24 replies to this topic

[Appogee]

My lancemates and I have been debating how we think Elo has been applied to MWO.

Can anyone provide a link to a definitive statement from the devs on this subject?

My belief is that the Matchmaker assembles two teams of about the same average Elo. Each player's Elo may be only modified based on the outcome of a match, based on the following rules:

• If the team with the ''best" average Elo wins, then no-one's Elo is modified.
  
• However, if the team with the ''worst" average Elo wins, then the Elo of each member of the winning team is increased, and the Elo of each member of the losing team is decreased.

So: individual scores, k/ds etc do not modify a player's Elo. Only the outcomes of the matches a player plays in will affect a player's Elo.


I'm looking for a definitive statement from the devs on this, please. I'm sure they've explained it somewhere previously.

[Roadbeer]

Magic space cats with pizza. That's how Elo and the Matchmaker work in MWO.

I can't find a better explanation, even from the Devs.

[BlackBeltJones]

This should be pinned

[-Natural Selection-]

Have you ever dropped into a single match and the groups made sense? MM is a myth, just a good thing to blame bad rounds on.

[Strum Wealh]

Appogee, on 20 April 2014 - 11:04 AM, said:

My lancemates and I have been debating how we think Elo has been applied to MWO.

Can anyone provide a link to a definitive statement from the devs on this subject?

My belief is that the Matchmaker assembles two teams of about the same average Elo. Each player's Elo may be only modified based on the outcome of a match, based on the following rules:

• If the team with the ''best" average Elo wins, then no-one's Elo is modified.
  
• However, if the team with the ''worst" average Elo wins, then the Elo of each member of the winning team is increased, and the Elo of each member of the losing team is decreased.

So: individual scores, k/ds etc do not modify a player's Elo. Only the outcomes of the matches a player plays in will affect a player's Elo.

I'm looking for a definitive statement from the devs on this, please. I'm sure they've explained it somewhere previously.

The post in question can be found here.

Also, this post may be of interest.

[Appogee]

Strum Wealh, on 20 April 2014 - 11:55 AM, said:

The post in question can be found here.

Also, this post may be of interest.

Thanks Strum Wealh, much appreciated.

For those who want a summary:

• At the start of a match, each of the two team's average Elo rating is calculated (ie the sum of each of the team's players' Elo ratings, divided by 12).
  
• If the team with the higher average Elo wins, then no player's Elo is adjusted.
  
• However, if the team with the lower average Elo wins, then the players on the winning team have their Elos adjusted up, and the players on the losing team have their Elos adjusted down.
  
• The amount of Elo adjustment is in proportion to the gap between the two team averages. That is, if a team which has a much lower average Elo beats a team with a much higher average Elo, then each player's Elo will be adjusted by a greater amount ... to a maximum adjustment of +/- 50.
  
• Your individual performance - your score, the number of kills or assists you personally made - make zero difference to your Elo rating. Only your team's performance can result in an adjustment to your Elo.

Edited by Appogee, 20 April 2014 - 12:20 PM.

[EgoSlayer]

The elo math is here:
http://mwomercs.com/...65#entry1626065

Note that this a a team averaged elo game, so your change is based on what your team average is vs their team average.

[Davers]

Appogee, on 20 April 2014 - 11:04 AM, said:

My lancemates and I have been debating how we think Elo has been applied to MWO.

Can anyone provide a link to a definitive statement from the devs on this subject?

My belief is that the Matchmaker assembles two teams of about the same average Elo. Each player's Elo may be only modified based on the outcome of a match, based on the following rules:

• If the team with the ''best" average Elo wins, then no-one's Elo is modified.
  
• However, if the team with the ''worst" average Elo wins, then the Elo of each member of the winning team is increased, and the Elo of each member of the losing team is decreased.

So: individual scores, k/ds etc do not modify a player's Elo. Only the outcomes of the matches a player plays in will affect a player's Elo.



I'm looking for a definitive statement from the devs on this, please. I'm sure they've explained it somewhere previously.

