Why does every MMO game copy the same foolish Elo system?

This thread wasn’t too salty tho. Unless you consider any complaint of any shape or form to be “salt mine worthy”

Yep - we have one - but I don’t think this is a post or thread that deserves to be in one.

whenever someone says MMO i think of…

What I think makes this problematic is that while individual skill with a hero matters, composition matters as well and often more than individual skill with heroes.

This is because of hard counters existing in the game. So limiting the range of heroes that players can play in ranked would I think be counter productive.

There are also issues about measuring individual player performance.

  • Firstly as @HipsterSkaarf mentioned different roles are more or less measurable.
  • Secondly measurable elements don’t necessarily reflect skills - take KDA for example - High tier games - especially pro games - typically have lower KDA than lower tier games (because everyone knows the game better and plays more sensibly) does this show us that high tier players are worse than low tier players?
  • Thirdly making any of these outputs “count” introduces perverse incentives - players would prolong already won matches to pad their KDA more for example.

While I agree there is something odd about taking something designed to work from an individual’s game outcome and apply it to a team based game it is worth noting that there is something odd about us playing a team based game primarily as a solitary player (side note I would love it if SEMC would split full party MMR from solo MMR).

But the main reason to keep the current system and why MOBA’s use elo or modified elo systems in general is that it works - as an epistemic truth finding device that aims to predict which players will beat which players it does a good job - a team of random T10 players will usually beat a team of random T9 players. It does this because although yes, role familarity, hero familarity, player behaviour and so on introduce noise into the system if you iterate it enough times that noise gets filtered out.


Personally 10/10 times i would take a bad comp over bad hero familiarity. Not being familiar with your hero imo is much worse because you dont know how to build for the hero and you can’t assess risks properly or time skills correctly.

Since i only play Taka, i hate Krul, Koshka, Glaive and Reim in particular but i will always still pick taka against them because im comfortable enough at avoiding what i need to avoid which can include a risky kaiten play that can cost the team the fight.

I always feel bad for the other team when i notice one player who clearly doesnt know what they are doing with their hero like ringo’s that think they are Jason Statham.

My point is, i hardly think it is fair to completely screw over someone because of mistakes someone else made. The team may not win but reward good play.

Second i hardly think it is right to allow people to run heroes they know nothing about in ranked. You may be a good player but if you dont know the hero you are using you are going to be completely off. Ive seen how much this messes with people in other games.

One game in particular actually rewards it. Thats the madness some devs are capable of.

In higher tiers people are more likely to run things that will give them as much as an advantage as possible, so there is a low chance of having comatose players. In mid to lower tiers it can be a problem. There is nothing more infuriating than getting a Saw that thinks he has no health bar.

The trueskill is worse than the Elo because it mixes casual mmr and ranked mmr. If you don’t play casual your “true” skill will be seen as very low…

We just need more stats

Absolutely incorrect.

TrueSkill is calculated based on wins and losses, as are most ranking systems whose methodology is public knowledge. If you haven’t played a certain mode, you don’t have a win or a loss, so it’s a non-factor in calculating TrueSkill.

A deep dive into the math behind TrueSkill is here.

(FWIW, VainSocial now separates TrueSkill by game mode … If you have access to a Discord server with my bot in it, you can send a!vg + your IGN to see your separate TrueSkill scores.)

You mean to say you are bashing me for recent changes… Lmao they have always been the worst API site… I never said I had no games just a few.

Basing trueskill on winrate isn’t at all possible because a win in VG gold is worth a million wins in just beginning…

Vainsocial was garbage and I stopped using it because of the nonsense like their trueskill which was out of whack…

I can’t explain why you’re wrong if you can’t understand the math. :man_shrugging:

I’m a language and history guy not math…

Each of the api sites has its own charms and uses. Vainsocial tends to do the best data/numbers work around - but their UI has historically not been the best (though I think its improving).

VGPro has a lovely UI - less useful data (especially mass data) but its well presented. They have also done an excellent job of marketing themselves especially to proplayers and by getting to provide the stats for VG8 etc.

I quite like the calmness of VGAura.

