages and totals wins, and a 79% correlation between the Star averages and how B1G teams do relative to one another (i.e., first, second, …, 14th best or worst).   (For matching four-year periods, in both cases.)   I interpret these high correlations to mean that the recruiting services are getting better at rating something.

Winning begets “stronger” recruiting classes; that should surprise no one.   But I found that correlations between 2012-15 wins and four-year Stars/signee/class averages remain about the same for these teams until one goes back to 2008-11 and before—where they drop off smoothly with time.   This indicates, to me, that projected-talent levels are indeed somewhat predictive of wins a few years down the road.   As I have said elsewhere, it is not merely the guys on the field that are driving wins; scout-team quality and the guys that are vying for the starters’ jobs also matter.

(On the other hand, correlations between 2012-15 Stars/signee/class averages and wins in previous tetrennials spiked for 2010-13 wins before falling off more ambiguously.   It is tempting to venture, from the spike, that how prospective college teams do in kids’ sophomore years most influences the kids’ eventual destinations relative to how they do in earlier or later years.)

It has often been said that MSU “does more with less”.   Does MSU do it better than any other Power-Five program, as a Sporting News analysis concluded (q.v. Part One)?   The data in the graphs have encouraged me to build a crude model to test the idea:

Construction:   Assume that an average recruiting class pool of 3.00 Stars/signee should win 29 games in four years (that being the nearest integer to the 2012-15 B1G mean).   Having found (above) that four-year Stars/signee/class averages vary more-or-less linearly, assume a deviation of 0.04 Stars/signee/class will change the outcome of one game every four years; e.g., teams with recruit averages of 3.04 and 2.96 Stars “should win” 30 and 28 games in four years, respectively.   Compare the resulting expected wins to wins indicated in graphed data.



adj. wins 43 28 38 31 30 25 50 34 29 19 21 17 12 30
expected 34 19 31 24 26 22 48 32 32 24 26 22 20 43
margin +9 +9 +7 +7 +4 +3 +2 +2 -3 -5 -5 -5 -8 -13

Green: “Coaches Up” Admirably

Teal: “Coaches Up”, Modestly

Black: Normal/Special [PSU]

Orange: Under-Develops

Red: Under-Develops, Quite Badly

Brown: “How the Bobby L. Williams is This Even Possible?!?!”


One cannot truly say that Michigan State football is “more overachieving” than Minnesota football right now.   But to the extent that Rivals and Scout are accurate in their overall talent assessments, the Spartans and the Gophers are achieving more with their talent than the other B1G programs—and MSU will sustain that level of relative development prowess longer than MN.   (If any program can catch MSU in this respect any time soon, it will be Iowa.   But again, this refers to relative development; Iowa is not about to become OSU West.)

It is ironic to see the expected wins for MI equal MSU’s actual wins—and it remains odd to see MI with as many wins as NU in four years.   (It is not like NU has facilities akin to the Gulos’, is it?)   But it is also revealing to see that MSU has had incoming talent ranked better, on average, than that of Nebraska’s and Penn State’s scholarship players.   MSU is an 11-2 program—but it must surprise some people that it has recruited like a 9-4 program.   For longtime fans, the surprise is probably along the lines of, “That well?!”

For those outside the region, the surprise is probably, “That’s all?!”


Returning to the second graph, one can still imagine the B1G as three-tiered—like it was two years ago.   Since then, however, the top tier has flared out with OSU and NE moving in different directions—and Minnesota has migrated from the slums and joined the mainstream.

The variability hidden in each data point differs, of course.   On the football field, Penn State has been the FBS’s most constant performer over the last six years; never more than one win astray from eight in any year.   MSU looks more constant in the graph, but that is because the 2012 & 2013 seasons (7-6 and 13-1) cancel each other out.   Those variations give MSU the third-highest standard deviation in per-year wins (σ = 2.6) among B1G teams.   (Iowa—having won 4, 8, 7, and 12 games in its last four seasons—currently has the largest per-year-win σ in the B1G.   Compared to quite a few other FBS teams, however, the Hawkeyes are relatively consistent.)

