OT - paging Craig Ross

Submitted by Chipper1221 on October 13th, 2022 at 11:54 AM

CR,  i just listened to the round table and found your experiment with net yards per pass attempt interesting.

 

As someone who gambles far too much I’m always looking for new stats to analyze for my picks. Do you have a running table with all the teams and this info? I’d love to use those numbers to my advantage this weekend. 
 

 

Chipper1221

October 13th, 2022 at 11:55 AM ^

Decided to make this a topic so that my fellow gamblers could chime in. 
 

Craig, even if you could just share the formula I’d like to play around with it and test it for a few weeks and monitor the results. 

Joined: 08/13/2014

MGoPoints: 16247

Hab

October 13th, 2022 at 3:16 PM ^

I think you'd have to cut in CR for a portion of whatever winnings that might come.  And I'm not sure you can afford his hourly rate.

Joined: 01/02/2015

MGoPoints: 39676

CR

October 13th, 2022 at 4:45 PM ^

Oh, sigh.

I wrote about this idea in Obscene Diaries and in Search for the Unified Field Theory. As to the former my guess is you can find it for about $1 or maybe less at my house. Since the publisher went into bankruptcy I get zero from that.one except a framed notice from the Court. Theoretically I get royalties on the latter but since I haven't seen a check in a couple of years, I assume no/few copies have been sold. Either that or the company is owned by John O'Neil and I am being stiffed. I prefer the latter. So it goes, Billy Pilgrim.

 The "formula" is simple. Take offensive and defensive YPPA after adjusting for sacks, since sack yards are lumped into run yards. Compare the differences between the two teams. So, this week that's a plus (approx) 2.5 favoring UM. Multiply by 5. You now have the "neutral field and no TO" variance between the two teams, of about 12.5 points. HFA is worth something. TOs are not predictable. SOS? Relevant, I would say. 

Why 5? I ran hundreds of trial and error and that seemed to fit outcomes. No Science High School.

A couple of things. First, it tends to be better at backward evaluation than forward evaluation. I have posted (at least once) on MGoBlog showing the ordering of team quality at season's end and it has done pretty well there, and variances are explicable. Second, it has worked reasonably well (prospectively) when I have paid attention to the idea in UM games. 

Shouldn't this be adjusted by SOS? Sure, but I haven't done anything more than eyeball this; this week, e.g, I see no real variance in the quality of the opposition, though I certainly could be wrong.  

But third, it was also no better than flipping a coin in NFL games when I played around with it a while back, though it did reasonably in playoff games. I don't suggest this is a way to handicap games. It might be a useful adjunct tool. Caveat emptor. 

 I would really be happy if some smart math guy would refine my crude algorithm or even conclude it is complete horseshit. Maybe someone already has; or maybe it is a flat earth theory. But, I believe, it is a pretty good after the fact idea, at least, worth playing around with..

Thanks for asking.

Joined: 10/23/2009

MGoPoints: 2368

Chipper1221

October 13th, 2022 at 5:57 PM ^

You’re a fan favorite. When i ran the numbers myself (removing Cade’s stats) I’m getting 3 point difference in Michigan’s favor. So, 15 on a neutral. 
 

Are you using this formula - 

((passing yards - sack yards) + (20 * TD) - (45 * int)) / (pass attempts + sacks) 

Joined: 08/13/2014

MGoPoints: 16247

CR

October 13th, 2022 at 6:24 PM ^

I used Cade's stats in the mix

Your idea is interesting.

Mine is

Pass Yards-Sack yards/Passes plus sacks events

minus

Pass Yards Allowed -sack yards/Passes plus sack events

As a team number then compared to the opponent

Times 5

So, I have UM as plus 4,48 (a number so big not sure I trust it) minus PSU  plus 1.91

Times 5 is 12.85

 

Joined: 10/23/2009

MGoPoints: 2368