Toribash
Read the thread?

As above: "There's this fun thing called "placement" matches, where if you show at the beginning that you're a great player, you'll transcend a rank that for some it takes thousands of games to reach."

Every single time ELO is used they have some kind of calibration method, if you have played any of the hundreds of games that use ELO you will know this. Placement matches are very very common I can probably name a dozen games that use it (Valve is particularly fond of it, as are Blizzard).



The downside is that we have to change the ranking system. This is a considerably downside so in lieu of actual advantages there is no point in changing............
Originally Posted by GnilRettemHC View Post
Read the thread?

As above: "There's this fun thing called "placement" matches, where if you show at the beginning that you're a great player, you'll transcend a rank that for some it takes thousands of games to reach."

I have read the thread, I think you are confused about what the Elo system is and what calibrating the system itself means. The placement matches are not part of the Elo system per say they just decide what your base Elo is. Calibration would be changing how much Elo you win or lose per game based on more/different factors.


Originally Posted by GnilRettemHC
Every single time ELO is used they have some kind of calibration method, if you have played any of the hundreds of games that use ELO you will know this. Placement matches are very very common I can probably name a dozen games that use it (Valve is particularly fond of it, as are Blizzard).


The downside is that we have to change the ranking system. This is a considerably downside so in lieu of actual advantages there is no point in changing............

Ok, now that not only are placement matches not a method of calibration, but they have been shown to not always work as if you play badly in 3 matches you have to play hundreds of hours to get the rank you would otherwise, and deserved to, have been at. In addition placement matches are far more common in team ranked games than 1v1 ranked games.

Please do cite a calibration method if you do have one.

Plenty of advantages are brought in, particularly the validation of the competitive side of the game. In addition changing shouldnt be that much work since the source code is freely available for Glicko-2 and afaik only 1 variable has to be determined and changed in the code.







Why dont you go play a game that uses Glicko-2 as its ranking system and judge for yourself whether its better? I mean alarm bells should be ringing when the game Elo was created for switches over to Glicko-2... Surely even you would admit that?
Last edited by SmallBowl; May 10, 2016 at 07:00 PM.
Don't dm me pictures of bowls that you find attractive.
Originally Posted by SmallBowl View Post
I have read the thread, I think you are confused about what the Elo system is and what calibrating the system itself means. The placement matches are not part of the Elo system per say they just decide what your base Elo is. Calibration would be changing how much Elo you win or lose per game based on more/different factors.

No, that's what calibration is.

Originally Posted by SmallBowl View Post
Ok, now that not only are placement matches not a method of calibration,

But they are, in fact they are THE MOST COMMON METHOD OF CALIBRATION. Almost every game that uses ELO uses placement matches.

Originally Posted by SmallBowl View Post
but they have been shown to not always work as if you play badly in 3 matches you have to play hundreds of hours to get the rank you would otherwise, and deserved to, have been at. In addition placement matches are far more common in team ranked games than 1v1 ranked games.

Well I've not heard of a game using literally 3 placement matches, but if you play significantly under your 'true' mmr for 3 games out of 3 games, then you are probably just not as good as you thought you were.

10 games is the usual 'minimum' for placement, and it's almost never binary. You should probably look up how placement matches work, because it seems you have no idea what is going on. I suspect that you got placed in wood in LoL and are a bit sour or something?

I've played 10's of thousands of hours of ELO based games, so I do know your complaints are imagined. If you underperform in every single placement match, then you have overrated yourself. If a "2200" player always plays at 1200 level, then they are a 1200 player, end of story.
Originally Posted by SmallBowl View Post
Please do cite a calibration method if you do have one.

P L A C E M E N T
L
A
C
E
M
E
N
T

Originally Posted by SmallBowl View Post
Plenty of advantages are brought in, particularly the validation of the competitive side of the game.

How about listing at least 1 then? Instead of just saying "oh there is a lot, trust me lol"


Originally Posted by SmallBowl View Post
In addition changing shouldnt be that much work since the source code is freely available for Glicko-2 and afaik only 1 variable has to be determined and changed in the code.

That's a huge assumption, but I can guarantee that it's less work not to change it.

Originally Posted by SmallBowl View Post
Why dont you go play a game that uses Glicko-2 as its ranking system and judge for yourself whether its better?

Because personal experience is extremely biased and I'd have to play hundreds of games at varying skill ratings in the same game under multiple rating systems to get even a marginally fair result.

But besides that, what game even uses glicko? Almost every single major competitive game in the world uses ELO.

Originally Posted by SmallBowl View Post
I mean alarm bells should be ringing when the game Elo was created for switches over to Glicko-2... Surely even you would admit that?

Should I be concerned that Deep Blue will beat me at Aikido? Don't say stupid things just because they are convenient.