A solution to feeders - please read!

[xenocide85]

Okay,

I think I've finally come up with a solution for feeders. I think there needs to be an additional barrier to entry for League Beta games in addition to 80% stay and 10 games played. I think that overall K:D needs to be tracked. I think that if a player is beyond a certain threshold, they should not be allowed to join League Beta. I think the threshold should be an average of their last 20 games, so they can continue to play qualify to get better. The problem is that in almost every league beta game I play, there are usually 2-4 players that are just giant noobs. They don't chat, they feed like crazy, they don't engage in teamplay and they don't appear to learn from their actions or possess common sense. For instance, a normal person would learn after 1 time, that charging into 3 tanks = death. Many of these noobs never appear to learn from their actions and habitually repeat the same mistakes, usually ruining the game for their team.

This solution allows noobs to learn and improve at their own pace, in qualify, and also allows league beta players to not have to put up with noobs ruining their games.

As far as the threshold goes, I think this needs to be voted on. I was thinking that an average K:D of nothing lower than either 1:2 or 2:3 should be tolerated (10 games average).

-xenocide85 / General_Pwn

[eSVau]

In comon sense topics with capitals are rudehock:

[AGGInator]

and you are not the first who wrote something like that..

[xenocide85]

Agginator - do you have a link to the previous posts? Is there a reason it hasn't been adopted so far?

[AGGInator]

huh whot? difficult to find.. was not in specific threads. more additional to other topic posts.
dunno how to search for something like that. i did forget the content..
i do remember that equiem suggested something like implementing k/d ration to bot limitations. but i'm not sure 'bout. i only know that i read ideas like that a few times :wink:

[xenocide85]

In comon sense topics with capitals are rudehock:

eSVau, can you please apply your "common sense" to paladon's post about the battle tanks tournament then...

[xenocide85]

Okay, so I went back and did some analysis. I pulled random players that I remembered from my last 10 games, placed them into categories and pulled stats on them. This is what I found, of the 30 or so players I identified, at least 14 were solid, 6 were question marks and 9 were feeders or at least really noob players (didn't engage in team play, defend cps, suicided, didn't learn from their actions, etc.). **Note: It's worth mentioning that at least 60 players played in my last 10 games, and while I did not grab every good or question mark player, I did try to grab most feeders. They are over represented here, just for the purposes of stat comparison. My analysis does not suggest that we should remove 33% of players from league beta**

What I am trying to do is identify the K:D ratio that should be required for folks to enter League Beta. The highest 'feeder' % was 66% and the lowest good or decent player was 63%. Since the data I used is a cumulative total, I suspect that the player with 63% K:D, Dennyx_Omni, who played a perfectly good game today, has actually gotten better since he started League Beta. I checked and he's played almost 60 games. I don't have isolated stats for his last 10 or 20 games, but I guess his K:D would be most like 85-100+ %. If we remove him from the mix, the lowest K:D% is GEN_Schwarkzkopf at 81%.

Since I can personally vouch for these 2 categories of players based on my in-game experience, it seems like our cut-off should be be around 75%, probably no lower. That means that if your K:D is worse than 3:4 for your last 10 or 20 games you will need to play qualify in order to gain access to League Beta.

So let's apply this standard to the question mark players I identified. As it turns out, some of those players have really good stats so they were probably just having a bad game, which is fine. Some question mark players are clearly not ready for league beta, having 40ish% K:D. Another question mark player looks like he was just barely good enough to be playing in league beta.

Please apply this standard to players you encounter moving forward and give me feedback. I think if the map owners implement this, it will drastically help reduce feeders from league beta, as well as help feeders/noobs get better by playing in qualify - everybody wins.

