Warrior General Games

Calculated Elo Ratings for the Oceania Pro League after 3rd round playoff wk19 games (and final rating for those who are out of the playoffs)...

* Canberra Strike Alpha 144.0
New England Burning Pride Alpha 131.2
* Brisbane Gods Zeta 126.0
Tongan Crimson Auha Zeta 99.2


* - Conference champions

-----------------------------------------------------------

Predicted Win Probabilities for upcoming wk20, Oceania Pro championship game based on Elo data...

Canberra Strike Alpha 0.526
Brisbane Gods Zeta 0.474

-----------------------------------------------------------

Upcoming Playoff Game of the Week (top combined Elo ratings):
championship game - Canberra Strike (17-2-0) at Brisbane Gods (16-3-0)

Upcoming Playoff Game of the Week (closest Elo rating difference):
championship game - Canberra Strike (17-2-0) at Brisbane Gods (16-3-0)

-----------------------------------------------------------

caveats: This data is based on a data set of 19 points/team (games). Larger data sets are recommended for reliable trend/regression analysis.

Additionally, the different conferences were combined assuming that both conferences had identical overall conference strength.

notes: Elo "K value"=24, starting Elo rating of zero for all teams starting at season's start. "Actual results" - Win=1, Tie=0.5, Loss=0

Additionally, factors such as home field advantage, score margin, and team chemistry changes are ignored.

yay were finally the underdog

Thanks for a whole season of this

Yeah, major thanks are in order. You definitely went above and beyond in doing this all season.

sweet, we're not the game of the week anymore

yeah man... i have no idea how you figure these out but im grateful it is always something i check each week of the season.

Originally posted by Vilita
http://en.wikipedia.org/wiki/World_Football_Elo_Ratings

Mine is pretty close, but I have used a constant K factor even into playoffs. I may up the K factor for playoff teams next season.

I have used the original Elo formula, so I haven't incorporated a G factor (with N = touchdown difference) for score difference thus far. It's not in Dr. Elo's original formulae (built for chess rankings, which don't have point spreads ), but I will include that next season to improve the ratings.

It won't take too much more to add it, once I build up the spreadsheet.