Stats models and rankings

[ibeattetris]

And here’s this:
Miami still in the ACC title race (they only have one loss)


Upset (we’ll show up in “did we really lose that bad?” this week, on the other side of the for Miami)

I’m surprised we had 43% postgame win expectancy according to him.

 

[cpf2001]

While we’re indulging in weird stats:


For future reference, it’s the GT-UM game, Miami’s odds of winning the game, in incredible whiplash.

Interesting to look at it that way and see that most of the time, that fumble doesn’t cause a loss since the other team likely won’t be able to score quickly enough from that field position.

In retrospect it felt inevitable and even live it wasn’t that shocking that they scored but if the offense had fizzled it would have actually been very unsurprising.

 

[ibeattetris]

No clue how I've missed this until now, but BCF toys provides drive level outputs for all games: https://www.bcftoys.com/2023-ps-w06#291
Would have been nice to know this when I was looking at things after the Louisville/Ole Miss games.

The overall game split for defense is good: https://www.bcftoys.com/2023-dgs#georgia-tech
Here is overall: https://www.bcftoys.com/2023-ogs#georgia-tech

GT held them to 1.54 PPD after surrendering 3.88 to BG. In Miami's previous three games their PPD was: 3.44, 3.73, and 4.75.

I doubt we can consistently rely on forcing five turnovers, but even if we remove the 5 drives that resulted in turnovers, we held them to 2.85 PPD on the remaining drives.

 

[slugboy]

No clue how I've missed this until now, but BCF toys provides drive level outputs for all games: https://www.bcftoys.com/2023-ps-w06#291
Would have been nice to know this when I was looking at things after the Louisville/Ole Miss games.

The overall game split for defense is good: https://www.bcftoys.com/2023-dgs#georgia-tech
Here is overall: https://www.bcftoys.com/2023-ogs#georgia-tech

GT held them to 1.54 PPD after surrendering 3.88 to BG. In Miami's previous three games their PPD was: 3.44, 3.73, and 4.75.

I doubt we can consistently rely on forcing five turnovers, but even if we remove the 5 drives that resulted in turnovers, we held them to 2.85 PPD on the remaining drives.

Miami’s offensive numbers that matter are top-25; even after Saturday. They’re #17 in points per drive.

Texas A&M is the only ranked team Miami has played, but they scored 38, then 48, then 48 again, then 38, and then—against us—20.

Great defensive performance.

 

[roadkill]

Miami’s offensive numbers that matter are top-25; even after Saturday. They’re #17 in points per drive.

Texas A&M is the only ranked team Miami has played, but they scored 38, then 48, then 48 again, then 38, and then—against us—20.

Great defensive performance.

Not trying to rain on our defensive parade, but the rain seemed to affect both QBs, particularly in the first half.

I do think the interceptions were not a lucky break but a consequence of our DBs being in the right place at the right time as well as a bit of disguising coverage. And we displayed better tackling too.

 

[slugboy]

Not trying to rain on our defensive parade, but the rain seemed to affect both QBs, particularly in the first half.

I do think the interceptions were not a lucky break but a consequence of our DBs being in the right place at the right time as well as a bit of disguising coverage. And we displayed better tackling too.

The interception were smart plays. One was a dumb play by Van Dyke, but even then it was smart defense

I saw some passes affected by the rain, but I think Miami’s team speed was a bigger factor. Unless the BC game is terrible, I think our improved play was the bigger factor on our side.

On defense, if one or two players don’t go all out, or make a bad play, the other nine or ten suffer. It doesn’t even have to be the same player every down.

Giving up long first downs will exhaust a defense.

Every defense is a “bend but don’t break” defense—the question is HOW much bend and how much break? If you tune it wrong, you either give up too many big plays or too many first downs.

Missed tackles weren’t that problem, though. A good tackle takes more effort each play, but it gets you off the field faster. Gang tackles mean everyone runs more on each play, but for fewer plays. I hope the coin dropped this weekend

 

[slugboy]

We're in week 6. At this point, almost all of the bias should be out of the stats models--if not now, in a week or two.

FEI does not "like" us. It doesn't like Deion's team, either.
After Miami, our defense has risen to "below average" or "bottom quartile of the P5". Our offense fell to average. I'll take the W, though.

