Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
ELO System % Correction Factor proliferates selfish players
#21
Of course you're right Aggi, alot of farming tends to help a build, I'm just saying the ability to farm creeps does not define a person as skillful.

For example: What good is the farming if the guy feeds? Runs up a lane, gets some creep money and dies. Rinse and repeat.

Creep income is finite. Creeping on it's own doesn't really contribute to the team, it just puts money in the player's pocket. Taking CPs, building down and players down is contributing.

- Rob
Former Chieftain of Clan toaf on US EAST  *Retired*

Just playing for the fun of it now.
Reply
#22
First gents, I'd like to start by saying this is just a rant, and is not directed at anyone here.

I find again after a few more games that some of the most reckless of players are being rewarded by this correction factor. I just had a game with one guy who had a heli with 2400 hp and over $10,000 worth of weapons! The guy fed like crazy but farmed lots of creeps. Is this skillful, hell no, it's people like this that ruin it for the rest of the team. Sometimes inappropriate spending is worse than having their help in the first place.

A true team player or support player is not getting a proper rating from this system.

1. This amount of credit awarded for kills in the correction factor needs to reduced points to consider the amount of money fed to the enemy from deaths. I've seen games ruined, for example wyrd_cow went 60-63, completely screwing our team by farming and dying constantly. He farmed us all poor, plus handed over tons of FREE money to the enemy from his higher bounty. He doesn't even realize his recklessness is akin to team sabotage, but for this he gets rewarded by 15% loss reduction, as well as an attitude that he's a skillful player. It's false logic.

You know what else is INCREDIBLE? wyrd_cow record is 16 wins, 13 losses. That's 29 games, 20 of them he left. THis means 4 of the games he left, he was awarded a win for! Check it out, it's insanity! http://league.btanks.net/player.php?playerid=4650

2. Thus, Leavers should not be awarded wins, PERIOD. I've seen many cases where someone leaves at very start, getting a win (even though a large penalty) it produces a FALSE ELO; which leads to bad game balancing. Sure their rating is correct, but balancing is done by ELO.

3. If you capture a CP by using a barricade you should get credit for the capture whether you are physically present on the CP at the time of capture or not. (You should also get the gold for it too.) This is a map bug. In fact if you should get extra credit if you use a barricade; this should offset the stupidity caused from player greed; already promoted by the existing stats correction factor system.

4. Capturing CPs should be weighted higher in credit than ANYTHING else in the ELO correction factor. Taking CPs tip the scales of the game. The ability to capture or defend CPs is pivotal. Players building to only farm ruin games; CPs fall quickly when a player continues to spend on weapons and nothing on getting a bigger tank.

5. Going AFK should be heavily penalized in the correction factor. Nobody should be getting 15% loss reduction if they afked 4 minutes. That alone could have caused the loss. -afk and -afkai should not stop afk time from accumulating.

6. We should start banning people with penalties of over 1000...it's unfathomable how they are still playing. These are the game ruiners.

7. Points should be awarded for the number of games played. I'm sick and tired of a game being "balanced" with some noob who is 0-0 and yet the game is being "balanced" with a rating of 1500. Having someone who has never played a league game being put on your team and having a probability of success based on a 1500 ELO is ludicrous. Being banned and made to re-qualify should not be advantageous for anyone; and neither should creating many multiple accounts...it's all attempts to manipulate a flawed points system. Starting over should not 'help' your rating in most cases.

(a) After 100 games, a rating of 1500 is very good. The fact of the matter is that most players are better if they've played alot, so they should get a small token of ELO credit for games played. (Somewhere in the range of 0.5 - 1.0 bonus point per game played in good standing. For example 100 games played would result in a rating of 100 additional points higher than 1500 if you were 50-50.) This type of thing is already done for players offsetting penalties, but I propose this should be awarded also to players with no penalties as well, and count towards ELO.

(b) After 100 games a fighter build will have an average of 200 more ELO points in rating than a support player. Why? Because the last hit gets credit for the kill. The ELO correction factor system does NOT reward teamwork or support players, it penalizes them. A well rounded team does not consist of all airships or medivacs heavy in weapons, however some players use this same build almost every time cause it's easy to farm, and with more weapons, it's more likely they'll get the last hit/kill credit. Because they choose to take this role, they get the high rating in this system. "Let someone else stun, then I can make the kill." It's the formula for success in this system. This is part of the reason I think the ELO correction factor is bollucks...a win is a win and a loss is a loss. Unselfish teamwork leads to having an under-rated ELO.

