Statistical Prediction: Alabama Versus Georgia

[selmaborntidefan]

I've been using this for post-season for a few years now. You can evaluate how effective or accurate it is by looking at previous predictions, all of which I've been willing to make public, including the one at this link:

https://www.tidefans.com/forums/showthread.php?t=279457&p=2966032&viewfull=1#post2966032


I make no claims to have any supernatural knowledge, don't bet on this basis. But looking at the numbers and figuring, it has been very good at determining the final margin or close to it even if not all the details.

PAST USES OF THIS MODEL
2013 BCS Title Game: Alabama 28 Notre Dame 7
2014 Sugar Bowl: Alabama Ohio State pick 'em (34-34)
2015 Cotton Bowl: Alabama 31 Michigan State 10 (I fudged it to 17 because I couldn't believe it)
2016 Title Game: Alabama 27 Clemson 17 (got the score wrong but we were up by 12 with seconds left)
Super Bowl XLIX: Seattle 28 New England 24 (hey, if Pete Carroll hadn't phoned Les Miles......)
Super Bowl XL: Carolina 28 Denver 23
2016 Peach Bowl: Alabama 30 Washington 16
2017 CFPNCG: Alabama 31 Clemson 17
Super Bowl XLI: New England 31 Atlanta 27

The model so far is basically 6-3. One loss can be chalked up to Pete Carroll's stupidity and another to the unfortunate combination of Bo's injury and two pick plays - not to deny Clemson their deserved due.

Btw - some predictions even in the above post were accurate.
Alabama DID get 31 points.
Clemson DID turn the ball over twice.
My prediction was ALSO almost incumbent upon us getting a NOT - which we damn near did if you recall.
And the biggest failure? Bo getting hurt. Seriously - if he'd lasted one more drive, we win.

So let me explain what I do.

1) Eliminate the cupcake games that inflate stats.
2) Deduct the scores from within the game in question to determine the other stats for scoring offense/defense (e.g. when evaluating LSU's corrected scoring offense, I deduct their cupcake games PLUS the Alabama or Georgia game - that's why the totals are different).
3) Get a total average performance based on all the data points (8 for Alabama, 9 for UGA)
4) Figure how much weight to give to common opponents
5) I did NOT eliminate the stats regarding rushing and passing yards. Will it substantially affect them? Maybe, but it's getting too close to the game, and I've never done it before.

Next post will contain the data followed by a post with a prediction.

 

[selmaborntidefan]

ALABAMA VS GEORGIA

ALABAMA OFFENSE VS OPPONENT DEFENSES (47.5 ppg)
Louisville 51 (51.8) - equal
Ole Miss 62 (37.3) – plus 24.7
Texas A/M 45 (24.1) – plus 20.9
Arkansas 65 (36.8) – plus 28.2
Missouri 39 (23.0) – plus 16
Tennessee 58 (27.1) – plus 30.9
LSU 29 (18.8) – plus 10.2
MSU 24 (11.6) – plus 14.4
Auburn 52 (19.3) – plus 32.7
Average – 19.8 ppg ABOVE what defense permits

Alabama would score 38 points.

ALABAMA DEFENSE VS OPPONENT OFFENSES (14.1 ppg)
Louisville 14 (19.1) – minus 5.1
Ole Miss 7 (26.1) – minus 19.1
Texas A/M 23 (25) – minus 2
Arkansas 31 (17.4) – plus 13.6
Missouri 10 (38.2) – minus 28.2
Tennessee 21 (20.4) – plus 0.6
LSU 0 (29.6) – minus 29.6
MSU 0 (20.5) – minus 20.5
Auburn 21 (22.4) – minus 1.4
Average – 10.2 ppg BELOW what offense scores

Three of these offenses are ranked 75th or below.
Alabama would hold UGA to 29 points.

GEORGIA OFFENSE VS OPPONENT DEFENSES (39.6)
South Carolina 41 (28.6) – plus 12.4
Missouri 43 (23.6) – plus 19.4
Tennessee 38 (29.4) – plus 8.6
Vandy 41 (24.7) – plus 16.3
LSU 16 (17.4) – minus 1.4
Florida 36 (21.4) – plus 14.6
Kentucky 34 (15.6) – plus 18.4
Auburn 27 (19.3) – plus 7.7
Ga Tech 45 (31.7) – plus 13.3
Offense averages 12.1 above average permitted by opposing defense

Georgia would expect to score 26 points. (This sets UGA’s average to be expected points total between 26 and 29 points).