Rk​	Team	Rec​	FBS​	FEI​		OFEI​	Rk​	DFEI​	Rk​
61​	South Alabama​	3-3​	2-3​	.04​		-.01​	51​	.07​	67​
62​	Arizona​	3-3​	2-3​	.03​		.10​	43​	.00​	73​
63​	Liberty​	5-0​	5-0​	-.01​		.15​	38​	.02​	71​
64​	Boise State​	3-3​	2-3​	-.01​		.11​	41​	-.04​	79​
65​	Georgia Southern​	4-1​	3-1​	-.02​		.01​	50​	-.19​	89​
66​	Marshall​	4-1​	3-1​	-.02​		-.49​	87​	.46​	36​
67​	Wyoming​	5-1​	4-1​	-.02​		-.40​	83​	.28​	54​
68​	Appalachian State​	3-2​	2-2​	-.03​		-.02​	52​	.16​	59​
69​	Baylor​	2-4​	1-4​	-.05​		-.17​	66​	-.06​	80​
70​	Michigan State​	2-3​	1-3​	-.05​		-.51​	90​	.31​	48​
71​	Louisiana​	4-2​	3-2​	-.06​		.31​	26​	-.40​	106​
72​	Western Kentucky​	4-2​	3-2​	-.09​		-.21​	70​	.05​	69​
73​	Pittsburgh​	1-4​	0-4​	-.10​		-.67​	107​	.37​	42​
74​	Georgia Tech​	3-3​	2-3​	-.11​		-.12​	59​	.03​	70​
75​	Colorado​	4-2​	4-2​	-.12​		-.03​	53​	-.37​	103​

 

[slugboy]

Here's the overall for the conference, including the "new guys" but not Notre Dame

Rk	Team	Rec	FBS	FEI	Column1	OFEI	Rk_1	DFEI	Rk_2
10	Florida State	5-0	5-0	.75	ACC	.94	12	.58	29
17	North Carolina	5-0	5-0	.60	ACC	1.13	9	.19	57
21	Clemson	4-2	3-2	.56	ACC	.19	34	1.29	2
24	Duke	4-1	3-1	.53	ACC	.30	27	.69	18
26	Louisville	6-0	5-0	.45	ACC	.41	22	.52	30
31	Miami	4-1	3-1	.38	ACC	.13	39	.60	25
43	Syracuse	4-2	3-2	.20	ACC	.19	33	.28	52
49	SMU	3-2	2-2	.10	ACC	.22	32	.11	64
56	NC State	4-2	3-2	.07	ACC	-.59	98	.62	23
59	Wake Forest	3-2	2-2	.04	ACC	-.04	55	.30	51
60	California	3-3	2-3	.04	ACC	-.15	64	.31	50
73	Pittsburgh	1-4	0-4	-.10	ACC	-.67	107	.37	42
74	Georgia Tech	3-3	2-3	-.11	ACC	-.12	59	.03	70
76	Virginia Tech	2-4	2-4	-.12	ACC	-.35	79	-.10	82
86	Boston College	3-3	2-3	-.24	ACC	-.20	69	-.37	102
88	Virginia	1-5	0-5	-.26	ACC	-.38	82	-.10	83
102	Stanford	1-4	1-3	-.43	ACC	-.58	96	-.37	99

 

[GetYourBuzzOn]

SP+ ratings are similar to FEI- GT improves to -.6 after the MIA win. I believe SP+ is having its best year yet with predictions; all stats are tracked here.

	TEAM	RATING	OFFENSE	DEFENSE	SPEC. TEAMS
60	Toledo	1.9	30.7 (39)	28.9 (78)	0.0 (65)
61	Mississippi State	1.7	30.4 (41)	29.0 (79)	0.4 (8)
62	UL-Lafayette	1.5	32.2 (32)	31.0 (94)	0.3 (15)
63	Nebraska	1.4	22.5 (93)	20.7 (24)	-0.4 (119)
64	Minnesota	1.2	24.1 (83)	23.2 (46)	0.3 (19)
65	SMU	1	29.5 (48)	28.3 (74)	-0.3 (113)
66	Houston	0.7	31.7 (35)	30.8 (91)	-0.2 (100)
67	Coastal Carolina	0.4	28.6 (52)	28.0 (69)	-0.2 (91)
68	Illinois	0.4	23.4 (85)	23.0 (44)	0.0 (67)
69	California	-0.4	27.0 (64)	27.1 (66)	-0.4 (128)
70	Georgia Tech	-0.6	28.0 (57)	28.3 (75)	-0.3 (105)
71	Wyoming	-0.9	23.2 (86)	24.3 (52)	0.3 (22)
72	Liberty	-0.9	27.0 (66)	27.7 (67)	-0.2 (93)
73	Purdue	-1.1	25.6 (74)	26.4 (63)	-0.4 (115)
74	UTSA	-1.7	28.2 (56)	29.7 (87)	-0.2 (94)
75	Baylor	-1.7	26.7 (68)	28.2 (71)	-0.2 (103)
76	Virginia Tech	-2.1	21.7 (96)	23.9 (51)	0.2 (44)
77	Marshall	-2.1	21.9 (94)	23.7 (49)	-0.4 (118)
78	Miami-OH	-2.2	22.5 (92)	25.0 (55)	0.4 (1)
79	Ohio	-2.6	21.3 (101)	23.5 (47)	-0.4 (131)
80	Colorado	-2.8	29.5 (49)	32.1 (105)	-0.1 (87)

