[Husker] A case for recruiting rankings

[Scott R Lawson]

Another interesting aspect of all this is to go to Rivals' site and
click on past top 100 lists, I believe it goes back to 2002 (maybe more
if you sign up) to see how many of the names on there you've heard of or
have made an impact in college. I can honestly say the 2002 list of the
top 100 players has about 10 names on it that I am familiar with, and I
like to think I follow college football a little more closely than the
average fan. To me, that says it all about recruiting rankings.

Scott in NY

-----Original Message-----
From: husker-bounces at tssi.com [mailto:husker-bounces at tssi.com] On Behalf
Of Brown, Laura (LLU)
Sent: Wednesday, January 23, 2008 2:58 PM
To: Steve Reichenbach; husker at tssi.com; nickchevance at gmail.com
Subject: RE: [Husker] A case for recruiting rankings

What I think would be more interesting would be to take all of the major
award winners or All Americans and see what their rankings coming out of
HS were.

Laura 

-----Original Message-----
From: husker-bounces at tssi.com [mailto:husker-bounces at tssi.com] On Behalf
Of Steve Reichenbach
Sent: Wednesday, January 23, 2008 11:49 AM
To: husker at tssi.com; nickchevance at gmail.com
Subject: Re: [Husker] A case for recruiting rankings

As was noted in another post, the records should have been staggered
to be after recruiting periods.  The most important missing elements
of the analysis are a quantification of the correlation, statistical
significance of the correlation, and variance in the error of
prediction.  That said, one of the truest observations of the article
is that this is "Duh!".  Of course, the recruiting ratings are an
imperfect measure of talent and potential and of course talent and
potential contribute to the likelihood of winning.  And, it seems
obvious that it is a causal relationship:  better players cause teams
to win more games (even if there are other factors).  Duh!

As noted in the post below, the real mistake is to interpret such a
correlation as being a basis for "confidently" predicting the future
of a single player or even a single team.  There are many confounding
variables (such as coaching and injuries) that significantly undermine
the predictiveness for limited outcomes.  That being said, it is a
causal relationship and it has been and almost certainly will continue
to be the case that over the universe of many players, teams, and
years, the relationship can be observed in the correlation.

Finally, one of the arguments against the importance of recruiting
ratings is a argument in favor of the causal relationship.  Critics of
recruiting ratings argue that (at least some) coaches are better judges
of talent and what a team needs than the recruiting ratings.  If that
is the case (as it well may be), the implication is that the coaches
are better able to judge something that is an important factor in
success, which implicitly acknowledges that recruiting ratings are a
measure of something quite important (even if they are imperfect).



> "There are three kinds of lies: lies, damned lies, and statistics."
> Benjamin Disraeli
> 
> First of all, Jon is correct.  The author is very good; he writes well
and
> clearly, and while the debate about the subject matter may go on for
years,
> the article is pretty easy to understand.  I enjoyed it.  Thanks, Jon.
> 
> Now, what we have here is a correlation.  The rankings, a rather
subjective
> measure of athletic skills and abilities, of recruits correlate with
the
> success of any given team on the field, as measured by wins and
losses.
> Some say its flawed.  You betcha.  Much like predicting the weather,
the
> massive amount of information available about the weather has to be
selected
> and simplified in order to make a prediction on an outcome that won't
take
> longer than it takes to actually observe the effect (i.e., have the
forecast
> for rain before it actually rains).  In this case, there may be
correlations
> between a whole host of factors, and the more data used and more
complicated
> the modeling, the more reliable the outcome (hopefully).  I'm not
convinced
> that ranking high school athletes is scientific enough to be more than
> informed guesswork, but its probably better than a correlation between
a
> running back's 40 times and the length of the team's cheerleader's
> eyelashes.  But because there are so many factors, which ones do you
choose?
> 
> And, in the end, does it demonstrate a cause and effect?  In other
words, is
> the correlation between Rivals rankings and winning percentage mean
that the
> more highly ranked recruits a team gets spells more wins on the field.
You
> can't say really, because that's what is really fun about
correlations.
> They are pretty easy to point out, but damn hard to prove.  Take the
link
> between smoking and lung cancer - the link was known for years, but it
took
> a long time to prove that link.  There are so many other factors that
go
> into the winning percentage factor alone (and that may be the only
"hard"
> fact in this study) that proving that link would be difficult at best.
> 
> And isn't that what sports talk is all about?  That's why we develop
all
> those statistics in the first place, and then shuffle them around
until it
> makes us go "Hmmmm, maybe it is the length of the cheerleader's
eyelashes
> after all!"  Its the wholesale conjecture about minutia that gets us
going,
> and makes us write long paragraphs to email lists when we should be
getting
> work done (ooops!).
> 
> A last note on the significance of statistics in sports.  I was at the
> Creighton - Drake game last night (no, I didn't wear a sweater but I
did
> drink some beer) and Drake was ranked 22/23 going in, but Creighton
was
> favored as much as by 7.  Did Drake have the better athletes and are
they
> the 22nd or 23rd best team in the country?  I have no idea what the
Rivals
> rankings were for their players, or for Creighton for that matter, and
I
> have no idea how either team would fare against the 21st or 24th best
team.
> What did matter is that one team rebounded really well, got shots
inside the
> paint (the easier, shorter shots) and got them to drop.  The other
team
> didn't rebound well, didn't get many inside shots, and even fewer to
drop.
> In short, the team that won scored the most points, and seemed to work
> harder at making that happen.  Now, I can get raw statistics for much
of
> that, but I'm not sure what the statistic to use to measure that last
part,
> the "effort" part (a confounder).  But its that last part that we'll
argue
> about because it is so hard to measure.
> 
> Nick
> -- 
> "Science is a good thing. News reporters are good things too. But it's
never
> a good idea to put them in the same room." Scott Adams
> _______________________________________________
> husker site list
> husker at tssi.com
> http://romaine.tssi.com/mailman/listinfo/husker
> 
> Note:  Comcast has been bouncing a lot of list traffic recently.  The
problem appears to be on their end.
> 

_______________________________________________
husker site list
husker at tssi.com
http://romaine.tssi.com/mailman/listinfo/husker

Note:  Comcast has been bouncing a lot of list traffic recently.  The
problem appears to be on their end.

_______________________________________________
husker site list
husker at tssi.com
http://romaine.tssi.com/mailman/listinfo/husker

Note:  Comcast has been bouncing a lot of list traffic recently.  The
problem appears to be on their end.