*I'm not saying our top rated players are guilty of never getting ground/stun tanks mid game...I'm just saying that the players choosing this build tend to have the higher ratings.

© Newer players are not interested in teamwork for a win, they're interested in personal stats. They're failing to save CPs just so they can farm their lane. They've been told their stats affects their rating and it affects proper judgement. I wish there was a way to penalize a player's ELO correction factor if they fail to help a CP that goes down...seen that alot.

8. The amount of credit given or creep kills should be reduced. I've watched guys on team cut eachother's throats using ultimates to get creep kills, instead of using them on heros...it's ridiculous.

- Rob
Former Chieftain of Clan toaf on US EAST  *Retired*

Just playing for the fun of it now.
Reply
#23
Do you know any of our factors and how all those values are weighted?
Marvin Wrote:The first ten million years were the worst and the second ten million years, they were the worst too. The third ten million years I didn't enjoy at all. After that I went into a bit of a decline
Reply
#24
Hi eSVau,

I can see what is already posted on the ELO http://league.btanks.net
Plus I can see the relation between the stats posted for each game and the corresponding % correction.

If all variables are approximately the same except one in a data stats comparison, then you have isolated it and can extrapolate it's weighting, especially if the factors are linear. (Linear/Matrix Algebra) (I'll bet dollars to donuts they are linear.)

This process can be repeated for each variable. Given enough data stats to mine you can determine a formula just from the data.

But let's not beat around the bush here, do you have the formula or know who does? Perhaps then the real discussion can begin.

Are you able to post the formula eSVau?

- Rob, P.Eng
Former Chieftain of Clan toaf on US EAST  *Retired*

Just playing for the fun of it now.
Reply
#25
Ok GEN, perhaps what you need to do before coming into this discussion you must consider the following actions:

1) Make 2 new accounts. One is your "support/no-farm" account and the other is your "greedy-bastard/all-farm" account. Then you must demonstrate how being a greedy sonofabitch really inflates your ELO. And you should save all the replays just in case anyone asks.

2) Consider the average Btanks player when you are creating a discussion. Most don't have a clue when to stun or ult. Just look at your last tournament game there were 3 guys on the other team (airships) randomly ulting waves of creeps. Hell one guy would ult the creeps and 1/2 sec later the other guy would ult the dead wave lol. It's really pathetic there is such lack of coordination in btanks to begin with that for GEN to suppose that most players know that being greedy rewards their ELO is preposterous. Most btanks players run around like chickens with their heads cut off only TPing sometimes and farming lots probably because they don't understand the mechanics of last-hitting in this game at all (hell, I don't even bother trying to do it). The point here is that they don't PURPOSEFULLY be greedy n shit GEN, all they are trying to do is compensate for all the last hits they failed to achieve earlier on and they are impatient to get their next purchase. Most of the time you can say well if a person dies its their own fault. There are few times when it really is another player's fault but at the same time players are predictable especially if they disagree with you. Using the information at hand there are ways to avoid a loss somewhat. This is where the great equalizer of "KSing" comes into play I guess.

The only really great point against your argument is that you play carry and one feeder lets the whole team down. At this point the ELO correction rewards the greedy bastard who may completely be at no fault whatsoever. But I think that's admittedly hard to say since this is a team game.
Reply
#26
Just to put a new idea into the discussion: Is it possible or wanted to make the stats factor be determined on the number of games? Like: If you start playing league your the stats will effect your point gain/loss a lot, but the more games you play the less their impact is. This would be something like the best of both worlds: higher stats value at the beginning to provide a fast approximate correlation of skill and elo for the balance function and selfish playing gives no advantage in the long run as the stats won't effect the points after (say) 50 games.
Reply
#27
Well, you are right. The formula does apply correctly for good players.

The problem is the ELO correction factor applies incorrectly for bad players who farm creeps, make alot of kills but also feed alot.

There is a simple solution: If the correction factor didn't consider creep kills and removed points for FEEDING, the good players would still be on top, as their kill to death ratio is good; but the feeders would be penalized.

Are you able to get the formula eSVau?

