i think 10chars
i think 10chars
We have increased the rate at which someone decays on the leaderboard. We were hoping to observe the change for a few days before making any sort of announcement, but you guys are too clever for that.
Justin has also changed the decay to update every hour, rather than once a day, so you should see movement happen a bit more often.
Let us know what you think!
We have increased the rate at which someone decays on the leaderboard. We were hoping to observe the change for a few days before making any sort of announcement, but you guys are too clever for that.
Justin has also changed the decay to update every hour, rather than one a day, so you should see movement happen a bit more often.
Let us know what you think!
Wait… every hour? So like if I don’t play for an hour, I’ll start decaying? Or you mean like if I don’t play for a week (or whatever the real minimum interval for decay to start is), the exact hour that 1 week since I last played hits, the decay will start?
Yes, after two hours it will start to decay but the decay is so low that you shouldn’t see any changes in such a short amount of time. It also ramps up the longer it’s been since the last time you played.
These changes to decay aren’t exactly a ‘new system’ – we’ve just increased the speed and severity that someone loses ranking. Like I mentioned in a previous post, we’re looking to overhaul quite a few things in the New Year, and this system is on our list.
For the time being, if players are logging in to maintain their ranking because they’re decaying at a faster rate – then in a sense the leaderboard decay is working. Players who have afk’d for months will begin to drop down in the list. With the changes to MMR and roster randomization players will be far less likely to play in stacked games when they do come back to maintain their position.
I don’t understand their philosophy on decay disappearing after you play a game. It makes no sense.
We’re going to be changing this, but you’ll have to wait for the details. :/
In perfect theory, if matchmaking put you against EQUAL opponents , would everyone have/be very close to a 50/50 win ratio in the long run ?
No, even assuming a very large population with abundant skill range coverage, a portion of the difference comes from a player raising/lowering their rating from the default starting position. It takes a while for your MMR to settle.
Mind that switching it to an hourly basis isn’t making you decay faster, (we did increase decay) it’s improving the visibility/perception of decay actively working.
If you have a good matchmaking and plenty of playerbase yes, it’s close to 50/50, unless you get really bad or really good suddenly, or you are in the extreme. But 90% of players should have close 50/50 in the long run.
Baldric is exactly right. Another component that can lead to a non 50/50 win ratio is when matchmaking has to give you a lower quality match because of off-peak concurrency.
Another component that can lead to a non 50/50 win ratio is when matchmaking has to give you a lower quality match because of off-peak concurrency.
What is “off-peak concurrency”?
Any concurrency outside the hours of the day when concurrency is at it’s highest.
The “off-peak” part I easily understand.
Concurrency is what I was missing.
Do you mean simultaneous queuing?
Ah, sorry. Yes, concurrency is the number of players playing at the same time.
So the moral of the story is: "if you want your leaderboard rank to be as accurate as possible, only play at peak times, say, 4pm to 11pm CST in North America.
No, it’s more like “if you want your win/losses to stay around 50/50 then you should play during peak hours.”
Glicko2 can handle adjusting your MMR properly when you’re matched against players of different skill levels.