(Note the implication of MSU’s per-year σ:   Were I to draw the appropriate error bar on MSU’s 2012-15 datum, it would stretch from an absurd 53 to a nearly-as-absurd 33 wins per tetrennial.   The latter would overlap with Rutgers’ error bar—implying a 2% chance that Rutgers will win more games than MSU in 2016.)

MSU is not scheduled to play Purdue; Nebraska; Iowa; and Minnesota this fall.   As long as MSU takes Illinois more seriously than it seems to have taken Purdue (and I think it will), MSU’s schedule looks favorable—especially with OSU, WI, MI, and two other B1G foes at home.

Sidebar: On The Other Hand—   If one represents the seasons in which MSU won at least nine games with a ‘1’ and other seasons with a ‘0’, here is what the Dantonio-era sequence looks like:

0, 1, 0, 1, 1, 0, 1, 1, 1, …

Hard to avoid spotting a pattern, is it not?   Well, I suppose it is even harder to see how MSU could lose five or more games this fall.   But State is replacing more starters than it has needed to since at least 2012—and having fewer offensive starters is a more sensitive risk factor for year-to-year victory-margin depression than is fewer defensive starters.   (And MSU has that problem, too.)   For a team that won several close games last year, that is not good news.   Of course, there is no intrinsic reason for the pattern to continue—any more than past mutual-fund returns predict future returns.   Still, I thought the pattern so far worth noting.

Next year, I expect the four-year win totals for NU, RU and (pathetically) PU to be lower than they are now.   (This is saying, precisely, that those teams should have worse records this fall than they had during the 2012 season.)   I expect the totals to improve for IL and IA—with the latter jumping near NE’s current total.   Ordinarily I would expect MI to hold steady, but Jim Harbaugh has a lot of returning veterans for 2016 and a very solid Year Two rebuilding record; it should improve, but no more than IA should.   And, of course, MSU’s totals should move markedly closer to OSU’s.

In Part Three, this exercise will be applied to the top “quintile” of 2012-2015 programs.   (Top 21%, to be precise; there is a seven-way tie for 21st place.)   I close for now with a few anticipated questions about today’s Post:

Why do you average Rivals and Scout data, but ignore data?   Rivals and Scout were noted blogger HeckDorland’s early sources; like him, I think two sources are better than one.   (Sometimes there are significantly different Star averages between Scout and Rivals.)   I agree that three are better than two; I should gather and average in 247’s ratings, too.   But it is my suspicion that the improved accuracy would not appreciably lower ranking sigma (uncertainty) enough to justify the many hours of additional work I would have to put in for all the teams and years I expect to cover in Part Three.

Why use Stars/signee instead of class rankings?   The Rankings seemed biased in favor of large classes—which I associate with oversigning and rapid student-athlete turnover.   If there is going to be bias, I would rather it favor programs that red-shirt a lot of their signees—which is, for the most part, what got MSU to where it is today.

How could OSU pull ahead of MSU in 2015 when they had the same number of wins?   These are graphs of overlapping tetrennials: Not connected dots from, e.g., discrete seasons 2014 and 2015, but more like blurred snapshots.   OSU’s totals improved from 2011-14 to 2012-15 because OSU went 6-7 in 2011 and 12-1 in 2015; i.e., its four-year total increased from 44 to 50.   MSU has barely moved since 2010-2013 because its 2010 and 2011 wins [22] were replaced by about the same number of wins [23] in 2014 and 2015 in the current tetrennial.   For 2013-16, OSU’s totals should not change much (it should win somewhere between 10 and 14 games), but MSU’s should increase because it should win at least nine games and it won only seven in 2012.

