The Sparks Achievement Ratings are precisely what the name implies: a measure of what a team has achieved. They are NOT "power" ratings, and they should NOT be used to try to predict future games.
Because the Ratings measure a team's achievement, margin of victory is not taken into account. Though score may be important for determining a team's actual strength, a win should be a win when considering where a team deserves to be ranked. The object of football is to score more points than your opponent, not to score many more than your opponent.
Likewise, because the Ratings are Achievement Ratings, all games are weighted equally. Early games count just as much as later ones.
A team's rating is the average of its individual game ratings. Each game rating for a team is determined only by the winner of the game and the rating of the opponent, but the game rating changes in the event of a harmful win or helpful loss. If a strong team beats a weak team, the strong team should not be punished simply for playing the weak team, and likewise the weak team should gain no benefit for losing to a powerful opponent. Therefore, these games are effectively disregarded. Moreover, any other game that dramatically changes a team rating will receive less weight. For more information on how these processes work, e-mail me. These calculations are all 100% objective and mathematically determined.
Ratings are recomputed every week. Each time, a team is assigned an initial rating of zero. The game ratings are then computed assuming every opponent's strength to be zero (the initial rating), and a new team rating is produced for each team. These new team ratings are plugged in where the zeroes used to be, and this yields another new set of ratings, with each iteration becoming more and more accurate. I continue recomputing the ratings until the ratings are accurate to 4 decimal places. For more information about the actual formulae used, e-mail me.
No preseason ratings whatsoever are used. This means that the ratings may seem bizarre early in the season, before enough games have been played to properly ascertain where teams deserve to be ranked. 2-0 Ball St could very easily be ranked ahead of 2-0 Florida St. Thus, although I do produce ratings for every week of the college football season, these ratings are probably rather untrustworthy until October or so.
The final output that you see when you view the rankings lists four items for each team. First, and most importantly, is the team's Sparks Achievement Rating. Second is the team's win-loss record in 1-A games only (only 1-A games are included in the calculations). Third is the team's strength of schedule rating (average of its opponents' ratings), and fourth is where the team's schedule ranks among all teams.
E-Mail Me with Any Other Questions