16 replies to this topic

[Darian DelFord]

While I will pretense this with I think how ELO is calculated and the MM work's is complete BS and the most ironic thing ever I do need some clarification

Why is it I with over 7000 matches just in my Jenner F, my ELO is treated the exact same as a Panther 10k which I have never touched in my life and is not even basiced out?

[C E Dwyer]

Because ELO is done on weight catagory, and not for each type of mech

Simples

[DeamoBD]

I believe ELO is by weight class. Maybe someone with more insight can clearify further.

Edit: haha beat me to it

Edited by DeamoBD, 05 April 2015 - 04:38 PM.

[FupDup]

Because they didn't feel like doing it per chassis, for some reason.

[Darian DelFord]

FupDup, on 05 April 2015 - 04:38 PM, said:

Because they didn't feel like doing it per chassis, for some reason.

Exactly my point, how difficult would you think that is? And more importantly

Do you think it is needed?

[El Bandito]

Darian DelFord, on 05 April 2015 - 04:46 PM, said:

Exactly my point, how difficult would you think that is? And more importantly

Do you think it is needed?

Yes. Not only by chassis, but by variants. Cause there is a big Elo difference between my STK-5M and STK-3H.

Edited by El Bandito, 06 April 2015 - 04:34 AM.

[QuantumButler]

The secret is ELO doesn't really work, the playerbase is too small so steering wheel underhive scrubs regularly get matched with Heimdelight.

[Ace Selin]

ELO is based on weight class. Similar mech weight pilot skills are somewhat transferrable anyway.

[Zoid]

It's a 12 v 12 game, in order to get your ELO anywhere near your "true" value you have to play a TON of games (since your effect on each game is small). Breaking it up any smaller than weight class would mean the vast majority wouldn't even come close to an accurate value on anything but their very favorite 'mechs.

Not that the current system does a great job of course, but this would make it even more silly.

Edited by Zoid, 05 April 2015 - 05:20 PM.

[Roadkill]

QuantumButler, on 05 April 2015 - 05:11 PM, said:

The secret is ELO doesn't really work, the playerbase is too small so steering wheel underhive scrubs regularly get matched with Heimdelight.

Incorrect. Elo works fine, it's the matchmaker that's not working correctly. Elo ratings are just one input into the matchmaker.

A layered approach to Elo would be much better than what we have and should not require massive changes to implement. We should have an overall Elo rating, plus ratings for each weight class, each chassis, and each variant. Once you have 50 matches in Lights, you'd use the Light Weight Class Elo rating instead of the overall rating. Once you have 50 matches in Jenners, you'd use the Jenner Chassis Elo rating instead of the Light Weight Class Elo rating. Etc.

Only the most specific (i.e. variant) should be used for matchmaking purposes, but all of them should be adjusted after each match.

Then when you buy a brand new Mech, you'd have a more appropriate Elo rating to use for it until you had 50 matches. It makes no sense to buy a brand new Stalker variant and have it use an Elo rating that you partially earned piloting Awesomes. It should use your Stalker Chassis Elo rating, which will be far more accurate. Then once you have 50 matches in the new Stalker, it would start using its own rating so that it could separate from your 5M and 3H ratings.

Really pretty simple.

[Fate 6]

Yeah it doesn't really make much sense. There's no real way to basic out a new mech without tanking your Elo in the process (unless you're basicing a Timber Wolf). It's a dumb system.

[Impyrium]

It'd definitely make more sense to have it per chassis, but I'm willing to bet that's a truck load of extra data that'll significantly increase the work load the MM has...

[The Massive]

It's not any of those things.

Match maker could find 2 perfectly balanced teams. One of those teams either by someone choosing to use VOIP, or a few mechs getting caught out, or by sheer blind luck, will find themselves in a stronger position.

Once that team has taken advantage of their stronger position it very quickly becomes a 12 - 4 win for them.

Then of course everyone cries MATCHMAKER IS BROKEN!! When it's simply the flow of battle which has decided the game.

Even the pro competitive matches where all mechs are meta and all players are equal skill (as close as anyway) sometimes end 12 - 4. You gonna try and blame MM for that?

Matchmaker is fine. And MWO has to be the hardest game in the universe to calculate ELO for.

[Sean von Steinike]

You would think they could build in a modifier to lower your ELO a little with a new chassis. Not to noob level, but a little lower than usual.

[MrZakalwe]

Why does it matter? There aren't enough players in this game for an ELO system to function; I'm a scrub and get dropped into games with the best players this game has to offer

So stop worrying about it.

[Vandul]

ELO is a crutch.

[CDLord HHGD]

El Bandito, on 05 April 2015 - 05:06 PM, said:

Yes. Not only by chassis, but by variants. Cause there is a big Elo difference between my STK-5M and STK-3H.

If they differentiated Tiers by variant, then they need to differentiate ELO by variant. What they ought to do is not calculate ELO by weight, chassis, or variant. They should calculate it by what performance tier the mech falls into.....