Regardless of how the voting goes down, I’m very proud of what we were able to accomplish at Idaho. We were able to go 9-2 despite a rough qualification schedule that featured three matches against 2122 and two against 5499. Though the playoffs didn’t go as hoped, I am still proud of the fight we put up at the highest-level regional we competed at this year.
Hey Matt, if 900 makes it into the top 25 then we’ll come back to Rumble this year!
Also, if we make it into the top 25 this week then I’ll personally meme every post Andrew Schreiber makes for the next year and I’ll stop mentioning that the Q&A people made me buy a dictionary in every other post I make.
This is a really valid concern that I have of Elo, investigating this question more was already on my long list of interesting things to investigate. The best way I can think of to test for regional strength would be to look at the average Elo ratings change of every team in a region before and after championships. Regions that see the largest Elo boost coming out of champs probably were undervalued because of their region going in. I’ll probably look into this eventually.
The easier way to look for signs of regional variation is to compare how good the Elo predictions were at the championship event with the district championships, since districts are by far the most isolated regions and the championship event is by far the most diverse. If the championship predictions are appreciably worse than the district championship predictions, we might conclude that there is something (like varying regional strength) that is causing the predictions to be poorer.
The Brier score for quals matches at the championship event in 2016 was 0.172, and the Brier score for quals matches at week 7 district championship events in 2016 was 0.184. This certainly doesn’t absolve Elo completely of the concerns above, but it is sufficient proof for me that Elo tracks fairly well across regions.
Sure, I can post them on the final top25 thread. All I’m doing is copy-pasting values from here though, and anyone is free to do the same for these or other metrics they are interested in. Here they are for currently registered teams:
Thank you for all these positive discussions so far. We are very interested in looking at a mathematical system to weigh the votes for the 2018 season as this will be similar to other sports polls. No matter what we choose somebody will be not happy with it we do want a balance especially as there are more and more events… Another would be to break down by certain regions too.
Last week was rough with so many events and also results in less votes. This week are close to as many votes right now as we were last week so on a personal note I think there will be better distribution.