Isn’t four years too short a period to define a “program”?   Did you forget my opening paragraphs already!?   But seriously, the answer is maybe.   The average number of seasons on the job at the beginning of the 2015 season for the 65 Power Conference teams was five years, so five years may be as good a term as four years.   But student-athlete turnover and B1G-coaching turnover is still closer to four years than five, so I am comfortable with my original notion of four years as a full, yet responsive, reflection of Coach So-And-So’s guys and the wisdom and commitment of their Universities.   Besides, it turns out that the order of B1G teams (i.e., wins, most to least) is the same for 2011-2015 as it is for 2012-2015.

There are other persuasive arguments for taking a longer view, and I will pay some respect to those in Part Four.

Whaddayamean “Adjusted” Wins?   Ain’t Real Wins better?   Not when comparing teams that play schedules of substantially different quality.   The B1G Results graph is a not-very-imperfect model that accounts for this by discounting some of the wins Maryland and Rutgers accumulated by playing ACC, Big East, and/or American Conference schedules in some years.   In effect, I subtracted one win from MD’s totals for each of the tetrennials except for 2012-15; I left that one unaltered.   RU is more complicated; one win was subtracted from the first four tetrennials, and two were subtracted from the three most-recent tetrennials.   More must be said about this in Part Three; for B1G teams, the biggest adjustment [RU from 27 to 25 wins] was hardly worth mentioning.

Why is Penn State “Special” in that table?   Because it was the only team that was not allowed 85 scholarships for a time (2012-2014).   One could say, in this context, that PSU is an orange in the apple cart that is B1G Football.   The scholarship disadvantage it had earned surely affected PSU’s ability to win games in 2012-2015—and did so independently of its Stars/signee/class average.   We will not be able to quantify how PSU under-develops or coaches up its players for a few more years with my crude model, but it is fair to say that PSU did not do as badly as that -3 margin indicated for 2012-15.


Here is the table that I made and was trying to share last night regarding the B1G MBB Pennant Race:


35 Responses  
  • jerseyjohn writes:
    February 15th, 20164:32 pmat

    i’m speechless. there’s really nothing for me to say except “impressive” the one thing that really grinds me is that we skipped penn state in the years you alluded to. we could have really gotten a leg up in the series history.

  • USMC Retired writes:
    February 15th, 20164:32 pmat

    Good stuff there, ’89.

    @OSU could be a chore and we NEED to play really well to beat Whiskey

    When on looks at the ‘Big Picture’ Coach D just seem to be doing as well as Jerry Kill was, BUT Kill never won anything. I think what’s missing in the data is that the Sparans won or tied for the B1G Title half of those years and when we didn’t OSU won it Whiskey notwithstanding

    • 89 Chemistry writes:
      February 15th, 20167:25 pmat

      Thanks, USMC. I never get tired of looking at the paradoxical realities of MSU & Michigan over the past several years.

      The data accounts for MSU’s Title wins by counting B1GFCG victories–which only two teams per year get the mere opportunity to compete for. The implication with Kill/Claeys is that, had MN been able to attract talent similar to State’s, then it would have been Wisconsin 2014, Iowa 2015, and a serious threat to beat whoever wins the B1G East this year. If Dantonio had been recruiting talent with less obvious potential than Purdue’s all these years, then MSU would not have won any outright titles, either.

      These are, of course, only ifs. Winners often attract more good athletes, and that was inevitable with Dantonio after 2010. It could be that the Gophers are where MSU was about six years ago and is on the cusp of winning the B1G West.

      To dismiss programs that get 7-6 results out of 4-8 talent is a legitimate point of view–and I am certainly disinclined to look at how 100+ teams perform on the field vs. on signing day. But I did want to look at the B1G, and was stunned to see MN recruiting at the bottom of the heap despite its frequent contention in its (admittedly not-so-tough) Divisions.

      As for this fall, OSU is a home game. The thing about the WI game is that it will be right after the trip to Notre Dame.

      • USMC Retired writes:
        February 15th, 20167:37 pmat

        My first ‘paragraph was abut the basketball schedule, not football

  • 89 Chemistry writes:
    February 16th, 20164:30 pmat

    In today’s press conference, Tom Izzo indicated that Kenny Goins would be out for two to four weeks. (The latter case would keep him out of the B1G Tournament.) Izzo said that compensating for Goins’s absence will require Schilling and Clark to play more minutes–and that Davis “needs to raise his game a little”.

