ELO Discussion

[Pook]

Yes, we could.

Tiger said he didn't see the value in reprocessing all the old games when a complete wipe will happen with the introduction of R3 in Early December.

[Raveen]

I think the suggestion was as an alternative to a complete wipe base data off the current formula with all the old game data. It'd at least be an interesting experiment to see who ends up where rank-wise, I suspect that the old stackers gain without losing problem would rear it's head though.

All in all, A total wipe is bad but maybe it's the best option available.

[Grim_Reaper_4u]

QUOTE (Pook @ Nov 22 2006, 05:12 PM) Most likely impossible to play twice as much as someone else yet not have more experience than they do, but for the sake of argument, sure.

You are kidding of course right pook? Some peeps just have a natural feel for a game. I've seen some 6 week newbs fly ints better than 5 years vets. Time-in-game means nothing esp if you keep wiping stats all the time. So if under R3 Weed all of a sudden gets a life and only plays once a month then he deserves to be a Vet1 for a year? Some old vets will never play enough to reflect their true ELO value (my best guess is it will take 1000 games/player to get accurate ELO). Since some of the "kids" of the beta years are now adults with jobs/families they have less time to spend on Alleg, however their skills are still similar.

What I'm saying is that someones ELo (in its current proposed form) takes too fricking long to get its correct value for a large portion of the player base. During this period balanced games will suck because there is no balance (after all having a difference between a superb player and a lousy player of 3 or 4 ranks is nothing, and we all know there are lots of lousy players who don't qualify as newbs anymore, just look at me /wink.gif" style="vertical-align:middle" emoid=";)" border="0" alt="wink.gif" /> )

It will take literally ages before some of our ueber vets get expert ranks and thus they will skew the balance in every game they play, which is especially painful in small games where 1 Weed or Culm on 1 side is often enough to win /wink.gif" style="vertical-align:middle" emoid=";)" border="0" alt="wink.gif" />. You will probably counter "but hey they will rise in ranks fast that way" but if they play a lot less than the rest it is still not fast enough for the peeps who are on the other team /wink.gif" style="vertical-align:middle" emoid=";)" border="0" alt="wink.gif" />

1000 games/player is an awfull lot of games Pook, too much if you ask me /smile.gif" style="vertical-align:middle" emoid=":)" border="0" alt="smile.gif" />

[Greator_SST]

...I think all this talk is pretty silly. I enjoy the game, rank or no rank. But here's my only quibble with ELO. I don't think the current ranking system makes enough of a distinction between ranks. Here's my logic.

Player Team ELO Player Team ELO
..15.......... 15............13.............13
................................13.............26
..15.......... 30............11.............37
..18..........48.............15.............52
..17..........65.............14.............66
..16..........81.............15.............81
..17..........98.............13.............94
................................11............105
..16..........114........... 12.............117
..18..........132........... 15.............132

Even though, team one is down two players, I still think that team has a significant advantage. My point being, there is often a big difference between a 13 and a 15.

I don't know, now that I've written it out, maybe this would equate to a very good game, but maybe not.

[Pook]

QUOTE (HotBird6 @ Nov 22 2006, 10:04 AM) Say (theoretically) there were 2 people with exactly the same level of skill and weren't improving.

One played 14 times per week, the other 7.

QUOTE (Grim_Reaper_4u @ Nov 22 2006, 12:57 PM) You are kidding of course right pook? ...

Read what I wrote carefully.

He's saying 2 people at exactly the same skill level. I simply said that it seems impossible that one person playing twice as much as another doesn't have more EXPERIENCE playing.

His entire post was based around an incorrect assumption anyway... he was really talking not about ELO but in the time it takes for the newbie modifier to go away.

[SNAFU]

*emerges briefly from under stone*

To clarify... I wasn't talking about newb ELO.

I was thinking, as Grim mentioned, about someone who was a vet that didn't play often enough for the system to get an accurate estimation of their skill.

But as already established, my assumptions are incorrect, in some way beyond my comprehension.

There'd be a lot less hassle if no-one was able to see their rank/elo or whatever. Sure you wouldn't be able to see if the game was stacked until it finished, but most people know their own skill level relative to other players without having a number tell them.

*slithers away*

[jgbaxter]

Well, if autobalance is being used then seeing the rankings of players wouldn't matter, as it can be done hidden... and even if it wasn't it wouldn't matter in that case if it was shown. /smile.gif" style="vertical-align:middle" emoid=":)" border="0" alt="smile.gif" />

[Pook]

QUOTE (HotBird6 @ Nov 22 2006, 06:16 PM) *emerges briefly from under stone*

To clarify... I wasn't talking about newb ELO.

I was thinking, as Grim mentioned, about someone who was a vet that didn't play often enough for the system to get an accurate estimation of their skill.

But as already established, my assumptions are incorrect, in some way beyond my comprehension.

There'd be a lot less hassle if no-one was able to see their rank/elo or whatever. Sure you wouldn't be able to see if the game was stacked until it finished, but most people know their own skill level relative to other players without having a number tell them.

*slithers away*

Since you start at the middle (1500), then someone who doesn't play much doesn't really affect the rest of the system - since the affect someone has on the system has to do with how far they are away from the middle.

[Grimmwolf_GB]

Someone mentioned decreasing the distance between ranks in an old thread. I think it would be a good idea: 100 elo is a huge difference. Maybe make it 50 instead?

[jgbaxter]

Decreasing the distance would mean that an expert like Weed would be less of a bonus to a team.

I don't agree with that, actually, I believe increasing it would make more sense. /mrgreen.gif" style="vertical-align:middle" emoid=":D" border="0" alt="mrgreen.gif" />

Well, not increasing per se, however creating a k-factor (yes, even after a year I'm still talking about my 'special' "k" lol) whereby players have a lower adjustment for games the higher their elo goes is good for several reasons. One, it gives something towards what you mean Grimm, ranks have more of an equilibrium once players start to plateau into the veteran ranks. Two, it means that you won't see as much of a large fluctuation from game to game and day to day with players once they've become closer to having elo realize their actual skill level.

Pretty simple really. Let's say a game is worth 1-32 pts with 16 as the average more or less. The point change is actually a matter of the k-factor, such that a novice would have a k-factor of 32, thus gaining the full breath of a max. adjust of 32, an intermediate would have a k-factor of 24, a veteran a k-factor of 16, while the expert players k-factor would be 8.

Thus the higher in elo you go the slower any changes would be. Perfect, probably not, better, I think so.

/smile.gif" style="vertical-align:middle" emoid=":)" border="0" alt="smile.gif" />