7 replies to this topic

[Pjwned]

Quote

The only relation between Elo and a BattleMech is that the player has 4 WEIGHT CLASS Elo scores. This just means that you might be great in a light, but not so great in a heavy.

Am I wrong or does this mean that a player with an excellent ELO score in a light will be put in with the newest players if they decide to change things up and play a heavy instead? This quote makes it seem like there isn't any kind of general ELO score even though a player with a high ELO in another weight class is clearly going to be more experienced than somebody who just started playing.

I do agree with having ELO scores for each weight class but if that's the only consideration when matching people up that seems like a poor idea.

There's also the issue of how ELO is calculated, but I thought I would save that argument for other threads. For the record though, if the matchmaker has a low average discrepancy, then to me that calls into question how ELO is calculated for each player.

I realize that there are other threads to post about ELO but this particular issue stuck out to me and I thought it deserved its own thread.

Edited by Pjwned, 22 January 2014 - 04:02 PM.

[PanzerMagier]

There isn't. People's skills are radically different in various weight class mechs.
You have completely separate elo for light, medium, heavy, assault.

If you've only played lights, you'll start with a fresh elo in the other classes.
Just keep in mind if you perform well, it will only take 3-10 matches for your elo to adjust accordingly.
So don't think it'll be easy mode from there on out.

How elo is exactly calculated, we'll never know cuz PGI will never tell the truth.
From estimations it's based off of your K/D W/L ratios in every waight class.

Edited by PanzerMagier, 23 January 2014 - 06:03 AM.

[anubis969]

PanzerMagier, on 23 January 2014 - 06:01 AM, said:

How elo is exactly calculated, we'll never know cuz PGI will never tell the truth.
From estimations it's based off of your K/D W/L ratios in every waight class.

Actually they did tell us exactly how it's calculated:

http://mwomercs.com/...65#entry1626065

Neither your K/D nor your W/L ratio plays any part in how the game calculates your Elo score.

[Roadkill]

anubis969, on 23 January 2014 - 06:52 AM, said:

Neither your K/D nor your W/L ratio plays any part in how the game calculates your Elo score.

That's slightly misleading. While your W/L ratio is not used to calculate your Elo rating, each of your wins and losses is used in the calculation at the time that it occurs.

[anubis969]

Roadkill, on 23 January 2014 - 12:56 PM, said:

That's slightly misleading. While your W/L ratio is not used to calculate your Elo rating, each of your wins and losses is used in the calculation at the time that it occurs.

There is nothing misleading to what I said at all. I said that the game does not use your W/L ratio to calculate your Elo and it doesn't.

In fact your comment isn't entirely accurate because technically speaking the game does not look at each of your wins and losses whenever it calculates your Elo like you suggest. Yes your Elo score comes from your past game results but they are not used directly in the calculation.

You start the game with a slightly below average Elo score, from there the game adjusts your Elo score at the end of each match depending on which team won, the disparity between your teams average Elo score and the opponents, and whether or not the MM predicted the outcome of the match correctly. Your new Elo score is then saved and used for the next match. As such the game does not look at your past wins and losses when it calculates your new Elo score, it just looks at your old Elo score.

That said now I'm just being pedantic but then it was you that didn't want people to get mislead .

[Warblood]

ELO = Win/Loss raito per weight class (start:38min)
http://soundcloud.co...-heat-scale-elo

[anubis969]

Warblood, on 23 January 2014 - 05:25 PM, said:

ELO = Win/Loss raito per weight class (start:38min)
http://soundcloud.co...-heat-scale-elo

Check 40:20. They point out how Elo is actually calculated which contradicts their previous assertion that Elo is based on W/L ratio. As said previously in this thread, Elo is based on your wins and losses but not on the ratio.

By the way, sorry to Pjwned as your thread seems to have gotten derailed and I fear that I am partly to blame. I only posted to correct some misinformation, I never intended for this to turn into a discussion .

Edit: In an effort to get things back on track; while I can see your point Pjwned regarding all players starting at the same Elo in a given weight class regardless of experience, when tonnage limits come in we will almost certainly have a single general Elo score . To my knowledge they haven't said one way or t'other but I can't see any other way for it to work. The reason I say this is because when tonnage limits are in place you will be choosing your 'mech after the MM has found you a team rather than before, like it is now.

Edited by anubis969, 23 January 2014 - 06:12 PM.

[Mawai]

ELO is not based on W/L ratio.

The process as PGI has outlined is the following:

The matchmaker takes an "average" ELO for both teams and then uses this to predict which team will win the match.

1) If the MM prediction is correct then NO ONE's ELO is changed. Neither the winning nor losing side.

2) If the MM prediction is incorrect then the difference in the team average ELO values is used to calculate an ELO adjustment applied to each player on each team. The winning team ELO is increased and the losing team is decreased. The maximum change allowed in one match is capped (I think the number mentioned was +/- 50 points).

That's it. Often whether you won or lost will have no effect on your ELO. On the other hand. if you won every game where your side is the underdog you would have a very high ELO. In fact, the only way to really get a high ELO is to beat teams with a higher average ELO ... which means that you are dropping on a team either solo or part of a group ... your ELO is substantially higher than your team mates so that the overall average is lower than the opposing team ... you win and your ELO goes up. This is somewhat indirect evidence that the ELO spread between players on a team can be quite high.

In any case, ELO has nothing to do with your W:L ratio ... if every win you had was predicted and every loss not then you would have an ELO near zero with a 1:1 ratio. Similarly, if every loss was predicted but every win was not your ELO would be close to 2800 with a 1:1 ratio. So ... although ELO is based on wins and losses ... it has nothing to do with your number of wins or losses or their ratio.

Edited by Mawai, 23 January 2014 - 06:02 PM.