Vgminer never really clicked for me, but I get that some folks love it.

Vain University was the best but it died…

We really need to stop the copy and paste jobs. Vainglory as it is a clone, the matchmaking system is a clone. At some point there has to be someone who actually tries to improve things not ctrl + c || ctrl + v their way into peoples wallets

ELO has been used in games like Chaotic TCG (RIP) over a decade ago. It was 1v1 so it made sense. Because of team play MMORPGs and MOBAs took the system then slapped MMR on to it which is only based on wins and losses. A metric that forces 50% win rates.

If you are going to force 50% win rates then allow proper coordination in solo through voice chat so the best players are the ones that can coordinate with a team not carry a team. No that won’t happen because of toxicity right? then develop metrics that is factored into matchmaking not just wins and losses.

In a recent dev stream they confirmed that voice chat was being looked at and may or may not be be applied to vainglory. They definitely hear players but just looking for the best options as voice chat for alot of devices could reduce performance. I don’t see it coming anytime soon but I do believe it will eventually make an appearance but of course this is just an opinion.

If their implementation of voice chat is as horrible as their implementation of text chat, we’re doomed.

I’d prefer that they leave voice chat to services like Discord, which do it extremely well.


I believe Discord has or is developing an api that enables their voice chat and stuff to be embedded into games.

To me that would be the smart solution.

1 Like

How would that be easier or any different than what Discord has to offer currently? You’d need a Discord account to use their api anyway so I don’t really get it.

edit : never mind i get it

you cant force a win rate.
ths game uses elo to match Players in a way, that their elo would predict a win for each side at 50% ideally.
that of Course is never going to be a reality.
and any Kind of metrics would Need to be factored with the game that happened and the other Players in that game to be even remotely usefull. this is going to be a try and error Thing and it will very likely not improve anything.

1 Like

You should gain as much elo as you lose, matchmaker is a computing program that is fair. This might seem skewed to you because if you beat lower level players you gain less and if you lose against them you lose more elo, but overall you get matched by people averaged to your skill and will gain and lose equal amounts.

Vainglory is a team game, without any teamwork you will not to do as well. Imagine if skill tier was just based on your individual skill and not team skill. In the higher tiers, everyone would be great at 1v1’s, but when it comes to a teamfight or anything requiring skill, they will become like your teammates, not listening to pings, not helping out, running into 1v3’s etc.

Usually people don’t charge into enemies, if he did it might have been a lesser skilled player than yourself. In that case you are expected to hard carry the game, and there are many things you can do to help. E.g. if you are carry, you can try to kill the taka when he is killing your ally, if you are captain you can flare the place where taka is to show your ally not to go there, etc.

Elo does consider individual performance. It only looks at how many matches you have won vs how many you have lost, and against what skill opponents and allies. It doesn’t look at kills/assists/deaths/etc. It seems fair to me.

If you have a greater than 50% WR you should be rising in rank, and if you its lower than 50% you will be decreasing. If its lower than 50%, keep in mind that out of all the matches you have played, you are the only constant. You allies and enemy skill is approximately similiar, so it is only you that has caused this winrate.

You understand that what you just said is the definition of forced win rates right? In an extreme case lets say devs wanted to make a certain group of players likely to lose 70% of their games they just have to match them against teams with really high MMR a little more than 70% of the time and do the reverse at least 30% of the time.

In that case you would say that it is match fixing correct? Devs are trying to guarantee a win rate. if this were happening and it was proven, there would be outrage. However for some reason its not considered fixing if the MMR is even. Its instead considered “fair” matchmaking.

They literally guarantee 50% win rates for most players by evenly balancing teams by MMR. You can have a high MMR but be teamed with someone with low MMR and matched against a team where players each have average MMR.

The result is a low MMR Saw constantly running head first into an enemy team near a turret.

The only reason why win rates are in effect forced is because there is 0 communication. Without communication matches are just MMR simulations. If you simulate games where AI teams have even MMR enough times, you are going to get 50% win rates.

The whole is not always equal to the sum of its parts. Just because the MMR is even does not mean that the effect of having a bad player is negated since team MMR is even (because 1 person has really high MMR)