Good or decent players:
name | ELO ( | overall average K:D | K:D % (if < 95%)
btank 1533 33:29
ORPG 1569 17:13
Dennyx_Omni 1454 14:22 63%
ShadowMeteor9.1 1527 39:33
General_Pwn 1506 20:23 86%
Subbaba 1678 43:16
killerfuerst 1566 30:29
entangL 1489 37:32
Soeldner 1545 23:20
ckibibi 1720 29:23
reddama 1514 18:32
willireichert 1582 33:24
adriankeke 1567 25:17
gen_schwarzkopf 1488 18:22 81%

Question mark:
ThinkTank.Klaus 12-27 1626 34:22
arrow_of_god 11-17 1492 17:20 85%
uLow.4.ever 19-25 1477 27:28
norbit 6-12 1443 7:16 43%
IrrerIvan 12-23 1487 15:36 42%
nobody-shadow 1490 13:17 76%


Feeders:
oormoo 10-64 1466 11:23 47%
parasympathetic 25-57 1478 9:34 26%
tonop 6-29 1469 9:20 45%
fiberoptik 8-23 1469 14:25 56%
a.c. 13-36 1469 9:28 32%
netzero 9-32 1496 10:15 66%
mykukusobig 9-43 1441 12:21 57%
Henker4u 42-71 1443 19:54 35%
killsalot 32-48 1368 6:17 35%

[GEN_Schwarzkopf]

If a Kill:Death ratio standard is only applied to new players trying to get into the ranked bot, well okay, I'd say that's fine.

While I agree in principle with the idea, I would not apply it to existing ranked players as it could penalize pro support players; and thus is not a deterministic indicator of skill on it's own.

If you want an example of support view my Tournament Replay with Korea#2; it's a clear demonstration of how a player can provide good support, supply important consumables like tp breakers, detectors and upgrades and lead to a victory. I'm not saying it's good to sacrifice one's ability to fight near the front line, but I am saying it's possible to do well for the team without having alot of weapons and scoring alot of last hits.

I have seen games were support players with few weapons also had the fewest deaths on their team, but they had the worst Kill:Death ratio simply because they hadn't scored many last hit kills.

It's too bad kill credit is determined by the last hit. Often in key battles there are major assists, but the person with the most weapons tends to suck up all the money and kills like a black hole. It doesn't make the assist less important, especially stuns and disables, but they don't get credit in money or stats for it. Even DoTA tracks assists; and certainly we should too...then we'd clearly really see the importance of teamwork and good support in a numerical value.

I would like to see a way to get rid of the worst feeders, but the line drawn would have to be very low. Honestly I'd start by focusing on getting rid of leavers first...by upping the leaver standards to 85%.

- Rob

[xenocide85]

Thanks GEN,

I know your play style is usually more of a support player, and I often play support as well. If I have a bad start, I always almost go Hunter and don't even focus on getting guns. I completely agree that any solution implemented should not alienate these folks.

However, for this solution to work:
1.) the threshold HAS to be >= 70%, ie 7 Kills : 10 Deaths as last 10 or 20 game average. If you don't you will still be letting in players that don't talk, suicide, feed, and really have no clue what they are doing. These players need to play Qualify until they are ready for the League.
2.) this fix HAS to be applied to players already in the League. That's entire problem we're trying to fix right, remove the really bad noobs that are ruining games right now.

Now trust me, I understand your concern about alienating support players, I think what you will find is that there really aren't many support players that have < 7.5:10 K:D. The strategy behind my approach is to get vetting from the community. So by all means, if you find a solid support player who's overall K:D is lower than 7.5/10, let me know (just search for that player's stats, bring up their page and divide total average kills by total average deaths).

From my analysis, there seems to be a very clear divide between players who know what they are doing, and players who don't, in terms of average K:D. In my post above, players like NetZero and Fiberoptik were really bad and severly imbalanced the teams. They have K:Ds in the high 50s and low 60s. We need to make sure we set the bar high, so the noobs understand they have something to shoot for.

So in my proposal, players would need the last 10 games average K:D to be > 7.5:10. This variable should always be tracked, no matter if in Qualify or League. So if this ratio ever dips, that player should be locked out of League until they fix it. On the very VERY rare cases where you find that a competent player is locked out, I think it would be far simplier to just have a thread for players to petition this restriction on a case by case basis. I think this is the best system to efficiently keep noobs out. Trust me, I've run the numbers, and I don't think there are many players with K:D ratios below this that are actually not noobs.

