Sagarin's Predictor Rating and Gap Analysis

Sagarin Ratings

The Predictor is very interesting, IMHO, because it is supposed to be the best measurement for how a game between two teams will go down. It's not perfect, obviously, but it can spark some interesting conversation.

For example, using the Predictor only, we see the following:
1. Georgia - 95.10
2. Alabama - 95.02

That's a difference of 0.08

The #3 team:
3. Iowa - 87.46

That's a difference of 7.56 between Iowa and Alabama, or 94.5 times the difference between the top 2 teams of that method.

A hypothetical team that is as far behind Iowa as Iowa is behind Alabama/Georgia would be ranked #19, between Oklahoma State and Nebraska (on the Predictor Only method).

In the other metrics as well, Alabama/Georgia is always very close. In Golden_Mean (utilizes the actual SCORES of the games in a different way but is also completely SCORE BASED and thus should be a good match for the PURE POINTS in terms of predictive accuracy for upcoming games), Alabama holds a 2.77 point advantage over Georgia, which is the only time where the gap vaguely resembles the distance to #3 (Iowa), where the difference is 3.4 between UGA and Iowa.

In the Recent model (score-based and weights RECENT play more heavily than earlier games. Its effect will become
more pronounced the longer a season goes if a given team happens to have an upward or downward trend), Georgia holds a sliver of a lead over Alabama, 0.28 (but larger than the Predictor method), which we can compare to the #3 team Iowa and a difference of 3.52 (12.6 times the gap).


The only gap closer is the one between Oklahoma and Iowa when you average all the scores together (the basis for the primary ranking), which are 0.01 points apart. Alabama and Georgia, using this method, are 0.97 apart; compared to the gap between Georgia (#2) and Oklahoma (#3) of 6.08, 6.3 times the difference; a hypothetical team with the same difference from the #3 team would fall at #13, just below Iowa State and just above Clemson.


I like looking at things from a "gap analysis" standpoint because sometimes two things can be right next to each other in the standings, but when you break it down further you can see that the difference between the silver and bronze winner might be 5 times the gap between the gold and silver.

Case in point:
Alabama has played, to date, the #50 SOS per Sagarin, while Georgia has played the #33 SOS. The difference between the numerical score is 1.03.

If you go up by the same number of RANKS (17), you find Clemson at #16, but you also find that their numerical score is 2.75 higher than Georgia's. If you go up by 1.03, the closest is Kent State's SOS ranking of 27 at 71.10.


Or another way to look at it:
In 2020, Alabama led the SEC at 48.5 ppg; Florida was second at 39.8. The gap between Alabama and Florida was 8.7 ppg; a hypothetical team that scored 8.7 ppg less than Florida would place them at #7, between LSU and Missouri.

In 2021, Georgia leads the SEC in scoring defense at 4.6; Texas A&M is second at 12.6. The gap between the two is 8.0; the hypothetical team in this circumstance would fall at #9, between Florida and Tennessee.

I like this kind of data because it gives weight to statements like "there's a huge drop-off from #2 to #3." People love lists, but sometimes there's a big difference between two points that may still be adjacent once you narrow the field down to a simple ranking.