ELO System % Correction Factor proliferates selfish players

[GEN_Schwarzkopf]

Gentlemen,

As you know the bot tallies up stats like creep kills, tower kills, player kills, and uses that as the determinant of "skill". Player stats are used in a % Correction Factor in the ELO calculation; rewarding 'good stats', but actually penalizing up to 15% of a win reward for players with the lower stats in the game, regardless of how good or bad. Players who have taken on a supportive role in the team have low stats and are penalized for this.

First of all I like the idea that stats are collected, as over time it may be an indicator of favorable performance, that is obvious in most cases.

The fact of the matter is that % Correction Factor rewards greedy, selfish farming and intentional 'conveniently' late arrivals to a CPs and important battles and arbitrarily labelling it as skill. It rewards players for dashing to safety faster than their ally instead of saving players in jeopardy. It rewards buying airships with big farming weapons and no stun so the last hit is most likely to come from the you. This leaves the responible players to buying tanks that stun, consumables like nets, teleport breakers and detectors when critically needed, but trading off easy farming. The stunners enable the farmers to make kills because they got the last hit from all their weapons. (This is passed off as skillful and acceptable, while if the stun made the last hit, they are 'kill stealing'.) The farmers are rarely lead the team into a real battle...not unless there is a bunch of creeps or towers they can take out...but they do show up when the enemy is weak for the last hit. The concept of last hit kill reward is a problem in and of itself...it's lead to truly ignorant players.

Do you know any players like this? I think we all do. They can be especially useless during key parts of the game, yet we depend on them because they have robbed the map of it's finite resources only to invest it into weapons so they can outfarm their allies some more. Should this person be rewarded with more points for a win or less points lost for a loss? What a sham!

Since when is the greediest farmer actually the most skillful player?

The ELO system works well for 1v1 games like chess. I recommend the point 15% 'correction factor' for the "best" or "worst" player be removed completely. It sets up the ideal that we have a rat race of farmers on each team trying to cut-eachother's throats to kill the most stuff with complete disregard to strategy. This is a TEAM game and the supportive play like buying upgrades when it's called for or allowing others to farm are just as important. Most important is good communication, co-ordination, positive morale and mutual respect among allies. The correction factor does not reward or support team play and does not belong in the rating system of a team sport.

The team is in it together to win...there should never be reward to focus on the self. The % correction sends the wrong message and should be removed.

- Rob

[Prog]

You are partially right. The thing is a bit more complex tho.

-Players who concentrate mostly on keeping their % factor high play are more likely to lose as they make frequently mistakes by doing so (using spells badly, cptp too late, etc), insofar their own actions which are intended to gain points by abusing the Stats are their own countermasure as they lose additional games by doing that.

-Pure ELO needs a lot of games played to display skill distributions, but we can reduce the amount needed with looking at the stats. As our ELO is used to balance the games it is key to have players as fast as possible ELO ranked accordingly to their skill.

It is obviously no easy task to decide the amount of influence of player stats, as playing selfish (stealing etc) should not be abuseable. I think with the amount we currently have people playing selfish lose enough additional games in comparison to people who don't to make abusing the statfactors not worth the effort. Nevertheless I might be wrong on this, tho I don't know anyone who is better ranked than he should because of playing selfishly.

[olivercamel]

-Players who concentrate mostly on keeping their % factor high play are more likely to lose as they make frequently mistakes by doing so.

I don't know anyone who is better ranked than he should because of playing selfishly.

I agree with it. I also don't know any stealer who is highly ranked. Unless you get this "someone" as a proof, I don't think there is a need to remove the correction factor.

On the contrary, I even think it is needed to enlarge the correction factor to make a pointer differ more within the team.

[Equiem]

On the contrary, I even think it is needed to enlarge the correction factor to make a pointer differ more within the team.

Totally agree with this part. Loosing a game where you had no chance what so ever is never fun and trying to play it out well even thou your team has already given up should be somewhat rewarded. Probably more then it is now.

This is when considering this:

It is obviously no easy task to decide the amount of influence of player stats, as playing selfish (stealing etc) should not be abuseable. I think with the amount we currently have people playing selfish lose enough additional games in comparison to people who don't to make abusing the statfactors not worth the effort. Nevertheless I might be wrong on this, tho I don't know anyone who is better ranked than he should because of playing selfishly.