Please keep in mind, the REAL PROBLEM is that more than 50% of games are ruined because 1 or 2 noobs feed, don't communicate, never properly defend or attack. Personally, I'm getting so frustrated by it, that it makes me question whether I want to continue playing. I hardly consider any of the games I play to be good, now that votekick is removed and noobs are in almost all games.
(votekick is NOT alternative solution to this, they should both be implemented)

[GEN_Schwarzkopf]

I'd support the policy to be applied to existing players if the threshold is low enough to not affect good support players. I'm not sure what that number is without being able to see the entire database, but I'd aim to remove the bottom 10th percentile at most.

The number would need to be based on at least 20 or more games. I see some of the guys %s are very low because they have only played 2 or 3 games.

Profile of a feeder:
killsalot is a perfect example of someone who needs to get kicked back into minor leauge/qualify bot.
http://league.btanks.net/player.php?playerid=111

They guy's rating is 263 and has an ELO of 1376.6 with a leaver penalty of 1113.63. This guy's record is 46-59, leaving 45 of these games....almost half of them. It baffles me how he managed to squeak above 80%, but shows us that our leaver standards are too low.

With killsalot's Kill:Death Ratio of just over 1:3 or 36% I'd say his name is pure irony.

- Rob

[Equiem]

huh whot? difficult to find.. was not in specific threads. more additional to other topic posts.
dunno how to search for something like that. i did forget the content..
i do remember that equiem suggested something like implementing k/d ration to bot limitations. but i'm not sure 'bout. i only know that i read ideas like that a few times :wink:

I proposed several changes concerning both stay ratio and KD a few months back but it never cought on. Everyone just said it was over the top. I still stand by some changes thou.

Profile of a feeder:
killsalot is a perfect example of someone who needs to get kicked back into minor leauge/qualify bot.
http://league.btanks.net/player.php?playerid=111

They guy's rating is 263 and has an ELO of 1376.6 with a leaver penalty of 1113.63. This guy's record is 46-59, leaving 45 of these games....almost half of them. It baffles me how he managed to squeak above 80%, but shows us that our leaver standards are too low.

With killsalot's Kill:Death Ratio of just over 1:3 or 36% I'd say his name is pure irony.

Very much agree, this isnt even close to good enough to play in any league. Why even join if you plan on leaving?

[Prog]

Just 2 points:

*If you complain about feeders, start using the 1505+ points bot! We introduced the bot for it only to be joined by 50%+ winning players, but noone uses it because you probably have to wait a couple of minutes. If everyone who complains about feeders just joins that bot the waiting time would obviously decrease. So the first point would be to use what is avaible already. (Also note: when you don't join that bot because you think you'll have to wait and all others do the same you all are the reason for the waiting time in first place - sorta self fulfilling prophecy.)

*The database of the bots is bigger than the one of the leaguepage. It's older and inherits the qualify games as well. So someone with "46-59, leaving 45" might have played like 100 more games not tracked on the league stats without leaving. Additionally the leaves might be at the very end. The result would be him having 46-59 (45) with an average stay above 80% or even 90%. Leaguestats just underdetermine the stay rate.

[GEN_Schwarzkopf]

*The database of the bots is bigger than the one of the leaguepage. It's older and inherits the qualify games as well. So someone with "46-59, leaving 45" might have played like 100 more games not tracked on the league stats without leaving. Additionally the leaves might be at the very end. The result would be him having 46-59 (45) with an average stay above 80% or even 90%. Leaguestats just underdetermine the stay rate.

I'm glad that at least the 80% considers the last 20 games.

The 1505+ bot never fills in US EAST during the evening prime gaming hours. The only chance to get on the 1505+ bot is when Europe is still on.

Even a 1475+ bot would keep killsalot (or other known noobs) out and might lower the bar low enough to fill a game. (It just wouldn't keep out those who have 0-0 games, so I'd recommend it also requires 10+ rated games as well as an entry requirement of 1475+.

It's funny if you look at the top 200, you'll find a large amount of players under 10 games...people are just creating new accounts and abandoning them if they accumulate too many losses.

