S&P+ rankings: Overperformance or overcorrection?

[J.R. Juniors Junior Jr. 1084]

On 9/29/2019 at 11:06 AM, texifornia said:

We're #16. OU is 3, aggy is 23

 

Nebraska at #39 and Boise at #50? Yeah, no.

Edited September 30, 2019 by J.R. Juniors Junior Jr.

[texifornia 56945]

Man that FEI model (which isn't Connolly) haaaaates us.

[Burt Macklin 19023]

3 minutes ago, texifornia said:

Man that FEI model (which isn't Connolly) haaaaates us.

FEI has historically liked Herman/Orlando teams, but it’s more about per drive efficiency, so the LSU game hurt us quite a bit and while we had great YPP against OSU, our failures to finish a few drives held us back. I expect FEI will have is higher than SP by the end of the year, in part because we’ll do a better job finishing drives than we have so far.

 

[Magus Ossis 2799]

If only there were a stat that measured what percent of FBS games you actually won. Not sure how you’d code the algorithm for something like that.

[satyanash 14484]

Texas 38, WVU 24 is SP+'s prediction for the week. Gives us a 79% chance of winning.

So far this season, SP+ is 58% ATS (although Bill speculates it'll start to drop once preseason projections fade out of the formula)

 

[NeverMarryAStripper 21906]

We do a weekly CFB poll in the office. 12 games selected with the Vegas Wednesday spread. For 4 weeks I’ve used Sagarins’s Predictor for my evaluations and gone 34 out of 48 (70.8%) One week I just went with my gut and went 4 out of 12. I’m still 38 out of 60 (63.3%) overall and in 1st place.

[OrangEngr 1035]

9 hours ago, texifornia said:

Man that FEI model (which isn't Connolly) haaaaates us.

Meh, our overall rank isn’t that much different. It’s just in love with Baylor and TCU, for some inexplicable reason

[Walden Ponderer 12639]

South Florida -11 at Connecticut?!?!?!

Somebody's smokin' something. And I know absolutely nothing about Connecticut; I just know I wouldn't bet on South Florida if they were playing Manor.

[Pato del Muerto 52494]

Those efficiency based numbers don’t, can’t, account for teams that put backups in for reps or because of injuries, and how often they do so, and what impact it has. 

Had Texas run the score into the 50s or above with its starters vs rice or latech, which also likely results in a better defensive performance late, it’s overall numbers would look better. 

Idk how other teams have handled their lopsided games. 

[texasstrong12 4668]

9 minutes ago, Walden Ponderer said:

South Florida -11 at Connecticut?!?!?!

Somebody's smokin' something. And I know absolutely nothing about Connecticut; I just know I wouldn't bet on South Florida if they were playing Manor.

UCONN is awful. They won 1 game last year and might only win 1 or 2 games this year. UCONN's one win this year was a 3 point win over Wagner. 

Edited October 2, 2019 by texasstrong12

[TheFlyingBoat 906]

28 minutes ago, Walden Ponderer said:

South Florida -11 at Connecticut?!?!?!

Somebody's smokin' something. And I know absolutely nothing about Connecticut; I just know I wouldn't bet on South Florida if they were playing Manor.

When I saw you complaining about the spread I actually thought you were complaining the other way. There are DII teams that could beat UConn. They could lose to a team of ghosts.

Edited October 2, 2019 by TheFlyingBoat

[Walden Ponderer 12639]

10 minutes ago, TheFlyingBoat said:

There are DII teams that could beat UConn.

Well, this weekend will be a good test of that hypothesis, then, because if USF can beat them, then yes, there are some DII teams that could beat them, too.

[TheFlyingBoat 906]

Internet is dumb ignore 

Edited October 2, 2019 by TheFlyingBoat

[gmr548 16580]

Those efficiency based numbers don’t, can’t, account for teams that put backups in for reps or because of injuries, and how often they do so, and what impact it has. 
Had Texas run the score into the 50s or above with its starters vs rice or latech, which also likely results in a better defensive performance late, it’s overall numbers would look better. 
Idk how other teams have handled their lopsided games. 

