[Husker] HuskerMath - Is NU's recruiting on Solid Ground?

[Steve Reichenbach]

> I'm not sure I buy the underlying premise that recruiting numbers
> translate into on-the-field performance.

That was the other point on which I was going to comment.  Even the
best correlation he claims is less than 0.6.  While, the relationship
between recruiting ranking (which itself is corrupted recruits of
more successful programs getting better ratings) and winning almost
certainly is real, the correlation is quite low, which makes for a
wide range of uncertainty.  This page cites a range of uncertainty
for W/L% of 20%-30% (and doesn't say what the probability for those
ranges is).

Also, on the nature of statistics, they are pretty good about forecasting
average behavior, but not so good at forecasting individual behavior.
Some coaches will routinely out-perform and others will routinely
under-perform relative to what is only a loose statistical relationship.