Improved Rating System

[Ironsights]

I beleive the current rating system is somewhat confusing, and most folks consider it meaningless. I think an improvement could be made, using a simple formula, that would make things much clearer, and more useful as a skill rating.

 

First, ratings -can- go negative.

Base rating, ie total newb never played before, is 0

Kills give +2 rating: allowing kills to affect rating makes the rating more meaningful to player skill

Deaths give -1 rating: again, it reflects player skill. Making it count as half a kill reflect the fact that you sometimes have to die for strategic advancement, such as a tank sacing himself to save the rest of the team.

Assists give +1 rating: again, this reflects player skill. Even if the player isn't finishing off enemy heroes, being in the fight is worth more than playing farmville all game long. Also rewards team players.

A win gives +10 rating: Winning, in the end, is the point. If your team wins, you should get rating to reflect that.

A loss is -10 rating: Wins and losses should hit you equally.

A surrender should be -5 rating: this does two things, one it encourages people not to tower hug for the last 30 minutes of a losing game, and two it allows people to surrender if two or more players drop in teh beggining of the game with less loss of rating.

Leaving/Dropping the game should be -15 rating: Rage Quits are really not acceptible.

 

Surrender option should be disabled once the artifact stops firing: This is to prevent players from base hugging to farm up a KD and then surrender once the artifact stops helping them to "steal" the win from the winners. Alternatively, the surrender option could be set to disabled after the three base towers fall, again to prevent "win stealing".

 

I beleive that this change to the rating system would make it much more useful to players in pubs an ih alike, and could help solve the problem of base huggers prolonging a game by 20-30 minutes.

 

 

Edit: had forgotten to add dropping to the count.

[Mus]

Dropping makes you lose rating? That's hardly fair.

 

In my opinion rating is pointless anyway, even if we use your suggested system. There will always be games where players will have to leave or want to leave.

 

Eg: I join a game, enemy team has new players on it and they feed my team over and over. This game is a waste of time and I always leave.

 

I join a game, enemy team is losing badly but does not want to surrender, that's fine I just leave because I don't want my time wasted. I already won.

 

I join a game and something comes up forcing me to leave the game.

[InserT]

I dont understand why you should only lose half the points if you surrender...

Its really no difference between surrendering and losing by getting your artifact killed.

Wouldnt be a good thing to encourage ppl to surrender after 20 minutes of the game.

[Takeray]

Why should a support get less rating than a carry?

[CrazySoldier]

So a tank goes 1-10 vs a carry who goes 5-5 why should they lose more points if they are playing their role correctly.

[Ironsights]

Dropping makes you lose rating? That's hardly fair.

 

In my opinion rating is pointless anyway, even if we use your suggested system. There will always be games where players will have to leave or want to leave.

 

Eg: I join a game, enemy team has new players on it and they feed my team over and over. This game is a waste of time and I always leave.

 

I join a game, enemy team is losing badly but does not want to surrender, that's fine I just leave because I don't want my time wasted. I already won.

 

I join a game and something comes up forcing me to leave the game.

 

right now, you DO lose rating for dropping...that is already in the game, I am keeping it there.

 

I dont understand why you should only lose half the points if you surrender...

Its really no difference between surrendering and losing by getting your artifact killed.

Wouldnt be a good thing to encourage ppl to surrender after 20 minutes of the game.

 

I have it set so you lose less for surrendering to prevent the hour and a half long games where several good defenders that have no hope of ever pushing out hide behind the artifact/last tower or two and drag the game out for another 40 minutes needlessly. Giving the option to surrender and prevent huge point loss will make for games that are over actually ending.

 

 

Why should a support get less rating than a carry?

 

In theory, they shouldn't. Supports often times rack up a LOT of assists, and could easily get MORE rating than a carry. Supports do get a few kills, but they also get a LOT of assists, meaning they will still be gaining rating. Furthermore, supports often bail out once the carry falls, meaning they end up with fewer deaths than the carry, meaning less points lost.

 

However, if the fear of support rating plummeting becomes a big issue, it would be easy enough to allow kills to grant only one point. I chose to say two points because honestly, getting the kill IS important.

 

Additinally, if a support is good, the win loss ratio should also be good. This will also drive up rating. I think it would all balance out.

 

 

So a tank goes 1-10 vs a carry who goes 5-5 why should they lose more points if they are playing their role correctly.

 

The fact that one assist cancels one death should defeat this issue. A tank that went 1-10 SHOULD have been doing his dying in team or group fights meaning that he should get two or more assists per death, thus equalizing or surpassing the carry he was dying to.

 

For instance, the tank in your example went 1-10 and the carry went 5-5. Now lets look at assists: The tank did his job, so you have to assume that the tank's team won the team fight at least each time the carry died. So, 5 deaths, 5 team fights lost, an average of 3 kills per team fight won --> 15 assists would sound reasonable. Thus the tank would be 1-10-15 --> 6 rating :: The carry tends to have fewer assist, since they are the ones usually doing the killing. A score of 5-5-5, also pretty reasonable, results in the carry having a rating of 10.

Now, given that you are saying the tank has done his job well, and the carry is clearly doing worse, we will assume the tanking player's team won, and for the sake of arguement lets say they won by destroying the artifact. Lets also assume this was each palyer's first game, so their rating was 0 to start.

 

Thus:

Tank: 0-->+6 from battle-->+10 from winning == 16 Rating

Carry: 0-->+10 from battle--> -10 from losing == 0 Rating

 

It is true that carries will often out perform in battle, but supports and tanks should make up for that in winning. Of course, a winning carry stands to gain a higher rating pretty fast. Even if kills were 1 point, this would still be true. Honestly, the better players play all styles, sometimes carry, sometimes support, sometimes tanks...thus their ratings should always be good. However, poorer players tend to stay with only one style, often only one hero, and this system would reflect that by giving them a lower rating.

 

Additionally, I believe this would allow us to see the difference between Good Pubbers and "Pubstars". I do not wish to cause offense, so I'll not name him directly, but there is a Shadow only player, often called a pub star, who I shall cal SM. Well, after a recent match I looked at SM's scores...he has an ok rating, a good KD with few assists, and a solid win/loss ration...the scary part was his left games: these totaled nearly 30% of his total games...thus nearly 1 out of 3 games, this pubstar raged out...With the system I proposed his rating would be severly hampered by these rage quits. I think that would be a good thing.

[Mus]

The current rating system does not provide a method to measure a players 'skill'. Your proposed system would probably be better, but I still wouldn't look at somebody's rating to see how good they are. I would still leave games as often as I do now, for the reasons I posted above. In my opinion, rating is stupid and should be removed completely.

[Ironsights]

The current rating system does not provide a method to measure a players 'skill'. Your proposed system would probably be better, but I still wouldn't look at somebody's rating to see how good they are. I would still leave games as often as I do now, for the reasons I posted above. In my opinion, rating is stupid and should be removed completely.

 

Your entitled to your opinion, and I'm entitled to mine.

 

Let's leave it at that.

[LOLLMAO]

The whole concept of rating is flawed since:

1. It does not contribute to the overall gameplay and conclusion of the game.

2. It may increase BM from pubstars as they are expected to have a "High" rating.

3. There are way too many feeders in pubgame giving a misdirection of actual skills.

4. Many carries tend to kill way more than 5 heroes and thus will probably earn high points regardless of game result.

5. Most people play for the leisure of the game, not for personal feats.

 

Then again, this rating system is probably better than the current rating system.

[Ironsights]

The whole concept of rating is flawed since:

1. It does not contribute to the overall gameplay and conclusion of the game.

2. It may increase BM from pubstars as they are expected to have a "High" rating.

3. There are way too many feeders in pubgame giving a misdirection of actual skills.

4. Many carries tend to kill way more than 5 heroes and thus will probably earn high points regardless of game result.

5. Most people play for the leisure of the game, not for personal feats.

 

Then again, this rating system is probably better than the current rating system.

 

That last bit is all I was hoping for. This is my attempt to make a simple, understandable rating system as oppossed to the one now that seems to randomly generate point gain/loss and at least for me always TAKES more than it gives, which is frustrating and leads to palyers such as myself more or less ignoring the rating.

[Mus]

What if people start demanding that you have a certain rating before you can play ih? That would really suck.

[Rem]

if that's your issue your rating is easily modified via bank files and tbh if your bad enough that your rating sucks under this system your not really wanted in an IH match anyway and noones going to pick you for their team, but a better rating system would open up a few extra paths in development potential for features so Im all for it.

[Mus]

I already play ih, my rating is below zero and yet I'm not a bad player, I am actually quite good. My issue is if people will start to stupidly use rating as a way to determine if someone is good at the game. The rating system is crap, this proposed system will be an improvement but it certainly won't be definitive.

[Midknight]

There are two main problems with the current rating system:

 

1) The ELO system designed to rate single players in 1v1 situations. You beat your opponent, you go up and he goes down. The amount depends on the match up. But when you try to apply that to a 5v5 situation, the system breaks down unless it's always the same 5 people on the team and they all share the rating.

 

2) All it gauges is winning vs losing. It does nothing to measure how good of a carry, tank, support, whatever you are.

 

Solutions:

 

1) It needs to be a system that "learns" about players like the TrueSkill rating system and rates them based on a composite of different criteria. Of course, winning or losing is most important as that's the point of the game, but also how likely is it that you'd win based on the current match-up and your history.

 

2) It needs supplemental scores to show how good you are at different roles in the game. How good of a tank were you (how much damage did you tank compared to your team, how many status affects did you soak up, etc), how good of a DPS were you (how much damage did you output compared to your team, how many kills did you secure, etc), etc. These supplemental scores could feed back into the primary score. For example, if you have a high record as DPS and you perform as expected as a DPS, your primary score wouldn't change much. If you showed improvement, your score goes up; not doing as well may make your score drop somewhat. But if you have a very low record as DPS and in one game, you perform outstanding as DPS and pull a win, your overall score would jump up significantly.

 

Summary: Any new system needs an adaptive system that produces a primary score based on how likely it was for you to win or lose the match based on the match-up and your play history and supplemental scores that show how well you did at the primary roles (tank, DPS, support).

[Qlx]

So basically this is the current rating, K/D/A and W/L in one number.

I can calculate that on my own, thanks.

 

My suggestion: Rating must adapt, and make it possible for you to do well, even If your team sucks and you're forced to play lone-wolf style.

[residente]

So basically this is the current rating, K/D/A and W/L in one number.

I can calculate that on my own, thanks.

 

My suggestion: Rating must adapt, and make it possible for you to do well, even If your team sucks and you're forced to play lone-wolf style.

 

Nono, Raphael has 24k rating with 14k kills and 7k deaths (saw it last ih), I have 13k kills with 3k deaths and my rating is 20k, and we have similar w/l ratio so i think k/d/a is not in the formula.

 

ps we both have more assist than kills

 

ps2 how ELO works on LoL? TheLuxe told me there ELO matters and some asian called TungChu or MungVu or so, and that means he's pro.

[Ironsights]

while Midnight's proposal is clearly superior, my goal was a simple and easy to implement improvement, whereas what he is suggesting seems much more complex.

 

Either alternative still feels superior to what we have now.

[residente]

while Midnight's proposal is clearly superior, my goal was a simple and easy to implement improvement, whereas what he is suggesting seems much more complex.

 

Either alternative still feels superior to what we have now.

 

Ofc it feels superior to what we have now, with all respect to devs what we have now as ELO rating seems to be a lazy bad shap, and honestly idk why they did that

[BestPlayer]

I already play ih, my rating is below zero and yet I'm not a bad player, I am actually quite good. My issue is if people will start to stupidly use rating as a way to determine if someone is good at the game. The rating system is crap, this proposed system will be an improvement but it certainly won't be definitive.

 

lol

[lOvOl]

Star Battle has a decent (though certainly not perfect) system. Just copy it.

[klonaway]

Adding to first post : your propositions are reasonable, but they should keep the purpose of the ELO rating system : comparing players one vs another.

For that purpose, just add a multiplying factor (myELO - hisELO) to most of your propositions.

 

Globally, this ends up with the present rating system, taking in account K/D/A and hopefully some important features : output, tanking, etc...

[Ironsights]

Adding to first post : your propositions are reasonable, but they should keep the purpose of the ELO rating system : comparing players one vs another.

For that purpose, just add a multiplying factor (myELO - hisELO) to most of your propositions.

 

Globally, this ends up with the present rating system, taking in account K/D/A and hopefully some important features : output, tanking, etc...

 

reasonable indeed, and possibly making the number more meaningful