This is true, but Elo is not the only metric the MM uses.

[-Muta-]

Well that is a lot of math... I think I will forget all I have read about ELO and keep on having fun... At the end of the day understanding how it works will not make me a better/worse pilot nor bring more/less fun.

So I see you at battlefield!

[Lyoto Machida]

Roadbeer, on 20 April 2014 - 11:28 AM, said:

Magic space cats with pizza. That's how Elo and the Matchmaker work in MWO.

I can't find a better explanation, even from the Devs.

[Trauglodyte]

Quote

Thanks Strum Wealh, much appreciated.

For those who want a summary:

• At the start of a match, each of the two team's average Elo rating is calculated (ie the sum of each of the team's players' Elo ratings, divided by 12).
  
• If the team with the higher average Elo wins, then no player's Elo is adjusted.
  
• However, if the team with the lower average Elo wins, then the players on the winning team have their Elos adjusted up, and the players on the losing team have their Elos adjusted down.
  
• The amount of Elo adjustment is in proportion to the gap between the two team averages. That is, if a team which has a much lower average Elo beats a team with a much higher average Elo, then each player's Elo will be adjusted by a greater amount ... to a maximum adjustment of +/- 50.
  
• Your individual performance - your score, the number of kills or assists you personally made - make zero difference to your Elo rating. Only your team's performance can result in an adjustment to your Elo.
  

This is also why so many people incorrectly say that the ELO chooses a team to win and the other team to lose. That isn't the case at all.

Btw, the one thing that makes me sad is that efficiency level and tech tiers aren't factored in as an ELO factor. Player A in a brand new mech vs. player B in that same mech but with all of the bells and whistles isn't fair. What's more, if both Player A and Player B end the match with the same damage, number of kills, component destructions, assists, etc, Player A should be rewarded more because they're having to do so much more with less. Unfortunately, both players finish the game with the same rewards and we all go on smiling.

[Corbon Zackery]

It a huge secret however,

Basically as I remember it they brought in a outside guy to write up the ELO program. Probably due to time restraints.

This person muffed the poodle so they canceled his contract and tried to fix it themselves with disastrous results.

As far as I know since they deleted all the old closed and open beta post. There is a bug in the program that says your ELO is: 500

ELO will match you with any player from 0 -1000

What it doesn't do is match your ELO to weight tonnage. So your team could have a 35 ton Jenner who's ELO is 750

Enemy team is going to have a 700 - 800 guy but he might be in a atlas.

Now your saying well why do I get 30 losses in a row. Or why do I get Terra Thurma 5 times in a row. That's because ELO looks at your overall stats and says wow your over 1.00 at 1.96.

So it set you as the highest ELO will say 1000: Now its going to find other players 0-1000 this number changes every two weeks as the developers tighten and loosen it so your not searching for 3-5min for a match. If that's happens ELO just dumps you into one of the match buckets. That's been happening to some 4 man teams

If they tighten the number lets say 1000-500 It will try to search for any players 1000-500 then after 3 min dump any other people into the match it can find.

If they loosen the number players will be found faster but teams degrade.

Its intense, it doesn't work at all. However I have seen worse matchmakers on sites like Neverwinter and Star trek online. Were your dropped into PVP vs. Premade characters and bots.

Bots are the other issue. While there is no way to prove it. We think that MechWarrior online offsets lower server numbers with bots. Also the developers are always modifying the game. They can make hand of god changes at any time. I am not surprised if they dropped some of us into test beds for Clan weapons. Once again no way to prove that. This is all based on observation in spectate mode. Cross hairs bouncing around. Near perfect shots. CT coring extreme long range shots.



This is also why you get tonnage mismatches. ELO is pulling mechs based off pilot skill not both skill and tonnage.

Edited by Corbon Zackery, 20 April 2014 - 08:50 PM.

[Corwin Vickers]

That is incorrect.

You have seperate Elo scores for each weight class so it does take weight into account.