But even with a more aggressive curve I dont see anyone really abusing this and I can't say I've seen anyone really trying to "steal" kills to get a better rating. It gets contraproductive if you do it too much since your team will be loosing out in terms of income. Usually better to let your team get the kills if you're ahead already to ensure a win.

I'm still considering if an assist system that rewards players helping out on kills might be good.

[GEN_Schwarzkopf]

Gents,

I see your rationale of trying to distribute players by skill, but I've got to honestly say that a greedy farming strategy tends to keep you as the 'big dog' in the game. This ensures you maximize your stats, thus making greedy play the ideal of 'skill.'

I've personally taken on support roles, allowing my allies a greater share of the pie, and still set up pivotal attacks with a hunter to turn the game around. I could have barely any weapons and still change the game. My clan knows this.

The thing is I'll end up with less kills and less score, even if I have the same number of deaths as everyone on my team. At a certain point in the game all your allies fly over your head to farm, you end up with nothing.

When I play with my clan this is not a problem; but in a random team anything that encourages selfish play is going to attract assholes to the game like bees to honey. Have you seen guys yelling solo bot or solo top?

I for one enjoy taking on the tough places...but everyone knows the fastest way to get money is to solo a lane, hopefully against a weak opponent who might feed you kills.

There are many strategies that do not render high returns to the person, but good reward to the team. These strategies are a non-starter when they reward the player responsible with an 85% victory simply because they didn't take on a farming role.

I'm not saying the correction factor doesn't have some value as a performance indicator to help balance games, but seriously it does encourage farming. It makes a person question "How can I get the most money and kills as fast as I can?"

- Rob

[Prog]

You are right, the points difference from support/offensive strategies is indeed a bit of a problem we inherited with our system of capturing elo faster with stats. It's a question of priority: Do we want (most) people's ELO adequat faster or every strategy equally influencial to elo - I see no way to have both and I'm not sure which is the better one to have.

[GEN_Schwarzkopf]

To be honest I'm not sure either. It is a trade-off.

What has been done right is the penalty on leavers as the balancing system generally works well if nobody leaves, or there isn't a horrendous feeder or two on a team.

What can tend to happen is the team with the highest rated player tends to lose if the worst player is a massive feeder. This is especially true if the highest rated player is a massive farmer...it's double trouble.

This is not DoTA: Teamwork is so important that often a "carry" hero does not help the team to victory, they actually help them lose by hogging all the resources in tank cost + weapons thus having a downtime long enough for the base to be raped. Technically this person COST the team the game by running out to farm or be overly aggessive and dying, yet they are rewarded with at 85% loss.

Careless players can still farm massively while feeding the enemy with massive bounty. I'm not saying this happens in every case, but it happens alot.

I don't think the 85% - 15% is really dividing"good" players from bad in ELO points; that is already done with win-loss. It is however distinguishing "offensive" players from "defensive" players.

Ipso facto the most 'skillful' player is defined by ELO as the most offensive player with best win-loss record. If you do a spot check 'study' you will find in many cases the person with the best stats is also the biggest farmer (or 2nd biggest) farmer.

I must take this a step further...

Over the course of 100 games, this 15% could earn/save the player an average of 2 points per game, or result in a rating of 200 higher than a defensive player of the comparable win-loss.

That is a lot of spread, think about it.

- Rob

[Equiem]

To be honest I'm not sure either. It is a trade-off.

What has been done right is the penalty on leavers as the balancing system generally works well if nobody leaves, or there isn't a horrendous feeder or two on a team.

What can tend to happen is the team with the highest rated player tends to lose if the worst player is a massive feeder. This is especially true if the highest rated player is a massive farmer...it's double trouble.

I don't think the 85% - 15% is really dividing"good" players from bad in ELO points; that is already done with win-loss. It is however distinguishing "offensive" players from "defensive" players.

Ipso facto the most 'skillful' player is defined by ELO as the most offensive player with best win-loss record. If you do a spot check 'study' you will find in many cases the person with the best stats is also the biggest farmer (or 2nd biggest) farmer.

I must take this a step further...

Over the course of 100 games, this 15% could earn/save the player an average of 2 points per game, or result in a rating of 200 higher than a defensive player of the comparable win-loss.