GEORGIA DEFENSE VS OPPONENT OFFENSES (18.7)
South Carolina 17 (27.5) – minus 10.5
Missouri 29 (36.3) – minus 7.3
Tennessee 12 (20.8) – minus 7.8
Vandy 13 (27.4) – minus 14.4
LSU 36 (28.0) – plus 6.0
Florida 17 (29.8) – minus 12.8
Kentucky 17 (24.3) – minus 7.3
Auburn 10 (23.6) – minus 13.6 (34% of Auburn’s points came in 2 games)
Ga Tech 21 (33.5) – minus 12.5
Average holding opponents 8.9 ppg below their average

Georgia could expect Alabama to score 39 points. This puts the Alabama expected points total between 38 and 39 points. This finding for both the Alabama offense and Georgia defense reinforces the likelihood.

Four of these 9 offenses are 75th or below nationally

COMMON OPPONENTS
Missouri (Alabama 29, UGA 14, +15 Alabama)
Auburn (Alabama 31, UGA 17, +14 Alabama)
Tennessee (Alabama 37, UGA 26, +11 Alabama)
LSU (Alabama 29, UGA -20, +49 Alabama)

Throwing out the anomaly of LSU, Tide still beats common opponents by about 13 points more than UGA. In two games, Alabama gave up more points on defense – though note one Vols score was a pick six).

EVALUATION

The best overall offense faced by Alabama was Missouri.
The best overall offense faced by Georgia was also Missouri.
Each team will be the best offense the other has faced.
The best defense Alabama faced was Miss State, and they doubled the ppg permitted by MSU. The best defense Georgia faced was Kentucky, and they likewise doubled the output.

PASSING YARDS (KEY GAMES)

ALABAMA 168
Ole Miss 344 (133)
16) Ark St 282 (218)
Mizzou 271.6 (142)
ATM 262 (263)
LSU 214.7 (184)
Auburn 209.5 (153)

GEORGIA 175

Note: Alabama BY FAR will be the best passing offense UGA has played.

UMass 298.1 (287)
SCAR 284.5 (282)
Mizzou 271.6 (221)
LSU 214.7 (200)
Auburn 209.5 (172)

RUSHING YARDS

ALABAMA (114)
Miss St 226 (44)
Missouri 197 (70)
LSU 174.7 (12)
Auburn 164.1 (130)

GEORGIA (128.4)
Georgia Tech 334 (128)
Florida 209 (170)
Kentucky 201 (84)
Missouri 197 (172)
LSU 174.7 (275)
Auburn 164.1 (102)

 

[selmaborntidefan]

OVERALL EVAULATION

Lost in Alabama’s thumping of opponents by no less than 22 points is that save for the LSU debacle, no team has been closer than 14 points to Georgia at the end of their games, either. Georgia has also played a substantially tougher schedule than Alabama (28 vs 55 per Sagarin), but these numbers are weighted heavy because Alabama played Ole Miss and Arkansas while Georgia was obligated to play bowl teams Florida and Kentucky.



Alabama has only one more penalty on the year but 65 more yards in penalties.

Both teams have lost 7 fumbles. Georgia has recovered 9 fumbles, Alabama 7.

CAVEATS

I already mentioned the schedule, but there is another anomaly that such evaluations have never before confronted: Alabama’s defensive numbers are misleading because the first-team defense rarely plays beyond halfway through the third quarter.

Look at a couple of games:
Auburn had 120 total yards in the first half, 163 in the second against Alabama. But they also were outscored, 35-7, so these were mostly garbage yards. By contrast, against UGA the Tigers had 179 yards in the first half and 95 in the second.

Does this suggest Georgia makes better half-time adjustments than Alabama does? Or does it suggest that teams pick up more yardage because second-team players are on the field for Alabama?

Or maybe look at Missouri. Missouri had 138 yards, 3 turnovers, and a missed FG and trailed the Dawgs at the half only 20-7. In the second half, they had 255 yards and were outscored, 23-22. Alabama gave up 175 yards and ten points, went in for halftime of a 30-10 game….and Mizzou had only 45 yards in the second half.

LSU? Alabama surrendered 77 yards in the first half, 119 in the second…and 0 points in either. It should be noted that ALL of those yards in the second half came after LSU was trailing, 22-0. Georgia, on the other hand, gave up 246 yards to LSU in the first half…and 229 in the second. To put the entire thing in perspective, LSU had more yards in each half against Georgia than in the entire game against Alabama.

