Predictor Week 4

[LegendaryGT]

So last week's was a disaster. I'm blaming the combination of data from last year and dumb stats from demolishing two nobodies at the front of our schedule. Maybe it's just the optimist in me, but I still believe we are way better than we played, even though our stats against ND are way better than they should be due to two very very late touchdown drives. To make a long story short: there's no way this prediction is accurate, we've just been playing too inconsistently in several variables for this algorithm.

GT vs DUKE
Week 4 of 2015

GT - Offense
Yardage: 466 ypg
Performance: 48 (higher is better)
Scoring: 41
Efficiency: 11 (lower is better)
Performance: 32 (higher is better)
GT - Defense
Yardage: 325 ypg allowed
Performance: 31 (higher is better)
Scoring: 18
Efficiency: 18 (higher is better)
Performance: 38 (higher is better)

DUKE - Offense
Yardage: 406 ypg
Performance: 16 (higher is better)
Scoring: 29
Efficiency: 13 (lower is better)
Performance: 15 (higher is better)
DUKE - Defense
Yardage: 301 ypg allowed
Performance: 19 (higher is better)
Scoring: 17
Efficiency: 17 (higher is better)
Performance: 13 (higher is better)

Prediction:

Pure statistical: GT 36 (452 yards), DUKE 16 (312 yards)
GT best case: GT 46 (603 yards), DUKE 11 (262 yards)
DUKE best case: GT 29 (373 yards), DUKE 24 (380 yards)

GT talent: 5051
DUKE talent: 4312
GT talent of schedule: 7681
DUKE talent of schedule: 4428

Suggested talent adjustment: 19% towards GT best case
Final: GT 38 (482 yards), DUKE 15 (302 yards)

I'll do other games by request. I'm making a slight modification to the predictor poll, which I'll post later in the thread.

EDIT: Using only 2015 data:

GT vs DUKE
Week 4 of 2015

GT - Offense
Yardage: 493 ypg
Performance: 41 (higher is better)
Scoring: 52
Efficiency: 9 (lower is better)
Performance: 49 (higher is better)
GT - Defense
Yardage: 298 ypg allowed
Performance: 48 (higher is better)
Scoring: 15
Efficiency: 19 (higher is better)
Performance: 28 (higher is better)

DUKE - Offense
Yardage: 498 ypg
Performance: 45 (higher is better)
Scoring: 34
Efficiency: 14 (lower is better)
Performance: 0 (higher is better)
DUKE - Defense
Yardage: 249 ypg allowed
Performance: 38 (higher is better)
Scoring: 8
Efficiency: 28 (higher is better)
Performance: 97 (higher is better)

Prediction:

Pure statistical: GT 23 (287 yards), DUKE 15 (369 yards)
GT best case: GT 32 (379 yards), DUKE 11 (255 yards)
DUKE best case: GT 17 (171 yards), DUKE 23 (492 yards)

GT talent: 5051
DUKE talent: 4312
GT talent of schedule: 4628
DUKE talent of schedule: 2413

Suggested talent adjustment: 14% towards GT best case
Final: GT 24 (300 yards), DUKE 14 (353 yards)

 

[Akinji07]

USC @ Sun Devils pls.

 

[GTCrew4b]

The predictor had no way of knowing just how prepared and well executed ND's scheme would be to stop us. Had they lined up and played the way they did against UVA and Texas we would've been in line with the predicted outcome.

We are a hard team to predict due to teams using a modified defensive scheme against us.

 

[ScionOfSouthland]

Duke will score more than 15 points - just my two cents

 

[GTCrew4b]

Duke will score more than 15 points - just my two cents

why do you say that?

They amassed 17 against Tulane through 3 quarters and can't throw the ball downfield.

 

[jacket67]

Duke will score more than 15 points - just my two cents

I think CPJ is going to bring out the "3 yards and a cloud of dust" (splattering of mud?) offense for this game. I say that because we have limited experience at AB and WR, but the OL should be able to push the Duke DL enough to move the chains with the dive all day. The number of possessions could be really low for this game, so Duke getting 2-3 scores might be a real possibility. The weather will make it hard for either team to break big plays.

 

[LegendaryGT]

USC @ Sun Devils pls.

