MakeAPlay
DA
Interesting discussion @Simisoccerfan and @MakeAPlay. Since you both understand this stuff, I'll mention that my original shortcut was just to use the opponents' average RPI for element 2 (and opponents' opponents' RPI for element 3). My refinement was to go to the trouble to actually back out the result from the game played in computing the opponent's RPI for that game, per the following formulas:
Opponent loses game: (OW + 1/2OT)/(OW + (OL - 1) + OT)
Tie: (OW + 1/2(OT - 1))/(OW + OL + (OT - 1))
Opponent wins game: ((OW - 1) + 1/2OT)/((OW - 1) + OL + OT)
Considering the passage @MakeAPlay quoted above, it makes sense why this refinement ended up having so little impact.
Yup. There since the net results of all games played are zero (somebody wins and somebody loses or it is a draw) the spread between strength of schedule is very small thus reducing it's impact and increasing the impact of a team's record. This is one of the reasons why RPI isn't a clear indicator of a teams TRUE strength. The eye test is usually best but extremely difficult to watch that volume of games and impossible for you to watch enough GDA games to have a true idea. With colleges it is pretty easy to watch games and if you watch the top 20 or so teams enough you have an idea of who the best teams are. Because of this as soon as I saw the brackets come out for the College Cup it was pretty easy to pick out 3 of the 4 participants as they were MUCH better if you just considered the eye test (3 of the 4 teams in the College Cup were true possession teams by the way).
Keep up the good work it has piqued my interest and I don't have a club player any longer.