If you are crap on your light mechs your Elo score for that weight class will be low. If you are awesome with assaults then your Elo for that weight class will be high (assuming you are a team player.)

If you score 11 kills every game and do 10,000 damage but you lose every match because you aren't a team player and don't care about winning you can have a low Elo which also helps you in getting those 11 kills and 10,000 damage because the people you are fighting against are suffering from strokes.

[Jeffrey Wilder]

Appogee, on 20 April 2014 - 11:04 AM, said:

My lancemates and I have been debating how we think Elo has been applied to MWO.

Can anyone provide a link to a definitive statement from the devs on this subject?

My belief is that the Matchmaker assembles two teams of about the same average Elo. Each player's Elo may be only modified based on the outcome of a match, based on the following rules:

• If the team with the ''best" average Elo wins, then no-one's Elo is modified.
  
• However, if the team with the ''worst" average Elo wins, then the Elo of each member of the winning team is increased, and the Elo of each member of the losing team is decreased.

So: individual scores, k/ds etc do not modify a player's Elo. Only the outcomes of the matches a player plays in will affect a player's Elo.



I'm looking for a definitive statement from the devs on this, please. I'm sure they've explained it somewhere previously.

It's very simple. Once you hit your best ELO, you will get a win for every 6 to 9 losses.

[Sean von Steinike]

Blue swan jumps spaceship kangaroo because potato yield sign four and half milk. <-----kinda like that

[RedDragon]

Appogee, on 20 April 2014 - 11:56 AM, said:

Thanks Strum Wealh, much appreciated.

For those who want a summary:

• If the team with the higher average Elo wins, then no player's Elo is adjusted.
  
• However, if the team with the lower average Elo wins, then the players on the winning team have their Elos adjusted up, and the players on the losing team have their Elos adjusted down.

Wait a minute... since we all started with the same Elo, how could the system have started to work in the first place? When every player has an Elo of 500, the team's mean Elo will be 500, so no matter if they win or lose to another team of 500, there won't be an adjustment... so, if they calculate Elo in relation to Elo and not based on the sole number of your w/l-ratio, how the hell does that work?
(I wouldn't by surprised if no one at PGI caught this and we were playing with our start-Elo ever since )

[ImperialKnight]

when you win, your ELO goes up, when you lose, your ELO goes down. when you win too much, the matchmaker smacks you down by putting you in a team full of low ELO players. when you lose too much, the matchmaker tries to give you good teammates

[Zerberus]

RedDragon, on 21 April 2014 - 02:05 AM, said:

Wait a minute... since we all started with the same Elo, how could the system have started to work in the first place? When every player has an Elo of 500, the team's mean Elo will be 500, so no matter if they win or lose to another team of 500, there won't be an adjustment... so, if they calculate Elo in relation to Elo and not based on the sole number of your w/l-ratio, how the hell does that work?
(I wouldn't by surprised if no one at PGI caught this and we were playing with our start-Elo ever since )

All it takes is one new player with an artificially lowered Elo score (which new players have) to start the chain reaction and get the ball rolling for everyone

[RedDragon]

Zerberus, on 21 April 2014 - 02:58 AM, said:

All it takes is one new player with an artificially lowered Elo score (which new players have) to start the chain reaction and get the ball rolling for everyone

But we were all new players at one point, regarding Elo. So if the Devs did not decide that some of us should have lower (or higher) Elo scores as others, there would be no progress at all...

[Eglar]

What I would find interessting is the average amount Elo delta with each win/loss.

Let's assume that you get an average Delta of 15 Elo (with 50 max.) after each match. So basically. you can move up 100 elo after roughly 6-7 consecutive wins. Shouldn't be anyone who can keep up a > 1 Win-Loss Ratio be reaching or be near 2800 cap sooner or later?

So if a player feels that he lost the game after a few consecutive wins, isn't it simply because he moved up the elo ladder rather than the matchmaker trolling him? If that same player fails to carry his current elo, he'll naturally lose and drop down.

#publiceloplz

Edited by Eglar, 21 April 2014 - 04:50 AM.