USC vs ASU
Week 4 of 2015

USC - Offense
Yardage: 479 ypg
Performance: 22 (higher is better)
Scoring: 42
Efficiency: 11 (lower is better)
Performance: 24 (higher is better)
USC - Defense
Yardage: 405 ypg allowed
Performance: 10 (higher is better)
Scoring: 23
Efficiency: 17 (higher is better)
Performance: 29 (higher is better)

ASU - Offense
Yardage: 412 ypg
Performance: -4 (higher is better)
Scoring: 36
Efficiency: 11 (lower is better)
Performance: 24 (higher is better)
ASU - Defense
Yardage: 422 ypg allowed
Performance: 2 (higher is better)
Scoring: 29
Efficiency: 14 (higher is better)
Performance: 5 (higher is better)

Prediction:

Pure statistical: USC 41 (513 yards), ASU 23 (363 yards)
USC best case: USC 46 (601 yards), ASU 18 (346 yards)
ASU best case: USC 36 (446 yards), ASU 28 (391 yards)

USC talent: 11961
ASU talent: 6858
USC talent of schedule: 6909
ASU talent of schedule: 5592

Suggested talent adjustment: 32% towards USC best case
Final: USC 43 (541 yards), ASU 21 (357 yards)

Sounds reasonable.

EDIT: Using only 2015 data:

USC vs ASU
Week 4 of 2015

USC - Offense
Yardage: 533 ypg
Performance: 53 (higher is better)
Scoring: 48
Efficiency: 11 (lower is better)
Performance: 26 (higher is better)
USC - Defense
Yardage: 401 ypg allowed
Performance: 3 (higher is better)
Scoring: 18
Efficiency: 21 (higher is better)
Performance: 69 (higher is better)

ASU - Offense
Yardage: 435 ypg
Performance: 18 (higher is better)
Scoring: 28
Efficiency: 15 (lower is better)
Performance: 6 (higher is better)
ASU - Defense
Yardage: 350 ypg allowed
Performance: 32 (higher is better)
Scoring: 23
Efficiency: 15 (higher is better)
Performance: 53 (higher is better)

Prediction:

Pure statistical: USC 29 (471 yards), ASU 18 (470 yards)
USC best case: USC 46 (655 yards), ASU 13 (427 yards)
ASU best case: USC 21 (291 yards), ASU 24 (510 yards)

USC talent: 11961
ASU talent: 6858
USC talent of schedule: 4590
ASU talent of schedule: 4157

Suggested talent adjustment: 27% towards USC best case
Final: USC 33 (522 yards), ASU 17 (458 yards)

 

[bonham]

why do you say that?

They amassed 17 against Tulane through 3 quarters and can't throw the ball downfield.

Because history suggests that our defense is the perfect cure for ailing passing offenses.

 

[GT1992]

Sounds reasonable.

So we should be betting on Duke and ASU?

 

[LegendaryGT]

#1: ('BC', [122, 2, 0, 4083])
#2: ('NW', [121, 3, 0, 4354])
#3: ('OSU', [121, 3, 0, 3797])
#4: ('FSU', [120, 4, 0, 4567])
#5: ('ALA', [117, 7, 0, 22506])
#6: ('WVU', [117, 7, 0, 8694])
#7: ('MISS', [116, 8, 0, 6448])
#8: ('CLEM', [114, 10, 0, 2644])
#9: ('UGA', [113, 11, 0, 19207])
#10: ('STAN', [112, 12, 0, 6188])
#11: ('ND', [111, 13, 0, 2997])
#12: ('GT', [109, 15, 0, 4758])
#13: ('USC', [109, 15, 0, 4280])
#14: ('LSU', [108, 16, 0, 4619])
#15: ('NCST', [108, 16, 0, 3270])
#16: ('MIZZ', [107, 17, 0, 713])
#17: ('TENN', [105, 19, 0, 3048])
#18: ('OKLA', [105, 19, 0, 2864])
#19: ('FLA', [103, 21, 0, 1988])
#20: ('WASH', [102, 22, 0, 2612])
#21: ('SYR', [100, 24, 0, 2459])
#22: ('MSST', [100, 24, 0, 2374])
#23: ('MICH', [99, 25, 0, 1605])
#24: ('UNC', [98, 26, 0, 2436])
#25: ('WIS', [98, 26, 0, 1972])

Duke checks in at #28.

I'll be updating the OP in a second. I went back and excluded the rest of last season's data and it didn't crash, so we'll be 100% in this season's data from here on in.

[NOTE on BC. The predictor loves defense.]

 

[Akinji07]

Thanks TIA.

 

[GruffyMcGuiness]

Can you do UCLA - 'Zona?

 

[johncu]

I think CPJ is going to bring out the "3 yards and a cloud of dust" (splattering of mud?) offense for this game. I say that because we have limited experience at AB and WR, but the OL should be able to push the Duke DL enough to move the chains with the dive all day. The number of possessions could be really low for this game, so Duke getting 2-3 scores might be a real possibility. The weather will make it hard for either team to break big plays.

