Matt Hayes and lies, damned lies and statistics

You know what’s really getting tiresome these days?  The countless articles where a pundit “proves” that all this recruiting rating stuff is overblown.  Since not every five-star recruit turns into an All-American, what does that tell you, hunh?  (Never mind that the statistical likelihood of that happening is far greater than it is for four-stars, three-stars, two-stars, etc.)  And since every once in a while a two-star recruit does morph into an All-American, the whole system must be seriously flawed, right?

So let’s add Matt Hayes to the mix.  Matt’s done put hisself together one of them there scientific studies entitled “The best talent doesn’t always equal the most wins” that you ought to take a gander at if you’re in need of a quick chuckle.  His grand concept is to take an average of each D-1 team’s recruiting rankings for the past five seasons and match it against the average number of wins that team has had during that same time.  The bigger the spread between the two, the more evidence that a program has either underachieved or overachieved during that period.

How truly stupid is this?  Let me count the ways.

  1. His method correlates classes in the early years that don’t contribute.  How many players from a school’s 2004 recruiting class actually saw the field of play in 2004?  In most cases, damned few.
  2. Along the same lines, his analysis doesn’t take into account the talent level of the program prior to 2004.  A perfect example of that is his last place team, Mississippi.  By all accounts, Cutcliffe simply wasn’t a successful recruiter.  Orgeron was, but had to start significantly in the hole from a talent standpoint because of his predecessor.
  3. Strength of schedule isn’t factored in.  Look at the schools at the top of his list:  the Big East dominates that part, followed by the ACC.  Neither conference has really been at the top of its game in the last five years.
  4. Worst of all, the methodology cuts against the very best and the very worst teams.  If you’re the number one ranked team in recruiting, the best showing you can make under Hayes’ study is to break even.  Well, guess what:  that’s exactly what Southern Cal does.   So much for talent not equaling wins.  (That’s only good for #34 on Hayes’ list, by the way.)

That’s not to say that there aren’t overachieving and underachieving programs in college football.  Of course there are.  But Hayes’ model is a particularly inept way of sifting through the evidence to find which schools are which.

One thing is clear from what he compiled, though.  All things being even, a school’s chances to succeed on the field are better with good recruiting than not.  Seven of the top ten schools in recruiting finished in the top ten in wins during that period.  None of the bottom ten did.

Advertisements

2 Comments

Filed under College Football, Media Punditry/Foibles, Recruiting

2 responses to “Matt Hayes and lies, damned lies and statistics

  1. peacedog

    Meh. Dr Saturday already sagely noted that, yes, fewer 5*s turn out to be All Americans than 3*s. But the 5* players have a much higher chance to become AA’s. Of course, they should. . . there are vastly more 3 star than 5 star players (hundreds at minimum to dozens at maxmimum).

    And, of course Recruiting isn’t the end all be all. A highly talented, well coached group can and will still lose games. That’s without factoring in all that other stuff like SOS.

    Weak column, IMO.

    Like

  2. Low Country Dawg

    If you’re confused about the difference between correlation and causation, remember that beating the drums during an eclipse will always bring the sun back.

    Like