 

[slugboy]

I wouldn’t read too much into it, but if you’re in the upper right you’re doing well

 

[MWBATL]

Our stats in these systems are being seriously hurt by the BG game. I wonder what we’d look like without that factored in?

 

[alagold]

Our stats in these systems are being seriously hurt by the BG game. I wonder what we’d look like without that factored in?

Unfortunately ,that is how we played.It was almost as strange as the uM game overall.We were 21 pt favorites and got destroyed.Worst than MidTnSt game and almost asbad as citadel.

 

[slugboy]

This is an interesting stat—your team’s scoring margin vs. the average top 25 team against the same schedule.

We’re mid—it’s a more favorable view than other views of our team.



Full image:

 

[roadkill]

This is an interesting stat—your team’s scoring margin vs. the average top 25 team against the same schedule.

We’re mid—it’s a more favorable view than other views of our team.



Full image:
View attachment 14923

Maybe it's just a coincidence, but this ranking has us in almost exactly the same spot: https://www.colleyrankings.com/currank.html
Colley's is kinda the opposite process in that it ignores margin of victory. Also, it thinks we have the 17th toughest SoS.

 

[GetYourBuzzOn]

This is an interesting stat—your team’s scoring margin vs. the average top 25 team against the same schedule.

We’re mid—it’s a more favorable view than other views of our team.



Full image:
View attachment 14923

Thanks for sharing. Maybe a dumb question, but I'm curious how they are defining their top 25 FBS teams. Very similar to the power rank concept that we've discussed here, where those show point differentials over the average FBS teams (vs Top 25 teams). Also, I would guess that our narrow loss to LOU gives us that boost into the 50s in this system.

 

[slugboy]

Thanks for sharing. Maybe a dumb question, but I'm curious how they are defining their top 25 FBS teams. Very similar to the power rank concept that we've discussed here, where those show point differentials over the average FBS teams (vs Top 25 teams). Also, I would guess that our narrow loss to LOU gives us that boost into the 50s in this system.

Since it’s Kelley Ford, my guess is that he takes the average power rank of his top 25 teams on the power rank, runs it against your schedule, and compares scores vs predicted scores.

 

[iceeater1969]

I saw a post on twitter saying that Gt held Miami qb Van Dike to a QBR rating of 45.
Is that low?
As a comparison: Is it lower than Against Auburn?

(My eye ball test gave both qbs some slack due to rain, I was shocked at how many interceptions we had on him. TVD has a great release. )

Just looking for a stat that gives a ray of hope for our D.

 

[MWBATL]

I saw a post on twitter saying that Gt held Miami qb Van Dike to a QBR rating of 45.
Is that low?
As a comparison: Is it lower than Against Auburn?

(My eye ball test gave both qbs some slack due to rain, I was shocked at how many interceptions we had on him. TVD has a great release. )

Just looking for a stat that gives a ray of hope for our D.

Yeah, it's very low. TVD has a season average of almost 85. No one is above 94 and TVD is 10th best in the nation. And that's AFTER the 46 we held him to. I imagine it was the interceptions that hurt him so badly. Haynes King is 45th in the nation with a QBR of 66 (his QBR vs Miami was 59).

 

[slugboy]

Yeah, it;s very low. TVD has a season average of almost 85. No one is above 94 and TVD is 10th best in the nation). And that's AFTER the 46 we held him to. I imagine it was the interceptions that hurt him so badly. Haynes King is 45th in the nation with a QBR of 66 (his QBR vs Miami was 59).

Van Dyne was second in the country in QBR before he played us. He was around 94.
Now, he’s behind Maye at second in the conference.

Either an off day by Van Dyne, or great defense by us, or both

 

[MWBATL]

Either an off day by Van Dyne, or great defense by us, or both

...or the rain helped us A LOT