Frequently Asked Questions

Does OSCAR take into account strength of schedule?

Yes - all OSCAR?s algorithms implicitly take into account the fact that some teams play harder schedules than others. Some computer ranking systems tend to overrate teams with very good records against very weak schedules. OSCAR doesn't.

What about Home Field Advantage?

Home field advantage is used as an input parameter for OSCAR's algorithms. Conventional wisdom dictates that the advantage of playing on your own field is worth something like 3.0 to 3.5 points - but recent careful research by John Kros et al of East Carolina University suggests that the true figure is closer to 2.3 points. OSCAR's algorithms use this latter value - with the exception of the "PC" algorithm, which ignores home field advantage entirely.

Some computer ranking systems use home field advantage as an output rather than an input parameter - that is, they attempt to find its value using game data from the current season. In our opinion this methodology is questionable because there simply isn't enough statistical information in one season's worth of results to deduce anything meaningful.

Another interesting question to consider is - do some teams possess a larger home field advantage than others? While the answer may well be yes, OSCAR does not take into account such considerations because it is designed not to give any kind of a-priori bias to any specific team.

Are recent games weighted more heavily?

Yes - OSCAR's algorithms incorporate a "half-life" parameter. For example, in the "Default" Algorithm, this parameter is set to 8 games, meaning that OSCAR considers a result from 8 games ago to be only half as important as the result from a game played yesterday. As a consequence, OSCAR tends to rank teams with early-season losses higher than those with late-season losses, all other things being equal.

What do the team ratings actually mean?

The OSCAR rating of a team is measured in points. If the rating is positive, it represents the number of points by which the team would be expected to beat an average division I-A team. If it's negative, that's the number of points it would be expected to lose by.

What's with all the colors?

OSCAR's color scheme is very simple. Strong teams, conferences and game results are displayed in increasingly deep shades of green. Weak teams, conferences and game results are displayed in increasingly deep shades of red.

Note that OSCAR will generally consider a narrow loss to a very strong team to be a more impressive result than a narrow win against a very weak team, and the colors displayed in the Matchup Tab tables reflect this fact.

When exactly does one season end and the next season begin?

The OSCAR college football season runs from July 1 to June 30. As you scroll through from June to July, you'll notice all the team records reset to 0-0.

What's all this "Connected Components" stuff?

Two teams are said to be "connected" if the schedules for the current season connect them. For instance, if Florida has played Tennessee and Tennessee has played UCLA and UCLA has played Oregon, all in the current season, then Florida has become "connected" to Oregon.

Nowdays, the teams as a whole usually become connected to one another by the end of September. In years gone by, the teams would often take much longer to become connected - if indeed they became connected by the end of the season at all.

It's clear that the better connected the teams are, the more reliable OSCAR's rankings will be. OSCAR has no way to compare teams in different connected components without taking into account results from the previous season.

Connectedness is one of the pieces of information that can be accessed by tapping the Info Button.

Explain "Prediction Accuracy".

Oscar assesses its the accuracy of its predictions by comparing them with the results of actual games played during the 30 days following the Rankings Date - assuming of course that games were indeed played in this period.

Two measures of prediction accuracy are available as part of the information accessible by tapping the Info Button - the 30-Day Prediction Accuracy gives the percentage of games whose outcome OSCAR predicted correctly, while the 30-Day Prediction Error Standard Deviation is a measure of how far out, on average, OSCAR's spread predictions were.

Why do the rankings sometimes take a long time to calculate?

OSCAR's algorithms are all iterative, meaning that the same block of computer code is executed over and over again, the rankings becoming more and more accurate with each pass. Usually the rankings converge (settle down to constant values) within a few seconds - but in situations where the teams are very poorly connected, convergence may take longer. If you get bored during this time, tap the Abort button - and OSCAR will use the rankings from an intermediate stage in its calculations.

But weren't Oklahoma national champions in 2000?

Absolutely - Oklahoma were national champions in 2000. But OSCAR thinks that by the end of the season, Miami might have been just a shade stronger. OSCAR, like any computer ranking system, sometimes throws out interesting and controversial opinions.

Who wrote the rankings algorithms?

The ranking algorithms were designed by Glenn Watson, who holds a Ph.D. in mathematical physics from the University of Florida.

Despite this fact, OSCAR holds no intrinsic bias toward the Gators!