I personally didn't like the settings at all but I still like to see how everyone is doing :)

I checked some stats, the 4th place guy is 1-1 on the FFA's, the 5th place guy is 6-0 (was 5-0 before it updates itself). How does that make any sense? And the top two ranked players are 2-0. Is it an advantage to play less games in this season?

its all about who you beat

It doesn't rank players based on straight win count, it uses an ELO formula that looks at how highly rated the players you defeated are.

It has very little data right now, so some things get skewed. By the time the season finishes, it should make a lot more sense.

I know it ranks by how good your opponents are. But The Duke of Ben has 3 wins equal to Gruffalo's one. But 2 of the guys Gruffalo lost to are rated very high, so I guess it has something to do with that. I've just never seen ELO act so strangely, I've played around with it for the regular ladder and done my own simulations and rankings for random things even with very little data, and I've never seen a guy go 1-1 and be ranked above someone who is 5-0.

The only thing I can think of is something with how it is FFA instead of regular 1v1. But I think you're right, it should even out by the end of the season.

I think the biggest issue with what the numbers look like now is the ratio of wins to losses on finished games. A loss is a single loss, while a win counts as three wins, one for each other person in the game. In real terms, this makes a loss trivial to counter, especially when the loss is against a high rated player. A 1-1 record is actually a 3-1 record, and explains much better how people with losses can be ranked above people without. A loss hardly does any damage.

the elo system might have to be thought over to better applu to ffa agmes

Yeah, it doesn't make sense. Like Jupiter just won two games (is 2-1) and somehow is ranked 22nd? Mordred is 1-1 and ranked 10th.

Something seems odd to me.

I think the issue is a win/loss is counted 3 times basically. So if you play 3 top 20 players, it is counting that weight of a top 20 player 3 times, rather than just once in a 1v1. So if someone else wins against 3 people ranked in the 50s, it dings them 3 times for those people being ranked lower.

You would think beating 3 people ranked in the 50s would be easy for a top player (it would be on 1v1), but this isn't a straight up fight. They could all just gang up on you. I think an actual win needs to be a much bigger factor than losing to good players is.

Explain this one...

Bluemonkey (0-5) is ahead of HHH (1-4 with win over #1 on ladder)

Fizzer, did you change something about the ladder ratings for season X? There was a huge changeup in the ratings without very many games completing, and no games being completed for most of the people who changed rank.

Nope, I haven't touched it.

I think it's just because it's still early in the season, and Bayeselo produces weird results when there aren't many games. So even a game that didn't involve those people can change the rankings wildly.

The difference in this season from previous ones is that you get a rank after only 2 wins instead of the usual 7. This was probably a mistake, as these wild fluctuations wouldn't be as visible with a higher number. I'm still confident that it'll solve itself as the season winds down.

I think you're right about it fixing itself. I would also agree about 2 games being too few to get reasonable results. The only thing it really affects in the end would be the "highest rank achieved" stuff, but it affects people psychologically quite a bit. In 1v1 this is less of a concern than in FFA, since psychology can sometimes affect who does or does not get attacked, and why.

Highest rank is not shown in your profile anyway on Seasonal ladders. Just your end rank

not shown on your profile but it is on your ladder page

Correct. But nobody else will look at that. We had dozens of people that were #1 overall at some point without winning a game.

Oh, I mean I am #3 now. Early season rankings matter!