Stanford vs. USC: Odds, betting lines, and computer predictions

Kirby Lee-USA TODAY Sports

LINES AND ODDS

The sports books opened Stanford vs. USC with the Trojans favored by 4 points and has since moved to USC by 51⁄2. The change in the line reflects that early betters have laid more of their money down on USC to cover the spread.

Vegas lines and odds aren't meant to be predictive. They are made to entice betting in equal portions on each side of the spread. However, combining the line and the over/under usually ends up close to some computer models. After all, Vegas is very good at this and it is a lot of fun.

Combining the line and the over/under, the Vegas combo has Stanford losing 30-25.

2017 Stanford Vegas record against the spread: 1-0

PREDICTIVE RANKINGS

Some computer rankings are designed to provide a built-in predictive element by comparing the ratings of two teams.

Some of the best ratings out there come from Football Outsiders. The most famous is Jeff Sagarin's for USA Today, previously used in the BCS computer rankings. I've included a couple others from around the web as well.

FOOTBALL OUTSIDERS F/+

F/+ is a combination of Bill Connelly's S&P+ and Brian Fremeau's FEI ratings. The ratings are usually pretty solid in factoring all the many variables involved in ranking college football teams that have a relatively low level of common opponents.

F/+ does not predict a final score, but does give an expected margin. F/+ likes USC by 3 points.

The F/+ ratings list Stanford # 8 and USC # 6.

Football Outsiders site only lists the ratings in order, but Bill Connelly posts predictions on SB Nation.

Bill's S&P+ picks predicts The Trojans by 4 points (winning 30.9-26.8) and gives USC a win probability of 59.4%. Connelly picks Stanford at +5 1⁄2 to cover the Vegas spread.

SAGARIN

Stanford is ranked #14 with a rating of 86.85.

Southern Cal is ranked #13 with a rating of 87.14.

Sagarin’s formula currently values home field advantage as worth 2.41 points. So subtracting the difference between the ratings and then adding 2.41 in favor of the home team, USC, Sagarin has USC favored by 2 1⁄2 points (2.7).

BILLINGSLEY

Billingsley's ranking was also previously used by the BCS. With the BCS restriction to remove margin of victory no longer a consideration, Billingsley has created a version of his formula that accounts for margin of victory.

Stanford is ranked #13 with a rating of 101.218.

USC is ranked #3 with a rating of 102.734.

The difference is 1 1⁄2 (rounded down from 1.516) points. Billingsley does not provide a home-field advantage factor, but lets apply the standard 3 points to the Trojans and say Billingsley has USC by 4 1⁄2 points.

DONCHESS INFERENCE RATING

Donchess boils down the ratings directly into digestible scores and probabilities, no math required.

Donchess predicts a USC win 31-24 and gives Southern Cal a win probability of 72.3%.

COMPOSITE RATING

There are a lot of computer ratings that aren’t packaged with a predictive ability. There are also a ton of computer ratings, period. Kenneth Massey, whose ratings were also part of the old BCS computer formula, hosts on his site a composite ranking of 106 computer ratings across the internet.

In that composite, Stanford ranks #13 and USC ranks #5.


This week everyone favors USC to defeat Stanford in the LA Coliseum.

Comments

So the computers like USC but the human experts are picking Stanford.. interesting. Weather at game time roughly 75 Degrees compare to WMU game at 110 Degrees. I’m curious who Lee Corso’s picking in this game.. He’s fun to watch and pretty spot on with his USC predictions.

Corso likes the USC trojan helmet. Even katy perry picks better than him <img src="//fonts.voxmedia.com/emoji/unicode/1f61b.png" alt=":stuck_out_tongue:" class="emoji">

Yeah, it's tough to put a tree on your head

Corso actually injured himself once when he picked USC to beat Stanford back in 2013. He took a Trojan sword and had a mock duel with another journalist and ended up getting cut on the lower lip. This was the week after Stanford had dominated Oregon, so Corso went contrarian and was (unfortunately for Cardinal fans) correct. That defeat in the Coliseum (the only one of the last five USC-Stanford games played there) ended Stanford’s BCS title hopes. https://www.youtube.com/watch?v=Md1MHaDY7L0

worth mentioning . . .

The computer prediction models particularly for CFB have to rely an awful lot on data from a season ago for the first few weeks until teams have played enough games in the current season to provide sufficient current data. This is the main reason the old BCS system which relied partially on computers did not release there first poll until October usually.

Also worth noting is that ESPN’s FPI relies much less on data from last season and their model is giving Stanford about 60% chance to win.

One computer not mentioned: ESPN FPI

Stanford given a 60 percent chance to win

Lee Corso picked USC on GameDay show.. heard it on the radio this AM.

ESPN FPI is not a good predictor of actual games.. Why? It accounts for margin of Victory, yards per pass or run, data from past seasons, rankings, etc..

View All Comments
Back to top ↑