- Rob
Former Chieftain of Clan toaf on US EAST  *Retired*

Just playing for the fun of it now.
Reply
#28
Quote:Just to put a new idea into the discussion: Is it possible or wanted to make the stats factor be determined on the number of games?

Yes, I think this is a good idea.

I think a player who complete a game without AFKing or quitting should get 0.5 points per game added to their ELO.

The reason is because there are with no stats coming in with 1500 ELO and pros players with 100 games played with less than 1500 ELO.

- Rob
Former Chieftain of Clan toaf on US EAST  *Retired*

Just playing for the fun of it now.
Reply
#29
GEN_Schwarzkopf Wrote:Hi eSVau,
[...]
Are you able to post the formula eSVau?
Of course i'm able to since i've made itTongue
[attachment=1]

All stats are distributed into three levels (priorities), every level counts twice as much as the next lower level (i've called it gradation of 0.5).
[attachment=0]
The whole stats rating calculation using average values of all games and players for each stats with different weights - deaths are negative

and now my bed is callingConfusedhock:


Attached Files Thumbnail(s)
       
Marvin Wrote:The first ten million years were the worst and the second ten million years, they were the worst too. The third ten million years I didn't enjoy at all. After that I went into a bit of a decline
Reply
#30
Great work, this is very helpful.

Can you list the elements of priority? (Which stats are being considered and how much is each worth?)

Can you list which are higher priority?

- Rob
Former Chieftain of Clan toaf on US EAST  *Retired*

Just playing for the fun of it now.
Reply
#31
We are worried that revealing exact data will lead to even more selfish play-styles, which the map itselfs favours in some matters :?

We also will adjust those values again when buildingskills and afk tracking will work correctly. And the other average values taken seems also a bit offscale (10k games are more accurate than 1k gamesWink)
Marvin Wrote:The first ten million years were the worst and the second ten million years, they were the worst too. The third ten million years I didn't enjoy at all. After that I went into a bit of a decline
Reply
#32
Okay so we agree.
Former Chieftain of Clan toaf on US EAST  *Retired*

Just playing for the fun of it now.
Reply
#33
eSVau Wrote:We are worried that revealing exact data will lead to even more selfish play-styles, which the map itselfs favours in some matters :?

We also will adjust those values again when buildingskills and afk tracking will work correctly. And the other average values taken seems also a bit offscale (10k games are more accurate than 1k gamesWink)

Something is SERIOUSLY wrong with the correction factor, check this out:

http://league.btanks.net/game.php?gameid=9997

Why is the guy on the winning team with the worst kill:death ratio, almost the most deaths, lowest tank level, lowest tank worth, lowest level, least creeps killed getting the most credit in % correction?

Check out the replay if you need further proof the worst player w_Wchoi (4) got the most points for the win. That is whacked !

It seems apparent to me part of the reason is the building kills bug is still present, as well it's seems pretty clear we're giving skill points for spawning units. Allowing Units Spawned to factor into skill stats makes no sense whatsoever...even if it's a level 1 priority...it doesn't represent skill in any form.

- Rob


Attached Files
.w3g   LastReplay.w3g (Size: 3.11 MB / Downloads: 153)
Former Chieftain of Clan toaf on US EAST  *Retired*

Just playing for the fun of it now.
Reply
#34
Indeed that does look fishy and should not happen.
Reply
#35
The toaf Dream Team vs BTFG game replay shows an example of great teamwork and support play. Watch how red has only have 1 bombard rocket almost all game and buy lots of nets and tp breakers.

http://btanks.net/forum/download/file.php?id=1184

- Rob
Former Chieftain of Clan toaf on US EAST  *Retired*

Just playing for the fun of it now.
Reply


Possibly Related Threads…
Thread Author Replies Views Last Post
  please respect china players! (Galbatorix MrTown[UC]) guangzhou265 19 14,252 2015-03-29, 19:03:58
Last Post: Wapiti.
  Elo System ichwillrotsein 3 3,763 2014-07-27, 07:59:05
Last Post: RaptorXI
  deep hack players and discrimation guangzhou265 0 1,569 2014-04-26, 18:47:00
Last Post: guangzhou265
  pls change elo system galadion1 13 7,598 2013-12-27, 12:09:23
Last Post: Althend
  kick system Terror-Sheep 6 4,564 2012-03-09, 17:37:48
Last Post: dirtyterror



Users browsing this thread: 1 Guest(s)