Loading...
  OR  Zero-K Name:    Password:   

has elo been changed?

16 posts, 540 views
Post comment
Filter:    Player:  
sort
It used to be quite hard to go up in elo, but now I'm seeing everyone go up at least 200 points over the past few months. Has elo been changed? Do big team lobpots give a higher elo bonus now?

cheers
+0 / -0
4 years ago
elo is no more, long live WHR
+0 / -0

4 years ago
what's whr?
+0 / -0


4 years ago
It's been WHR (Whole History Rating) for a long time.

How it works is complicated. CHrankAdminDeinFreund should probably put an FAQ explaining how it works somewhere so that he doesn't keep having to answer questions about how it works. :D
+0 / -0

4 years ago
https://zero-k.info/Charts/Ratings

Documentation has always been there.
+1 / -0


4 years ago
If you're willing to read a PDF. :P
+0 / -0

4 years ago
Why do you want to know how WHR works internally? I feel that would only throw people off.
+0 / -0
4 years ago
I think the OP's question has less to do with the internal workings of WHR and more with why the ratings seem so strange and unintuitive. My own, for example took a considerable nosedive and I can't explain why. The idea is that we look to it to represent skill level, but it does nothing of the sort.

Key word: UNINTUITIVE
+0 / -0
4 years ago
Statistics is un-intuitive many times. Also, I find it rather ok, meaning if I loose games -> rating goes down, if I win games -> rating goes up. The actual value by which you go up or down in teams depends on the abilitiy of team mates, and that is hard to "get intuitively".
+0 / -0
4 years ago
Machines are good with numbers (well, not really, but we're not talking about rounding errors here). Humans prefer to go on a feeling instead, but we trust in some numbers. If I see someone with a higher WHR than my own, I expect that person to play better than I do. If it's lower, I expect them to play worse. That's just how the human brain works: trying to keep things simple. If that weren't true, there'd be very few questions about the way the system works. Whether it's hard or easy to collect statistics to do this accurately is besides the point. The point is, whenever something happens to that number that a player doesn't expect to happen, that person is going to get confused and have questions as to why. Technical explanations don't cover the "why" because they are technical, and that "why" is... well... emotional.
+0 / -0


4 years ago
quote:
The idea is that we look to it to represent skill level, but it does nothing of the sort.

It does represent skill level. But it considers the whole history.

For a contrived example to illustrate why this sometimes gives weird results:
Imagine a player who is amazingly good at jump micro, but crap at everything else.
He attains a high ranking solely from his awesome jump micro.
Then a widget gets added to the game which automates jump micro.
The next day you play against Mr. Jump Micro and beat him soundly, gaining a whole bunch of ELO.
The next 10 people who play vs My. Jump also completely thrash him.
The game will start lowering your ELO with cause easily visible to you because it realizes that actually, beating Mr. Jump isn't as impressive as it thought and something happened to Mr. Jump to make him much less prone to winning games.
+0 / -0
4 years ago
Would that make him less skilled at jump micro?

His wins and losses are scored, not his skill. You may have beaten him many times with the new widget, but if something happens and it breaks or gets removed after an update, you with your now-high WHR score would be expected to defeat him with his low score, but that's probably not going to happen. And yes, the scores will eventually get readjusted, but not before a good number of people get surprised at how good he's suddenly become. And yet... his skill level has not changed throughout the whole ordeal.

Maybe my definition of skill is just not the same everyone else's...
+0 / -0
So... if the people you beat in the past go up in elo, you go up in elo too? Wouldn't that just mean the elo goes up and up for semi-active players? If so, "top 20%" is more like top 80% if you're an active player (play more than twice a week or so)
+0 / -0

4 years ago
"If you're willing to read a PDF"

so you want other people to write even MORE text (that you need to read anyway) just because you don`t want it as a pdf?

+0 / -0


4 years ago
There does seem to be something up.

I've noticed a lot more silver/blue players. I've also noticed a lot more silver/blue players that don't seem like silver/blue and seemed more like jan '19 bronze. I've seen blue players spam hammers and nothing else, and a few other "wait, I thought you knew how to play this game" moments.
+0 / -0

4 years ago
Here are the news posts about all ladder changes since the introduction of the new colors:
http://zero-k.info/Forum/Thread/24634
http://zero-k.info/Forum/Thread/27775
http://zero-k.info/Forum/Thread/28940

The most notable change has been about the activity cutoff, which now simply includes all players that played at least one game in the last 30 days. Previously it was based on rating uncertainty. Due to this change, newbies now have a much easier time being ranked (but their rating starts at 1100 instead of 1500, so they have to fight all the way up the ladder). This means the ladders fill up at the bottom with newbies and make it easier to reach the percentiles needed for the higher ranks.
+0 / -0