  • jerseyjohn writes:
    February 16th, 20164:35 pmat

    Goins season has effectively ended. poor schilling breathes on a guy and fouls out. I find it interesting that javon bess is no longer in the equation.

    • 89 Chemistry writes:
      February 16th, 20164:56 pmat

      Izzo said something about not wanting to put Bess in for just a couple of minutes vs. IU. He wants his time in the game to be meaningful, apparently. [His performances in his four games since the Nebraska loss seem to back up the idea that he needs more time than Izzo is able/willing to give him to be productive.]

  • Ben Green writes:
    February 16th, 20165:14 pmat

    Intriguing stuff, 89 Chem. Thanks for taking the time. I particularly enjoyed the fact of, and the colors you chose for, the teams at either end of your WINS vs. WINS EXPECTED BASED ON RECRUITING.

    • 89 Chemistry writes:
      February 16th, 20168:33 pmat

      I was going to go with blue or red until I had a moment of Lubeckian inspiration.

  • USMC Retired writes:
    February 16th, 20167:06 pmat

    Acorns take down Paris, we win on Wed. and everyone(not named Alda, Md, or IU) in the B1G will be at 9-5

    • 89 Chemistry writes:
      February 16th, 20168:56 pmat

      PU still holds a tiebreaker over MSU. If WI beats IA next week, then it would presumably hold a tiebreaker over MSU. MSU will now finish ahead of MI & OSU as long as it wins out.

      To gain on specific top-three teams, MSU would realistically have to root for the Gulos; although the MD @ IN finale should knock the loser down to MSU’s level, the winner would probably be out of reach.

      As long as it keeps winning, State cannot be mathematically eliminated from the regular-season title until at least next week.

  • USMC Retired writes:
    February 16th, 20167:32 pmat

    Neither “Death penalty U” or the “we’ll take anybody U” (Baylor or ISU) can play a lick of Defense Which seems to be the way they roll in the B1G12 in both Football and Basketball.

  • LeeBee writes:
    February 16th, 20168:48 pmat

    Could go from 8th to 4th in one game.

    • 89 Chemistry writes:
      February 16th, 20169:01 pmat

      See my reply to USMCR [above].

      • USMC Retired writes:
        February 17th, 20168:08 amat

        So at this point we are the HARE, to the Alda’s tortoise. i think most early prognostications had the winner with 4/5 losses

        • USMC Retired writes:
          February 17th, 20168:11 amat

          Winning out and some other “upsets” will also help to alleviate, for the committee, some of the loses when DV was out or just coming back – the PU loss will just be there as an outlier and Cobs is a BAD loss.

  • Jeffrey Lubeck writes:
    February 17th, 20161:29 pmat


    This is really good stuff as always.

    Could you chart something similar on two different fronts?

    1. Legitimacy/Accuracy of Complaint Against Officiating (on TUS Site By its Members) versus Total Number of Complaints Against Officiating (on TUS Site By its Members)
    2. Legitimacy/Accuracy of Complaint Against Announcers (on TUS Site By its Members) versus Total Number of Complaints Against Announcers (on TUS Site By its Members)

    • USMC Retired writes:
      February 17th, 20162:40 pmat

      In the 1st case, it’s obvious and in the 2nd case it’s fingernails on a chalkboard

    • 89 Chemistry writes:
      February 17th, 20163:22 pmat

      Thanks for the attaboy.

      As for Project Complaint: By myself? No. But with full-time funding and about eight staffers, 89 ChemLube [“a limited liability corporation, as far as you know”] could whip up something covering the site’s entire history of the events you characterized by the end of the summer.

      • Jeffrey Lubeck writes:
        February 17th, 20163:55 pmat

        Starting the Grant writing effort tomorrow.

