Forum
Advisor Blog
Reviews
Strategies
Ideas
aaaaaaaaaaaargh stop asking me
|
Grunvagr wrote
at 8:46 PM, Sunday March 4, 2007 EST
okay - to not type this out for the 30th time... here's how the new ranking / top 25 thing works.
Rank - click on the top 25, it is the 0.00 value on the left. Think of it as how much ass someone has kicked in the past. (ie, were they ever #1 in elo, or top 5, etc, or how many times they played in the 1900s+ finishing the games in the top 10 and way up there) Elo - the number by your name and in your profile. Think of it like this, the higher you are, the more in position you are to ADD to your rank. So imagine someone with a rank of 1.52, if their elo is down in the 1700s, they probably wont add much to it. But if they are in the 1900s, 2000s and at the very top they can add significantly to it. Technical mumbojumbo: ____________________ How's this actually work? Rank has do with Elo - just like the previous scoring system did. It is calculated: 1/your place in elo meaning, say player A wins a game and has 1900 elo. That means nothing, what matters is how does it compare to others. If 1900 is the 10th best elo rating, then that person gets 1/10, or .10 added to their rank. if you finish a game with 1750 elo and you have the 1500th best elo rating in the game, you get 1/1500, lets just say thats tiny cuz my calculator just gave me an error HEHE. SUMMARY: Rank is how much ass someone kicked in the past. Having high elo compared to others = putting yourself in position to add significantly to your 'rank' value. Pros: What's really cool is if someone is #1 in elo and kicks ass, then has a few bad games and drops, the game respects them by keeping them in the top 25. So no more slipping from 2nd place to 26th cuz you tried to take over 1st spot. Now, people gradually drop off the top 25 as others pass them, rather than massive jumps. It rewards people for being great players. Ok, youve never had the lucky streak to get to #1 elo, but you play a lot of games in the top 10 of elo, well, that's 1/10 or 1/9 or 1/11 added to your rank, .1, .11 etc, good values. You'll get rewarded, despite never making it to the top. Cons: It's new, people always dislike what's new until they master it. It's weird at the moment cuz how do you know what position you are, how do you know what 1840 elo means compared to others? it doesnt really show you unless youre on the top 25 (at the moment), probably other cons but im tired of typing Note: people say playing often can pad your score. While this is technically true, u have to maintain high elo while you play a lot. And again, 1 game in 2nd or 1st place of elo instantly massacres all the gains someone else makes playing 70 games in 100th. |
|
no_Wolf wrote
at 4:31 PM, Wednesday March 7, 2007 EST Grun, you've missed my point, which is unfortunate, you put so much work into that post.
I do not believe that maintaining 1st is an equal feat to maintaining 25th. But it is not 25 times greater. |
|
|
Grunvagr wrote
at 4:38 PM, Wednesday March 7, 2007 EST it's hard to quantify anything exactly, wolf :)
who knows if it is 1/10th as hard or 1/25th, fact of the matter is that IF the values for rank were not as 'steep' as they are now, then players who play too much will get WAY too much. look at me, Ive played 300+ games, that's freggin rediculous. I have gotten as high as 3rd best elo, and have hovered in the top 10 of elo for over 15 games. I've played in the top 100 almost all the time, or at least, for probably 60% of the games I play, unless I drop to the top 250. if the values were any less steep, and a 30th best elo finish awarded that square root formula I would get .18 for that. It is simply WAY too crazy. I would be in the lead overall on the top 25 by an INSANE amount. am I the best player? hell no. am I a good player, hell ya ill too my own horn. Thing is, the new system works very well. And we have yet to see how it works over the long-term (weeks or a month). |
|
|
no_Wolf wrote
at 4:43 PM, Wednesday March 7, 2007 EST Except this formula would be applied to everyone else as well.
What's wrong with .18 for 30th? You tell me "it's crazy" but don't say why. It's less than a fifth of what finishing first gives one. |
|
|
no_Wolf wrote
at 4:45 PM, Wednesday March 7, 2007 EST Is anyone willing to raise their hand and say "I think staying first is 25 times more difficult than staying 25"? Anybody?
We may not know exact numbers, but there's an obvious intuitive limit to it, and this steps beyond it. |
|
|
Grunvagr wrote
at 4:54 PM, Wednesday March 7, 2007 EST wolf - I understand it would get applied to everyone else
the 1,000th best elo after a game would give 0.03 roughly, and the 50th best elo would give 0.14 5 games in 1,000th place (which is SUPER easy to do for the better players) = the same as someone who has the 50th best elo? That might be a 200 or more difference in elo, quite a massive difference in talent, I would argue. To answer your question: It is CRAZY because sheer volume of games played would determine who was in the top 25. Read my example in the previous post, I played 300+ games at the moment. I would be in #1 by a MILE. Would I deserve it? if someone else who was a great player played 300 games to catch up to me they might have a higher rank, but the fact that i have more time to goof around would dictate that I get to be #1?? that has a major flaw, it would FORCE people to play CONSTANTLY if they wanted to have any shot at the top, it would render 1st place not a recognition of talent but of time dedicated + talent (with the emphasis on time) I can't make that clearer. It just simply would not work. The top 25 would be a mess and constantly changing too. There is a certain prestige at the moment to being #1 in elo or rank. The community looks up to those players and aspires to get there. If that system was in place with the square root values it would mean the top 25 would be changing constantly according to who played often (among talented players) this is not a tiny point, this is kind of a massive problem, it simply would not make sense |
|
|
no_Wolf wrote
at 5:09 PM, Wednesday March 7, 2007 EST "I would be in #1 by a MILE."
