Parent DirectoryI've tried making bowl game count double, and found that didn't work well, mainly because you end up penalizing the teams that lose a bowl too much.
It is possible for a rating to suffer from a win or benefit from a loss. If Nebraska had struggled to a one-point win over Pacific, shouldn't Pacific be rewarded and Nebraska penalized?
One other thing ... Elo's chess system didn't take into account white vs black pieces ... I do consider home vs away. Big factor there ...
One thing I found out about working on my system for the past 15 years is to try it out on bizarre situations like I described above and see if it would give you results that seem reasonable. If not, then there's tweaking to do.
When I started I used 60 as the rating for an average team and limited the amount a team's rating could change in a given game to 5 points. After a couple of years I realized that that wasn't enough spread, so now it's up to somewhere around 550 being my average team ("somewhere" because I round so ratings can't go above 999 or below 100 so as to keep predicted pointspreads plausible) and 20 points the maximum per game. Works much better ...
Predictions are for entertainment purposes only.
For more information, click here.
Darryl Marsee / marseed@db.erau.edu