You know in other places where an ELO system is used a rating decay system is also used. Inactive players would lose points and slip off the board.

- Rob

[Prog]

I'm glad that at least the 80% considers the last 20 games.

True, didn't remember that. But still you do not see the 20 last games considered as qualify games count as well.



I'm not sure about decay, but I might be biased as I'm #1 and #7 (brag ;P) and don't play that much currently due to sc2^^

[xenocide85]

I agree with GEN:
1.) I've never seen more than 1 person in the 1505 bot in the past 2 months on US East

But I also think that using ELO as a criteria point for entry is wrong for 2 reasons:
1.) ELO is NOT calculated in qualify so any feeders/noobs that have a good enough % stay, will join the League and ruin AT LEAST 10 games before they are locked out
2.) ELO is cumulative, and so in the case of players like Dennyx_omni (http://league.btanks.net/player.php?playerid=3942), who has played about 60 games, while he has improved to be a perfectly good player, his older stats hold would hold him back.

This why K:D ratio % should be tracked as an independent stat across qualify and league beta. I think that it should factored in once 10 games have been played to qualify to see if a player can enter league beta, but after that point, this variable should track the K:D ratio for up to the most recent 20 games. If a player's average K:D ratio falls below some threshold, probably 7.5:10, they need to be kicked out of beta and play qualify until it's better.

This will ensure that:
1.) noobs and feeders are dealt with proactively
2.) noobs and feeders are not in league beta

Any objections to this method around potentially good players getting kicked out should be handled by posts in an exception thread. There will be FAR FEWER exceptions to this logic, if any, than there will be noobs who are kept in qualify. As far as exceptions go, there will also need to be a 2nd variable, some kind of exception flag, so that if a some kind of support player petitions and the community recognizes that he is a good player and should be in league beta, but his K:D is lower b/c he always plays support, or something like that, then this 2nd flag can be turned on, bypassing the K:D ratio.

The logic would be something like:

IF ([K:D Ration of last 20 games] < (7.5/10) AND ([Exception Flag] = FALSE) THEN [Cannot enter]

[Prog]

Just to make things clear: We certainly don't want to exclude anyone from league because they play bad.

So if you don't want them in your games, play on the higher skill bot. Note that if everyone has your thinking of "noone is at the 1505 bot so i won't join that", the bot will never have any players (->self fulfilling prophecy) and you are the one suffering.

[xenocide85]

Wow... I might as well quit now.

And as far at he 1505 bot goes, it's not "my thinking"... there seriously is NEVER anyone on that bot. I'll make a bet with you that if I even idled on that bot, it would still take FOREVER to fill up.

I think you and whoever else runs this map needs to seriously rethink valid solutions to feeders and noobs. I can honestly say that if you keep votekick out and you deliberately keep noobs and feeders in, I will stop playing this map. I think it's only a matter of time before others will too.

It's not fun to spend 2 hours of your life wasted on some game where, despite how good you were, your team was never going to win because one or two noobs completely imbalanced the game.

[TKF]

The ugly 1505 name scares people just make the admin rename the bot.

from
BTanks League Beta 1505+ #X

to
BTanks Pro League Beta #X


The 1505 point limit can be announced when you join the bot. If it's possible the bot can list ur current points and show how many you have if you are below the limit

[xenocide85]

Renaming it is worth a shot, but I'm not expecting much. I think the reason why 1505 doesn't fill up on US East is because 1.) less active players with this score, 2.) players in general don't want to wait for a game to start- regular league beta fills up fast.

But this will still alienate players that don't have a 1505 score, but who are also not noobs. I think the best solution is the one I proposed.

[Equiem]

The ugly 1505 name scares people just make the admin rename the bot.

from
BTanks League Beta 1505+ #X

to
BTanks Pro League Beta #X


The 1505 point limit can be announced when you join the bot. If it's possible the bot can list ur current points and show how many you have if you are below the limit

I really like this small but very important change. Think it would make it fill up a bit faster. Right now I rarely see more then 2 people in it at once.

Any chance you could post some figures how many games are played on that bot now? Like daily aswell as the numbers for the normal bot?