Kind of interesting Wilner look

Kind of interesting Wilner look - Printable Version

+- The CardBoard (https://thecardboard.org/board)
+-- Forum: C-House! (https://thecardboard.org/board/forum-4.html)
+--- Forum: The CardBoard (https://thecardboard.org/board/forum-5.html)
+--- Thread: Kind of interesting Wilner look (/thread-13446.html)

Kind of interesting Wilner look - washingtonismoney - 02-02-2017

I haven't thought through the methodology here yet, but Wilner's got an article purporting to show 29% of four- and five-star Pac-12 QB recruits become impact starters, versus 17% of three-stars.

Of course there are a lot of confounding factors here, but it reinforces that the QB failure of '16 may not be *coaching's* fault per se -- after all, there's a 50.4% chance (assuming, of course, QB success is independent) that two random premium QB recruits won't be an impact starter. The real problem was Shaw et al.'s failure to take a QB in 2015. (Which, assuming independence, would reduce the odds to around 35% chance of three consecutive busts.)

http://www.mercurynews.com/2017/02/02/college-hotline-pac-12-quarterback-recruiting-think-you-found-a-future-starter-dont-be-so-sure/

Re: Kind of interesting Wilner look - CTcard - 02-02-2017

(02-02-2017, 02:13 PM)washingtonismoney link Wrote:Of course there are a lot of confounding factors here,

One of those confounding factors is that most QBs don't pan out. Since usually only one QB plays at a time, and more often that not that one turns into a multi-year starter, that means by definition that many QBs will not contribute significantly.

Under the circumstances, maybe 29% of higher rated QBs becoming impact starters is pretty good.

----
[Warning - the following contains details and calculations.]

Wilner did not say that 17% of three star QBs become impact starters. He kind of implied 17% of all recruits become impact starters, but then noted that is not true because the denominator of 99 would include 2016 true frosh (who usually redshirt). In fact many QBs don't become impact starters until after their redshirt frosh season, so the correction should be larger than that. But, just taking out the 2016s gives an overall 21%.

What number he would come up with for three stars is tricky, since I don't know which QBs make up the 17 he thinks became impact starters. I would guess that beyond the 10 4 and 5 stars, he would have chosen:
Kelly (3 star, ASU), Bercovici (3 star, ASU), Liufau (3 star, Colorado), Mariotta (3 star, Oregon), Wilson (3 stars, Utah), Halliday (3 star WSU), Falk (no stars according to Wilner, but actually 2 stars).
There are 43 non-2016 three stars listed, or 6/49=14%

There are also just some nutty choices made by Wilner.
He chose Troy Williams, Utah's starting QB this past year, as an impact starter. He did not choose Manny Wilkins, ASU's starting QB this past year. Wilkins had a substantially better year than Williams - a higher PER by 12, more yds/gm by 20, fewer INTs, more TDs, etc.
With such small numbers, these kinds of crazy mistakes makes the whole thing pretty bogus.

I would say that the point you took away is indeed the one solid item here: most 4/5 star QBs do not become impact starters. To what extent that is because they simply don't become very good versus being beaten out by other 4/5 star QBs would take further analysis.

Re: Kind of interesting Wilner look - washingtonismoney - 02-03-2017

Woof -- I knew Wilner's article was trouble when I couldn't figure out the methodology for determining what made for an impact starter.

Re: Kind of interesting Wilner look - winflop - 02-03-2017

Whether you question the methodology or not, 50% is average. I'm not OK with average.

Re: Kind of interesting Wilner look - CTcard - 02-03-2017

(02-03-2017, 06:37 AM)winflop link Wrote:Whether you question the methodology or not, 50% is average. I'm not OK with average.

I don't know what that means.
It's some kind of complaint about Stanford's QBs, yes?