That is a lot of spread, think about it.

- Rob

Well, when you get to a certain level you need to get most of the farm in order to ensure a win.

Just consider this: Give a bad player 20k and he'll invest it in something suboptimal and then not use it in the best possible way. Give a good player 10k and he'll probably turn the game.

This is not DoTA: Teamwork is so important that often a "carry" hero does not help the team to victory, they actually help them lose by hogging all the resources in tank cost + weapons thus having a downtime long enough for the base to be raped. Technically this person COST the team the game by running out to farm or be overly aggessive and dying, yet they are rewarded with at 85% loss.

Careless players can still farm massively while feeding the enemy with massive bounty. I'm not saying this happens in every case, but it happens alot.

A great player wont feed, a good player might but you'll probably not see him at the top of the charts then. The thing about a carry, it really does work that way in btanks aswell. I'm starting to think this is the whole point that you're missing. You need to be that "carry hero" in order to win, you just can't be the defensive/supporting player and you should be rewarded for contributing the most.

Btw, its not as much as 2 points a game and if you're consistently getting the 0.85 in games you can't blame playing support since you should be getting your money elsewhere and shouldent be that far behind in terms of income. Be it spawning creeps or trying to destroy the base.

My train of thought might be lost somewhere, seriously hangover, but I'll check what I wrote later on.

[GEN_Schwarzkopf]

Give a bad player 20k and he'll invest it in something suboptimal and then not use it in the best possible way. Give a good player 10k and he'll probably turn the game.

I agree completely on this point.

A great player wont feed, a good player might but you'll probably not see him at the top of the charts then. The thing about a carry, it really does work that way in btanks aswell. I'm starting to think this is the whole point that you're missing. You need to be that "carry hero" in order to win, you just can't be the defensive/supporting player and you should be rewarded for contributing the most.

The creep money is finite, everyone cannot be a carry hero. If someone constantly flies over your head to farm where you are heading they preclude your ability to farm.

I haven't missed the point about the idea of carry hero, certainly allowing money to go to a good and responsible player will be beneficial because they won't die due to foolishness and cost a the team the win. I often allow a known ally to take creep and tower money when I find I'm doing well enough. Others might try to cut their allies throats to get the creep and tower money.

My point is that farming is easy to do whether you're a good hero or not. Having expensive weapons tends to increase probability of a last hit, thus awarding you the credit for the kill. Perhaps that's what I'm really at odds with: Last hit gets credit.

The ELO correction factor has it's merits, but it's not perfect. The farmer or offensive player is rewarded in this system.

If kill credit or even just the kill income was divided amongst all players that assisted in the kill (within the last 5 seconds let's say) it would fundamentally change the way we view skill in light of postive kill:death ratio.

- Rob

[GEN_Schwarzkopf]

Well, when you get to a certain level you need to get most of the farm in order to ensure a win.

No. This is only true if your allies won't make the right choice with the money.

Let's say you are in a Tournament and your allies are all pro Shamen.

If you have a Titan and your ally has a Demon tank and needs more money to get a top teir tank like a Titan, is the 25k best spent on a supercell for the Titan, or is the team better off with another Titan?

I think the answer should be obvious...two Titans are far better than one Titan with an extra supercell.

A well-rounded team IS better than a team with only 1 carry hero. If the carry hero is hunted and killed, the rest of the team falls apart.

-------------------------------

I do see your point, if you are with a bunch of douche bags and the only person you can trust is yourself, then I guess you're better off farming the shit outta creeps and all buildings still standing so you carry the team to victory. And why wouldn't you if it results in a great rating? If you did this in a clan game you would be a unpopular if you did this to your clanmates. (Unless they consent to this, and I'm not saying you do this. This is just hypothetical.)

The problem with this mentality is that when you farm your allies into the poor house it's hard not to misjudge their skill or potential.

I know people who do this, and only say they have much to learn with regards to teamwork.

- Rob

[Equiem]

No. This is only true if your allies won't make the right choice with the money.

Let's say you are in a Tournament and your allies are all pro Shamen.

If you have a Titan and your ally has a Demon tank and needs more money to get a top teir tank like a Titan, is the 25k best spent on a supercell for the Titan, or is the team better off with another Titan?