Pretty sure S&P has a garbage time qualifier based on score margin and quarter that encompasses a good chunk of those situations

Sent from my SM-G920V using Tapatalk

[dingleberryswitzer 2724]

10 hours ago, Magus Ossis said:

If only there were a stat that measured what percent of FBS games you actually won. Not sure how you’d code the algorithm for something like that.

That ranking would look like a mack brown depth chart.

1) Alabama or AuburnClemson or Florida, or lsu or  Ohio St or  Wisconsin etc.

12) Notre  Dame or Texas or Oregon or Washington, etc

[texifornia 56945]

38-24 was pretty close for our game.

 

[texifornia 56945]

Mizzou???

 

[texifornia 56945]

We dropped to #20. aggy up to #19, OU held steady at #3.

 

[Girdwood 118]

Missouri at 10 🤡🤡🤡🤡🤡🤡🤡

[satyanash 14484]

Win probability distribution for the week. Expected season win total is still at 7.9, with a 23% chance of beating blOU this weekend.

[Burt Macklin 19023]

24 minutes ago, satyanash said:

Win probability distribution for the week. Expected season win total is still at 7.9, with a 23% chance of beating blOU this weekend.

We’d have to go 3-4 the rest of the way to hit Bill’s preseason projection for us and 4-3 to meet his current projection for us.

Good luck wit that, Bill.

[texifornia 56945]

Bill's turnovers = luck weirdness is BACK - our win expectancy was 47%

 

Edited October 8, 2019 by texifornia

[Huckleberry 32859]

lulz

Love his continued hubris that turnovers are luck if they don't match his model. It's not like every preview talked about how Texas could count on multiple turnovers from Kendall. Oh wait, scratch that, that's actually exactly what every preview said.

[bluto 4928]

And the entire first half was Sam converting 3rd and longs RE his ehlinger was bad on passing downs

[HtownHorn 14678]

39 minutes ago, Huckleberry said:

lulz

Love his continued hubris that turnovers are luck if they don't match his model. It's not like every preview talked about how Texas could count on multiple turnovers from Kendall. Oh wait, scratch that, that's actually exactly what every preview said.

All 4 of those INTs were great plays made by the defense. Luck is the fumble on the last play of the game last year against OU, that somehow counted as something really big in his calculations.

[Valmy77 13226]

We were up 35-17 but it was luck that we won? 

[OrangEngr 1035]

15 minutes ago, HtownHorn said:

All 4 of those INTs were great plays made by the defense. Luck is the fumble on the last play of the game last year against OU, that somehow counted as something really big in his calculations.

That's where I struggle. Fumbles are true 50/50 balls? Sure. But I believe his interceptions percentage is based on an assumed "INTs/ passes defended" percentage, which feels incredibly nebulous to me.

[JBJ 3654]

30 minutes ago, Huckleberry said:

lulz

Love his continued hubris that turnovers are luck if they don't match his model. It's not like every preview talked about how Texas could count on multiple turnovers from Kendall. Oh wait, scratch that, that's actually exactly what every preview said.

He's got the idea that PBUs are indicators of interceptions even though there's barely any correlation.

[OrangEngr 1035]

1 hour ago, texifornia said:

Bill's turnovers = luck weirdness is BACK - our win expectancy was 47%

 

Ron/ "Actually, nobody won the third quarter[\howard]

[Magus Ossis 2799]

He has a funny way of saying, “my model was exposed as weak in this instance.”

[Tex Pete 20351]

50 minutes ago, OrangEngr said:

That's where I struggle. Fumbles are true 50/50 balls? Sure. But I believe his interceptions percentage is based on an assumed "INTs/ passes defended" percentage, which feels incredibly nebulous to me.

Some fumbles are random, others aren’t. See: Derrick Johnson in college. Recoveries also aren’t necessarily random, either. Some players are just better at getting the ball. 

[Walden Ponderer 12639]

51 minutes ago, Tex Pete said:

Some fumbles are random, others aren’t. See: Derrick Johnson in college. Recoveries also aren’t necessarily random, either. Some players are just better at getting the ball. 

You can improve your fumble recovery percentages with good scramble drills.