Exactly what I'm expecting. I'd look for a heavy dose of B-back runs, many of them called plays. I think it will end up looking a lot like the UGA game last year, although hopefully our defense will be a little more stout than it was in that game.

 

[LegendaryGT]

Can you do UCLA - 'Zona?

UCLA vs ARIZ
Week 4 of 2015

UCLA - Offense
Yardage: 452 ypg
Performance: 10 (higher is better)
Scoring: 31
Efficiency: 14 (lower is better)
Performance: -5 (higher is better)
UCLA - Defense
Yardage: 328 ypg allowed
Performance: 21 (higher is better)
Scoring: 14
Efficiency: 23 (higher is better)
Performance: 63 (higher is better)

ARIZ - Offense
Yardage: 606 ypg
Performance: 30 (higher is better)
Scoring: 54
Efficiency: 11 (lower is better)
Performance: 3 (higher is better)
ARIZ - Defense
Yardage: 386 ypg allowed
Performance: -13 (higher is better)
Scoring: 21
Efficiency: 17 (higher is better)
Performance: 0 (higher is better)

Prediction:

Pure statistical: UCLA 24 (485 yards), ARIZ 19 (478 yards)
UCLA best case: UCLA 31 (552 yards), ARIZ 9 (358 yards)
ARIZ best case: UCLA 20 (420 yards), ARIZ 25 (656 yards)

UCLA talent: 9043
ARIZ talent: 5876
UCLA talent of schedule: 4303
ARIZ talent of schedule: 783

Suggested talent adjustment: 33% towards UCLA best case
Final: UCLA 26 (507 yards), ARIZ 15 (438 yards)

I sort of expect it to be closer than this, if only because Arizona has a penchant for late game comebacks under RichRod. As you can see, though, Arizona ain't played nobody, and the predictor don't like that.

 

[Deke]

That's a lot of yards for a combined 41 points.

 

[LegendaryGT]

That's a lot of yards for a combined 41 points.

Yeah, it's predicting a pretty disastrous game from the Arizona offense / ST. They've been scoring fairly efficiently so far, but they haven't been outperforming anyone on their graph against common competition. That suggests the efficiency value of UCLA's defense is likely to prevail, which is very bad news for scoring.

Meanwhile, UCLA is getting a ton of credit for how they defended UVA. The UVA - GT - ND triangle is squarely in their graph and it makes them look really good to the algorithm. Their defensive efficiency is great, but it is also outperforming by a huge margin, even worse news for Arizona.

If it knows what's going to happen, you can expect some goofball stuff to take a lot of points off the board for Arizona, despite them moving the ball well. Think UCLA vs Arizona State from 2014.

Although I think we've seen repeatedly that it doesn't know what's going to happen.

 

[Deke]

I'll take an order of stalled drives with a side of redzone turnovers.

 

[DaltonJacket]

Can you do Utah vs Oregon?

Thanks

 

[05GTCorndog]

I lost most of my life savings betting the predictor last week. I've since sold a kidney and put another mortgage on my house. Quitters never win. Lets double down this week!

 

[LegendaryGT]

Can you do Utah vs Oregon?

Thanks

UTAH vs ORE
Week 4 of 2015

UTAH - Offense
Yardage: 362 ypg
Performance: 16 (higher is better)
Scoring: 31
Efficiency: 11 (lower is better)
Performance: 12 (higher is better)
UTAH - Defense
Yardage: 371 ypg allowed
Performance: -10 (higher is better)
Scoring: 18
Efficiency: 20 (higher is better)
Performance: 49 (higher is better)

ORE - Offense
Yardage: 560 ypg
Performance: 27 (higher is better)
Scoring: 50
Efficiency: 11 (lower is better)
Performance: 36 (higher is better)
ORE - Defense
Yardage: 471 ypg allowed
Performance: -5 (higher is better)
Scoring: 33
Efficiency: 13 (higher is better)
Performance: -1 (higher is better)

Prediction:

Pure statistical: UTAH 45 (480 yards), ORE 36 (561 yards)
UTAH best case: UTAH 66 (580 yards), ORE 24 (457 yards)
ORE best case: UTAH 29 (362 yards), ORE 48 (737 yards)

UTAH talent: 5955
ORE talent: 9293
UTAH talent of schedule: 4668
ORE talent of schedule: 2583

Suggested talent adjustment: 6% towards ORE best case
Final: UTAH 44 (473 yards), ORE 36 (572 yards)

It hates both defenses but hates Oregon's defense more. Utah has the statistical epitome of a bend but don't break, and the predictor is counting on that for the close win.