I'd like to see the proof for that. In fact, say something that isn't rhetoric. You say "it's be a mess" "it would be crazy". I say the top 25 would be fine, movement is not a problem, and no one holds the person in 1st in high regard, just the spot. I say it simply would work, and it's not a problem. Try me, I can say a lot. Do you think you can play five games to first's one, and maintain your spot, Grun? I don't know for certain, let's find out. Let's try it. Let's get some numbers to supplement hollow rhetoric. |
|
ryan2 wrote
at 6:27 PM, Wednesday March 7, 2007 EST "hollow rhetoric"
Wolf, Grun is trying to explain why the rankings work not trying to sell you anything. He's being helpful, you shouldn't call it hollow. And the rankings do work well. You have a 100th rating player in the top 5, (only being rewarded 1/100th what 1st gets but still in the top 5). Doesn't this show you that a 100th rating player can compete in the top 5? Another thing to look at when comparing the skill needed by a 25th rating player compared to a 1st rating player is the rating increases. The skill of winning the game for each player is rewarded by a rating increase, (if we forget about elo differences). For example 25th has 1800 and wins he goes to 1830. If 1st is at 1950 and wins he goes to 1980. I think this shows you that the skill of winning/losing is reward similarly. The ranking is another story and does not attempt to reward for skill in a game but rather over many games by looking at the relative rating. Does this show you that a 25th rating player gets rewarded for skill similar to a 1st rating player? (in fact because of ELO the 1st rating player will probably be rewarded less) |
|
the brain wrote
at 6:32 PM, Wednesday March 7, 2007 EST "Here's the key thing. How come players like Tzisc, Comik, Lothros, and a bunch of others manage to climb to the top 10 in elo or to the #1 spot in elo SO OFTEN? yes, the answer is they get on a lucky streak
BUT - the key thing is, they put themselves in a position where a lucky streak of stringing 5 1st place finishes in a row puts them there BECAUSE they are constantly skilled enough to hover in the top 100." Exactly my point, it takes skill to maintain your top 100 place, it takes luck to gain 1st. Getting a lucky streak is a matter of playing enough, and therefore this scoring still rewards those who play more often. |
|
|
Grunvagr wrote
at 9:50 PM, Wednesday March 7, 2007 EST There is no absolutely flawless ranking system. The current one, I find, has far fewer flaws than the previous one - so in that regard, it should stay for now (this is BETA after all). Can we improve it? sure, anything might be improved upon.
The hollow rhetoric comment was a bit silly.. if you read my long posts i've made several points, specifically to answer that, WITH data. Do I have specifics worth of hundreds of games info to see how it would work? no, but common sense says that someone who plays 300 games vs someone who plays 70, assuming both are top 100 players with similar skill, the person with 300 games will undoubtedly have more points under that proposed 'square root' formula. as for the Brain's comment - you can't even remove luck as an aspect of a game that uses dice. It's kdice, live or die by the rolls. Now, what we could do, is propose a system that is even better. The more I got to thinking, the more I thought that perhaps a percentage chart would be a better indicator of long-term talen (skill) at the game. For example, have a minimum number of games to be on the top 25. (50, or 75) then only count games that finish in the top 100 as a victory so the top 25 would show something like player A - 142 games, 77% player B - 75 games, 76.4% player C - 67 games, 76.3% player D - 324 games, 74.9% etc. But then, that system would also need to have some sort of bonus for top 5 finishes or finishing #1 in elo. At least it is a thought. I do hope this thread stays constructive at least, in the criticisms. |
|
XicaDaSilva wrote
at 9:55 PM, Wednesday March 7, 2007 EST How easy is to break top 25?
Well, right now: 1924 (12th) .08/game 1873 (53rd) .05/game 1839 (97th) .01/game the 25th ranked 1.50 So for a new player that can reach and maintain only 1850, needs 150 games. One that can maintain 1875 need 75 games, while one that can maintain 1925 needs 20 games. For entering top ten, you really need to reach and spend time with a top 3 ELO rating. Obviously, the secret is the maintain a high rating. How easy is to reach a top 3 ELO rating, let's say 1950? For a loner, no matter how skilled maintainnig a top 3 ELO rating is almost impossible. Is more than a steak of 1st place finishes. Just by playing the next available game at 1700 level tables is a very unlikely way of getting to 1950. Top playes seem to have many friends, many of whom have similar high ratings. By coordinating they can play at tables where the average ELO is higher. The top ELO ratings may change hands, but they don't leave the club. Friends can also subtly help by attacking others in a neutral situation, allowing their friends to get 4th/5th rather than 6th/7th. So the advice for the next Kdice geniuses, don't concentrate on the ranking system, the system that will allow somebody to reach the top exclusevly on skill will not be found anytime soon. while mastering the social aspect of the game enjoy the fact that you can play against the top players more easily now, since they are in the points competion. good luck! |