  • 89 Chemistry writes:
    February 17th, 20163:30 pmat

    MBB: Between tonight and tomorrow, Iowa, Maryland, and Indiana will play games in which all three are favored to win by about ten points.

    Tomorrow night, MSU hosts Wisconsin [9 PM, ESPN]. Tentatively, MSU is favored by ten points.

  • USMC Retired writes:
    February 17th, 20166:39 pmat

    Glad Glade may (I say May) be taking down the Alda’s

    • USMC Retired writes:
      February 17th, 20166:40 pmat

      They did, beating them 79-75; curiouser and curiouser

      • Jeffrey Lubeck writes:
        February 17th, 20168:48 pmat

        Hawkeyes looked completely disinterested as the game progressed. PSU thought about choking and giving a W away with 4ish left, but then decided to tough it out.

  • USMC Retired writes:
    February 17th, 20168:18 pmat

    Say what you will. but when college teams give up 93m points combined in the 1st half, I’d say neither team can spell Defense. It looks a little like the 2.7 seconds of the nab al-star game I watched the other night WOW (uh this is the everyones pick to win it all giving up 45 to a very limited(injuries) Duck team. Remember how we used to lament (and still did at the beginning of the year) how the Spartans couldn’t stop dribble penetration? those Spartans could teach a class to both of these teams, just awful.

    • Jeffrey Lubeck writes:
      February 17th, 20168:54 pmat


      For objectivity, I need to take my Green shaded glasses off. There that is better. Wow both teams look pretty much the same.

      How is Duke staying in this game – one down with 4 left.

  • Jeffrey Lubeck writes:
    February 17th, 20169:00 pmat

    Speaking of taking out opponents…

    Momma Mountain Lion and two baby cats in this picture across the road the other day (Zinc Spur) from the house.

    • USMC Retired writes:
      February 18th, 20165:42 amat


  • Jeffrey Lubeck writes:
    February 17th, 20169:15 pmat

    Three Top Ten teams go down Iowa (4), UNC (5) and Oklahoma (3) tonight.

    Watched the games and these teams all looked like MSU did during its three game scuffle.

    • USMC Retired writes:
      February 18th, 20165:47 amat

      teams are getting tired of the grind, esPn idiots are starting to snipe at Ben Simmons for not being all that, they forget he’s a Frosh and probably hit the wall recently. One thing you can always count on with esPn, if you don’t make them look good in their predictions (PREDICTIONS based on their pseudo intelligence) they will stab you in the back and twist it. I personally think that most of them need to be waterboarded, just for the experience – it gives one a sense of humanity (shows doesn’t it?)

  • 89 Chemistry writes:
    February 17th, 201610:10 pmat

    Iowa barely held on to beat the Gophers recently. At home.

    Will Penn State be a problem for MSU eleven days hence? That is too far ahead to fret over–but depending on how things go, that game may no longer be the gimme it looks to be in pixels.

    • USMC Retired writes:
      February 18th, 20165:38 amat

      Especially with the B1G homer refs calling 50+ fouls as they have in the past against MSU @PSU

      • USMC Retired writes:
        February 18th, 20166:19 amat

        My mistake, thought we were playing @Glad Glade, only Away games left are Acorns and then NJ. As delineated above by ’89, there will be some carnage amount those at the top of the standings going forward as they all play each other. A piece of the regular season title is on life support

      • 89 Chemistry writes:
        February 18th, 20168:15 amat

        MSU did go to and win @ PSU this year. It was Valentine’s first game back and right before MSU’s three game losing streak.

        Iowa & MD do not play again before the B1GT. They played once; MD holds the tiebreaker if those two finish atop the heap on March 6.

  • I Believe writes:
    February 18th, 20167:18 amat

    Tiger fans wishing to order the below T-shirt, can do so here:

    Lets All Cheer Our Baseball Team

Leave a Reply must be logged in to post a comment.

»  Substance:WordPress   »  Style:Ahren Ahimsa
© Copyright, The Usual Suspects - All rights reserved.
%d bloggers like this: