|
@patrick_oshag | |||||
|
The biggest challenge isn't building good models, its using them and sticking with them. pic.twitter.com/PtcFaHIfsz
|
||||||
|
||||||
|
Corey Hoffstein
@choffstein
|
21. sij |
|
Is that saying that on 4th and 1 on your own 10 you should go for it?
|
||
|
|
||
|
Andrew Miller
@millerak42
|
21. sij |
|
Yes, (although all 4th&1 not created equal) but the conversion rates on QB sneaks are extraordinarily high.
|
||
|
|
||
|
Gary Basin 🤳
@garybasin
|
21. sij |
|
Also principal + agent problems here. When you do the unorthodox thing, you risk looking dumb.
No one ever got fired for buying IBM
|
||
|
|
||
|
Gary Basin 🤳
@garybasin
|
21. sij |
|
Until everyone got fired
|
||
|
|
||
|
Brad Thibeau
@ThibeauBrad
|
21. sij |
|
ooo ooo, time for one of my favorite papers
nber.org/papers/w11270.…
|
||
|
|
||
|
mikedariano
@mikedariano
|
21. sij |
|
Brad, Massey has been talking about this (and others) paper on @WMoneyball podcast. He's softened on QB selection. Otherwise, great paper and great to see it followed (or not) in league.
|
||
|
|
||
|
Anthony Bardaro
@AnthPB
|
22. sij |
|
/1 These seem to express narrow probabilities of successful outcomes instead of incorporating broader considerations, like expected point values...
twitter.com/patrick_oshag/…
|
||
|
|
||
|
Anthony Bardaro
@AnthPB
|
22. sij |
|
/2 e.g. Getting stuffed on 4th & 1 from inside your own 10 yard line is probably something like a -5.5 point consequence -- disregarding the opportunity cost forgone from not strategically punting for field position... 🏁
|
||
|
|
||
|
Will Kane
@Kane_1200
|
21. sij |
|
I’m guessing those models don’t take into account second order effects like team morale and momentum, time on field for offense and defense, etc.
|
||
|
|
||
|
Jim Hutchinson
@HutchCPA_I
|
22. sij |
|
OK, planning on doing a simulation to see what works. pic.twitter.com/HMOptzX31l
|
||
|
|
||
|
James Barnsley
@JamesBarnsley30
|
22. sij |
|
The goalposts are in the right spot!
Must be an old model
|
||
|
|
||