Well, I would say that about 90% of the players will make the wrong choice with the money. Quite a few players would buy a Super Cell on the Demon Tank instead of getting a tank and even if they get the tank they will continue to waste the skills of the tank.

The whole point with this is that people are not "all pro shamen". You have to consider this.

A well-rounded team IS better than a team with only 1 carry hero. If the carry hero is hunted and killed, the rest of the team falls apart.

Obviously, but you dont really have much of a chocie for team-mates so you have to adapt to the situation. I adapt by trying to overrun and hunt the enemy as vigilant as possible.

The problem with this mentality is that when you farm your allies into the poor house it's hard not to misjudge their skill or potential.

The problem I usually see is when people are behind they make terrible choices like getting that one extra weapon they dont need and fall behind in health or tank tier then falling even further behind. So again, it all comes down to that it doesnt really matter what or how you do it, its still they who play.

If you could somehow give other players money or even better items I would totally buy things for my team-mates. This would be an awesome feature although easy to abuse.

[GEN_Schwarzkopf]

The problem I usually see is when people are behind they make terrible choices like getting that one extra weapon they dont need and fall behind in health or tank tier then falling even further behind. So again, it all comes down to that it doesnt really matter what or how you do it, its still they who play.

I see your point...the case is you can't trust unknown allies in league, so if you want a rating you must go alpha-dog all the way. In a league game you are probably right most of the time...unless you know your allies. The people I play with tend to get to know me quickly as I communicate plans often, especially in the late game.

My experience is different for one main reason:

I regularly play with my friends and clanmates in a league game, they'll all come with me, often having 6 or more of my clan members in the same game. Then we get shuffled. Since I know them all well, I know exactly what effect being outfarmed into the poorhouse has on not just good players but great players. When an outfarmed but great player finally gets some money they turn the game around.

Filling a league game with all your friends makes for a good game. (Unless you don't quite fill it and have 1 horrendous feeder paired with your top-rated shaman for the loss. -- Poor Ring- and Prohunter...) Not everyone is going to come out a winner though.

You see, not having money often means not having armor, speedpack or weapons to keep you from dying. The great players who get their throats cut by a greedy ally tend to get the speedpack and armor to ensure they don't feed, but lose their ability to farm effectively or last hit. The result: the same amount of deaths as the other allies, but hardly any hero kills, creep or building kills. This player may be the one to turn the game around right after they get the money for the tank they needed...but they sure won't be given credit in points or stats...not in league.

Not allowing your allies that need the money some of the farm is a greater liability, if they know what they are doing.

- Rob

[olivercamel]

Your post becomes more and more colorful.

[AGGInator]

i'm just waiting for single colorized letters

btw.. good arguments! i agree with some terms and hope there will be some changes soon.

[ssl]

Your post becomes more and more colorful.

Also my first thought on this post ^^ Sorry for offtopic

[GEN_Schwarzkopf]

:idea: LOL, I admit that using purple was a bit 'over the top'.

- Rob

[AGGInator]

:idea: LOL, I admit that using purple was a bit 'over the top'.

- Rob

looks more like pink then PURPLE or?

[eSVau]

i do not read colored posts, so i dont have an idea about this topic

[GEN_Schwarzkopf]

I just had an idea that could help improve the interpretation of statistics to properly analyze skill of both offensive and defensive/support players.

1. Increase the weight of the Control Points taken to very high. Games are won and lost by control points and quite often greedy players will neglect to help them. This shows teamwork and proper risk-taking when the time is right.

2. Completely remove the creep kills and tank value from the skill equation. Farming alot doesn't mean the player is skillful.

3. Kills stats: Should be determined by the (Kill bounty gained - Kill bounty lost from dying). Often a good player may die and give a small bounty when saving for a big tank then turn things around and make IMPORTANT high bounty kills at the end. This can only be properly weighed by kill bounty.

4. Pie in sky idea. Share gold for all kill bounty with team, or share gold with all those who hit or stunned enemy player in last 10 seconds.

- Rob

[AGGInator]

Farming alot doesn't mean the player is skillful.

btanks is about farming creeps.. pro's will tell you the same. if you creep less, you will have no chance..
if you creep much its high possible to become a strong player. creep and farming ARE/IS important to say how much skill someone has.