|
Post by ExDeath on Oct 23, 2007 1:47:09 GMT -5
ExDeath, either you really misunderstand the impact of RB's ELO system in practice, or you are being deliberately misleading; and given my respect for your analytical and systemological abilities I am inclined to say the latter. I've already debunked this for you privately, and I think I still have the excel files in which I did out ExDeath's Elo with Sasuke's match results applied. The result was a net gain of about 8-10 Elo. You would most emphatically not be in the 2100 neighborhood. Over the course of a 39-game rated win streak, I find it hard to believe that I would only gain 10 Elo points. 3 (4?) of the wins were against Toyotami, one of the wins was against you (Ghostlight). That alone is easily 30. Granted I do have 3 losses on the account, but 2 of those are to Tchichi and 1 is to Calmon, so there's no way your data could be accurate. Keep in mind, Nawglan's archive mostly just counts Sasuke's old games from back before he was reset to 1500. Edit: It's hard to analyze because I don't know which wins were kills and which were surrenders, but, I guess in theory I could actually be losing points over the course of the streak if it's all 1600 rated players surrendering. I don't know. In practice I would never take those matches anyway, but I think there are a lot of wins vs quality opponents that you're simply not accounting for. Not over 2100, but I would guess it at about +40.
|
|
taliesin
Ronin Warlock
Grand Master
Posts: 156
|
Post by taliesin on Oct 23, 2007 5:57:36 GMT -5
Over the course of a 39-game rated win streak, I find it hard to believe that I would only gain 10 Elo points. 3 (4?) of the wins were against Toyotami, one of the wins was against you (Ghostlight). That alone is easily 30. Nope. If you're counting playing Ghostlight, you're counting a VF as an ELO match. Toyo is about a hundred ELO below you. You'd currently score at most about 20 ELO for three surrenders from him, and it would be less than 30 even for three straight kills. This would rise some if the victories had been at his over-2000 ELO best, but I think two of them pre-date it and one post-dates it, am I right? Granted I do have 3 losses on the account, but 2 of those are to Tchichi and 1 is to Calmon, so there's no way your data could be accurate. Tchichi also being about a hundred below you, you'd be losing 13-14 points a surrender and more than that if he managed a kill. That by itself could wipe out the ELO gains from Toyo. Calmon, at over a hundred and fifty below you, would be losing you another 15-18 points depending on the kill/surrender thing. That's a lot of lower-ranked player victories getting nuked - fifteen to eighteen victories over 1750ish players, for example. While the players you mention are, in the Warlocks scheme of things, pretty highly ranked, they're not close enough to you in ELO for a win and a loss to balance out. At the hundred point ELO deficit mark, assuming surrenders, it takes roughly two wins to make back every loss. Most of the people you've been playing are rather over a hundred ELO below you, and wins count for correspondingly less. I do think you'd have some gains as a result of this as you've won a lot of games, and you'd definitely have a higher all-time best, but every loss on both accounts would drag you down a little more as a result of higher ELO, and every win would count for a little less as your ELO grew higher. Life at the top is tough.
|
|
|
Post by Slartucker on Oct 23, 2007 9:31:44 GMT -5
From my email to you:
So the final result of the math was 14 points up, but that given uncertainty in the algorithm and my always taking the most favourable possible result.
|
|
morzas
Ronin Warlock
Posts: 30
|
Post by morzas on Oct 23, 2007 10:46:17 GMT -5
It's worth pointing out that there ARE professional gaming bodies that use Elo systems. Chess would be the obvious one, since that's where Elo began. And they do not use decay. So the tennis analogies are unnecessary. Chess also doesn't suffer from the problem of a small community with an even smaller group of players that don't play anymore, but only log in to keep their scores at the top.
|
|
|
Post by Slartucker on Oct 23, 2007 10:54:31 GMT -5
I'm not sure that's any different from warlocks. Chess rankings lists don't necessarily drop the names of grand masters who are, for example, dead. As a result, it's an often discussed problem that it's hard to compare the rankings of players from different eras.
Anyway, who does that in Warlocks?
|
|
|
Post by Rycchus on Oct 23, 2007 12:11:47 GMT -5
I'm totally gonna keep playing Warlocks when I'm dead. ;D
Hey, I wonder if any of the disappearing acts have died? It's kind of a morbid thought, I know, but it intrigues me that they could have done and we (probably) wouldn't know.
|
|
taliesin
Ronin Warlock
Grand Master
Posts: 156
|
Post by taliesin on Oct 23, 2007 13:39:17 GMT -5
BRAAAAINS!
*chomps on Rycchus' head*
|
|
|
Post by Rycchus on Oct 23, 2007 13:56:46 GMT -5
I'm not dead yet.
|
|
taliesin
Ronin Warlock
Grand Master
Posts: 156
|
Post by taliesin on Oct 23, 2007 14:28:08 GMT -5
I am, however, a disappearing act.
Possibly even an undead disappearing act...
|
|
|
Post by Rycchus on Oct 23, 2007 14:45:10 GMT -5
And you want to make me undead too?! Dude, that's so not cool.
|
|
|
Post by ExDeath on Oct 23, 2007 14:50:42 GMT -5
Re: Slarty
I actually never saw that e-mail. I would guess then that there are about 13 more victories uncounted, including 1 more each vs freesoul and toyotami, but I'll concede the point that it might only end up around the 2060 mark and not 2100 as I initially speculated. Kudos for putting so much work into it!
|
|
Derfel
Ronin Warlock
Did I Do That?
Troublemaker
Posts: 283
|
Post by Derfel on Oct 23, 2007 15:42:44 GMT -5
I would assume, just in terms of probability, that there are at least 3-5 Warlocks on the inactive list that have actually passed away.
Just sayin'.
|
|
|
Post by Rycchus on Oct 23, 2007 16:10:30 GMT -5
What probabilities are you using? Just guesswork/"feeling" or do you have something on it...?
I'm curious now.
There are 1243 inactive Warlocks. (Then again, we're probably only interested in the ones we've heard of, not the ones who've signed up and disappeared. If we get rid of the 1500s, that leaves us with 540.) I'm not sure what the probability of one of them dying in a given time is though... how would you go about estimating that?
|
|
laxen
Ronin Warlock
Posts: 77
|
Post by laxen on Oct 23, 2007 16:35:38 GMT -5
Rycchus, was the "I'm not dead yet" an unintentional Monty Python reference or subtle genius?? ("He's NOT the Messiah, he's a very naughty boy!", "This parrot is no more! It is dead! It is deceased! It has gone to meet its maker! It would be pushing up the daisies if you hadn't NAILED it to its perch!" ) I have a feeling only you (and possibly Taliesin, being Scottish) will get this ...
|
|
|
Post by Rycchus on Oct 23, 2007 17:16:47 GMT -5
No no, apparently the Americans are all-too-familiar with Monty Python as well. I've made that mistake before. (Plus, there are more Brits than just the three of us about )
|
|