We have been asked by multiple readers to explain a little more in depth how we compile our PiRate Retrodictive ratings and what exactly this rating means.
By the way, if you have comments you would like to make, please visit our sister site and use the “contact us” link, at:
There are basically two types of ratings in sports like football and basketball. The first type is Predictive, and this type is used by sports bettors. As the name implies, these ratings are used to attempt to predict the outcomes of the next week of games. If State U has a rating of 106.5, and Tech has a rating of 102.3, then on a neutral field, State would be expected to beat Tech by 4.2 points. Of course, home field advantage could alter this predictive spread by additional points. The regular PiRate, Mean, and Bias ratings are predictive in nature. A team that started the season 0-4 and then finished 8-4 might be favored in a bowl over a team that went 11-1, because what the 8-4 team did in its first four games is not as important as what they did in their final games.
Retrodictive ratings or rankings look backward and rate the teams on what they have done to date. What the team did in its first game is just as important as what they did in their most recent game. It is an attempt to rank the teams by their performance on their whole body of work. That 8-4 team would most likely be rated well behind the 11-1 team on the body of work to date.
Our Retrodictive Rankings take considerably more time to compile than our regular PiRate Ratings. First, there are three components to the ranking. The first is rather obvious–wins and losses. The second is also obvious–strength of schedule. The third is well known to many, but you may not realize that in addition to strength of schedule, the strengths of schedule of all the opponents played is vitally important.
As an example, let’s say that Iowa opens the season with a win over Boston College. As the season goes on, Iowa’s rating is influenced by what Boston College does and what all the teams Boston College plays do. So, if Iowa has a bye after that first game, and Boston College proceeds to beat Florida State the following week, Iowa’s rating is going to go up almost as much as Boston College’s rating goes up.
Point differential can be included in retrodictive-style ratings, and we here at the PiRate Ratings have a rather unique way of including the type of score in a team’s wins and losses. It is not an exact science, but we like our version and believe it has merit over standard point differential. Here is a rundown on this process.
What type of win or loss was this game for the team in question? We rate wins and losses as:
A. Blowout
B. Decisive
C. Good
D. Fortunate
E. Tossup
A blowout win does not have to be by 40 points. If a team wins 28-0 and holds their opponent to 150 total yards, this rates as a blowout. If a team led 42-0 at halftime and then won 49-24, it still counts as a blowout. If a team led 24-20 midway through the third quarter and then won 45-20, this will usually not count as a blowout.
A decisive win is one in which the winner would have won this game close to 100% of the time but did not win in such a way that allowed the team to rest its starters for the final 20-25 minutes of the game. A 35-10 win with the winner leading 21-10 and scoring 14 points in the fourth quarter would be one example of a decisive but not blowout win.
A good win is one in which the winner would have probably won 7 to 8 times out of 10 with the same stats generated. A 10-17 point win is often a good win, unless the winner was outgained by more than 100 yards.
A fortunate win is one in which the winner won by 4-9 points but the outcome was always in doubt. In some cases, the winner had stats that looked more like they should have been the loser.
A tossup win is the same as a fortunate win but with the final spread being 1-3 points. All overtime wins count as tossup wins.
The better the win, the more points the winner receives and the loser gives up. If Georgia beats North Carolina 42-14, they receive more points than if they beat the Tar Heels 17-14.
For strength of schedule, we rate each of the 128 FBS teams from 10 to -10 in tenths so that the current top team is ranked as a 10, and the number 128 team is ranked as -10. FCS teams receive a rating of 5.0 to -15.0.
Let’s say that through 4 games, Oklahoma has played teams that currently rate in our strength of schedule as 9.2, 8.6, 5.4, and 1.3 points. The Sooners’ average schedule strength is 6.1. We are not done yet. Now, we have to adjust this number based on the strength of Oklahoma’s opponents’ schedule other than playing the Sooners. So, if that 9.2 strength team played a schedule to date that averages 5.8, and the other three teams played schedules with 4.3, 7.1, and -2.9, Oklahoma’s opponents’ strength of schedules average to 3.6.
Obviously, Oklahoma’s own strength of schedule of 6.1 is worth more than their opponents’ strength of schedule of 3.6, but their schedule strength portion of the equation is lower than 6.1 because of the weaker scheduling of their opponents. In this case, the algorithm we use through 4 games played will lower the overall strength to about 5.3.
We combine the score for the team’s wins and losses with the schedule strength, and it leaves us with a final number between 10 and -10. In actuality, no team approaches these outlier numbers. For instance, this week, Alabama comes in at 8.27 to be number one, while Texas State and Buffalo bring up the rear at -7.59. Because these are not predictive, you cannot use these numbers to predict the outcome of a game should Alabama play Texas State or Buffalo.
Of course, home field advantage is factored into the outcomes of the games. If Georgia beats Clemson between the hedges by a score of 17-14, it does not count as much as if they had defeated the Tigers 17-14 at Howard’s Rock.
We hope that clarifies and not confuses you further. We are better with numbers than with letters.
The Pi-Rate Ratings 