Luck is a small component, not a large one. Good turnover margins are a cause of wins, not a symptom of them.

[Girdwood 118]

Connelly showing his ass again. Clown.

[Sejjr 3246]

This fucking dweeb. Of course he has Mizzou at 10. 
 

and fuck you too, “Fozzz” for the neg. 

Edited October 8, 2019 by Sejjr

[texifornia 56945]

Still hoping aggy justifies its ranking, still thinking we're a paper tiger

 

[Johnny Sack 8397]

WVu scores on 4th down at the end of the game

[ChemicalHorn 355]

What is dipshit’s definition of “garbage time”? As stated, we had more than doubled their score with 10 minutes left in the game and they finally scored again with ~4 minutes left to cut it to a two score game. Even ESPN’s goofy win probability model had us at >99% that entire time.

Edited October 8, 2019 by ChemicalHorn

[NoName 16351]

6 minutes ago, ChemicalHorn said:

What is dipshit’s definition of “garbage time”? As stated, we had more than doubled their score with 10 minutes left in the game and they finally scored again with ~4 minutes left to cut it to a two score game. Even ESPN’s goofy win probability model had us at >99% that entire time.

for 2018 at least: 

Quote

Garbage time adjustments: The S&P+ figures used in the tables below only look at the plays that took place while a game was deemed competitive. Garbage-time plays and possessions have been filtered out of the calculations. The criteria for "garbage time" are as follows: a game is not within 38 points in the second quarter, 28 points in the third quarter, or 22 points in the fourth quarter.

 

[texifornia 56945]

He has OU covering the spread, just

 

[Pato del Muerto 52494]

I’ll be pretty disappointed if we only get 30. Unless we win. 

[Viper 12109]

23 hours ago, NoName said:

for 2018 at least: 

 

Seems like you'd want to be a little more granular in the 4th quarter

[sushihorn 3233]

Connelly just can't figure out Texas and it makes him a pissy, irrational little bitch.

[texifornia 56945]

10 minutes ago, sushihorn said:

Connelly just can't figure out Texas and it makes him a pissy, irrational little bitch.

The fact that it's a team he has two major reasons to dislike has to rankle him more than if it was, say, West Virginia. If it was Mizzou or OU he'd definitely be a lot more laissez-faire about it.

Edited October 9, 2019 by texifornia

[DanRydell 799]

You can improve your fumble recovery percentages with good scramble drills.

https://www.fromtherumbleseat.com/2014/7/7/5868733/intro-to-football-analytics-fumble-recoveries-are-random

[Valmy77 13226]

Sure recovering fumbles is a crapshoot but causing them isn't.

[Girdwood 118]

17 hours ago, texifornia said:

He has OU covering the spread, just

 

[DanRydell 799]

11 hours ago, Valmy77 said:

Sure recovering fumbles is a crapshoot but causing them isn't.

SP+ agrees with that, no? The stat attributes recovering more or less than 50% of fumbles to luck, it doesn't attribute causing more or less fumbles to luck.

Edited October 10, 2019 by DanRydell

[Valmy77 13226]

3 hours ago, DanRydell said:

SP+ agrees with that, no? The stat attributes recovering more or less than 50% of fumbles to luck, it doesn't attribute causing more or less fumbles to luck.

Does it? Bill suggested we only won against West Virginia because we got lucky with 4 interceptions which kind of suggested he thought all turnovers were luck.

[sushihorn 3233]

14 minutes ago, Valmy77 said:

Does it? Bill suggested we only won against West Virginia because we got lucky with 4 interceptions which kind of suggested he thought all turnovers were luck.

It's hard to tell where his model ends and his personal opinion starts.  It's very unprofessional and a perfect example of the Appeal to (false) Authority Fallacy.

[JBJ 3654]

6 minutes ago, Valmy77 said:

Does it? Bill suggested we only won against West Virginia because we got lucky with 4 interceptions which kind of suggested he thought all turnovers were luck.

Interceptions are based on his expected number of passes defended.

In his mind a PBU is a near-interception like a fumble caused is a near fumble recovery.

Problem is fumble recovery rate is pretty consistently 50/50 across all defenses and players.  Interception to PBU rate isn't.