Unofficial Playoff Committee SOS Ranks
True Playoff Committee SOS Rankings
Games played through week 13, 2015
DISCLOSURE: The Playoff Committee noted, last year, that they use their SOS to rank teams. It is not any of the public SOS ranks fans and writers frequently reference. The committee, as it turns out, states they do not use an SOS ranking. They just call winning percentage averaging an SOS ranking.
Noted Selection Committee expert, Stewart Mandel tweeted out the following (see below) about the committee’s ‘SOS rankings’. This is basically an in-season version of the commonly referred to NCAA pre-season strength of schedule rankings. The committee takes it a step further and considers the winning percentage of the opponents of a team’s opponents on their schedule. An example of this, for Auburn. is in the little chart to the right.
Committee doesn’t use an SOS ranking. It looks at opponents’ record and opponents’ opponents record.
— Stewart Mandel (@slmandel) November 12, 2014
Stewart further advised me that for FCS teams, only the loses count against the composite opponent record. So if LSU played McNeese State (8-0), only the losses are added to the composite record of opponents for LSU (more on this later). So if you end up beating an FCS team that goes 11-1, you are only dinged for the one loss by that FCS opponent.
I have not been able to determine, and thus assume it counts as nothing, the strength of home versus the road. In the Big 12, the PAC 12 and soon to be in the Big Ten and ACC, some teams have five home conference games, while the other half have five road conference games. That is a significant difference in schedule strength.
To create the ‘SOS’ ranking,we averaged the records of the opponents into the records of the opponents opponents to get the composite. All three are broken down and ranked below into opponents record, opponents opponents record and the average of the two records.
Thru Week 13 Selection Committee Unofficial SOS Ranks
HIT THE RESET BUTTON: I am annoyed the CFB Playoffs ran a commercial this weekend telling you to “Forget the Matrix”, but I take it that we are on the right trail with our SOS ranks, modeling and stance that there is no true ‘eye-test’. I like models that can be used to find the best team, or project winners of football games. Never perfect, they are nevertheless a good starting point to determining team quality. There is only one national SOS ranking system that could be used for this ‘predictive’ value. Most SOS ranks, like this current one used by the Selection Committee, if you use them to pick bowl games and November outcomes, are around 50% correct. The Playoff Selection Committee says they use this formula for the SOS because it can be easily quantified and it is understandable. This week we saw Clemson drop 29 spots playing Wake Forest and Alabama RISE 5 ranking spots for playing Charleston Southern. We are certain this dead on and if told otherwise, it means the committee changed their formula without any transparency.
It is broke. Fix it. I am officially offering my help, with the assistance of other national CFB analytics minds, to come up with a new SOS formula to better separate like resume teams for the Selection Committee.
Like it or not, this is exactly how the formula was spelled out to me.
[click on image to enlarge]