kenany
commented
9 years ago @0BuRner Hmm, I wasn't even sure if they were using Elo but actually looks like they might be. Several factors could account for the disparity:
- Their
Kvalue is lower, so teams are only gaining a few points with each win. I'm using the "usual" default ofK = 32 - They don't take into account the point differential of matches. It's kind of hard for me to confirm this since all they provide on a team's page is a little graph. My calculations take into account the difference in score between two teams, using the same algorithm that FiveThirtyEight uses for NFL ratings.
- The default Elo rating for a new team is less than mine. There are a lot of teams on that leaderboard that are all tied at 1,000 Elo, so maybe that's their default value? Mine is 1,500.
Of course, we might never know for certain why their values are different from mines. The biggest issue here is that they have absolutely zero documentation on their rating-calculation process. One benefit of what I'm doing here is that my calculations are completely open-source, so it is clear how I have arrived at the current standings. Frankly, something seems very off with their ratings if the best teams in the world only have a ~1,500 Elo rating (that's the default rating in my program and in chess, I think).
0BuRner
Hi, It's not a real issue but I'm wondering why is there so many difference between your ELO and this one : http://www.gosugamers.net/counterstrike/rankings#team ?