In other words, Alabama’s first-string defense doesn’t “really” give up 168 passing yards per game, and the rushing defense doesn’t “really” give up 114. Georgia’s numbers are slightly misleading as well but not nearly as much as the Alabama defensive numbers.

What does this all mean?

Statistical analysis cannot tell us certain things, including the following important ones:
Injuries to key contributors
Bad officiating calls that flip momentum
Turnovers
Stupid coaching decisions like passing for a touchdown from the two-yard line with the Super Bowl championship hanging on the execution of the play


If BOTH teams give their average output for the year, we have the final score of:
Alabama 38
Georgia 28


However.......there's more to it than this.

1) Is there a home field advantage?

Under normal circumstances, I'd say yes. But Alabama has played more often in Atlanta in recent years than UGA so long as we're limiting our count to domed stadiums that either exist or were imploded but missed on The Weather Channel thanks to the MARTA bus.

So I would not give one in this instance. (I would have last January btw).

2) The 13-point average (corrected) performance of Alabama versus same opponents has enough data points to make a one-point increase for every touchdown margin. This creates an additional -2 for UGA or a +2 for Alabama.

3) Neither the penalties nor the turnovers provide a substantial margin of difference to benefit either team. Nor does the SOS. Georgia-Kentucky might have been a big deal...but when you then remember Kentucky lost to Tennessee........


Thus I make a final prediction of:

Alabama 38
Georgia 24

Do not be surprised if the first quarter ends in something like a 14-14 tie that conjures up images of Big 12 defenses. I expect some high energy before the settling into the rhythm of the game.

 

[Redwood Forrest]

You have overwhelmed me with data but I will take it. There is a lot of work in those stats.

 

[FF4bama]

Your word is good enough for me, Selma. I’ll take it!

 

[NationalTitles18]

Looks about right, give or take. We shall see.

Thanks for the effort and Roll Tide!

 

[deliveryman35]

We should win this game comfortably and by double digits. Whether that happens or not will depend on preparation and execution.

 

[cubzwin]

Rat poison

 

[ALA2262]

With an O/U of 64 and a point spread of 13.5, Vegas is expecting a score of 39-25. For that reason, I debated using a score of 38-24 on the score thread.

 

[selmaborntidefan]

With an O/U of 64 and a point spread of 13.5, Vegas is expecting a score of 39-25.

I did not know that.

Seriously.

 

[RTR91]

That’s about the same prediction I got when I did my stats earlier in the week in another thread. Still suspect Alabama eclipses 40 somehow.


Sent from my iPhone using Tapatalk

 

[ALA2262]

major tidefan

BamaNation All-SECJoin DateJul 2006LocationBoaz,ALPosts1,505Post Thanks / Like

Re: Pick the score: Alabama vs. Georgia (SECCG)

Alabama 38-24.​

 

[bamabelle1991]

WOW. I remember now why I failed Stats 101! haha! Super impressive!

 

[RTR2u]

That made me "swimmy-headed" selma, but I'll take it!:smile:

 

[cuda.1973]

WOW. I remember now why I failed Stats 101!

Don't be so hard on yourself. Hey, in your profession, it is probably a plus! :biggrin2:

(You deal in words, and making them say something other than what the rest of us think they mean.) :biggrin2:

 

[Bazza]

-snip-
Thus I make a final prediction of:

Alabama 38
Georgia 24
-snip-

Yeah yeah......very impressive. But can you tell us when that FG is going to be kicked?

 

[selmaborntidefan]

That’s about the same prediction I got when I did my stats earlier in the week in another thread. Still suspect Alabama eclipses 40 somehow.


Sent from my iPhone using Tapatalk

Same here.

If we don't top 40, I think it will be because we're leading like, say, 35-7, and go into "end the game" mode early.

 

[selmaborntidefan]

Yeah yeah......very impressive. But can you tell us when that FG is going to be kicked?

Sure.

When the ball is snapped on a fourth down play to a holder seven yards behind the center, and Bulovas takes four steps or fewer forward, kicks the ball, and it lands between two uprights and above a crossbar.

 

[B1GTide]

Sure.

When the ball is snapped on a fourth down play to a holder seven yards behind the center, and Bulovas takes four steps or fewer forward, kicks the ball, and it lands between two uprights and above a crossbar.

 

[JBama_in_PCOLA]

Very impressive. One thing also to consider that I didn't really see, the offense hasn't really played all that much in the second half this season either, albeit more towards the end of the season they did. I expect a championship because of that.