16 replies to this topic

[DeathlyEyes]

It was discussed in the town hall meeting that PGI was considering a new matchmaker that used a new system to determine "skill" and factoring in some sort of battlevalue system. I am fully on board with this.

In the current system, a high Elo player could drop in a DRG-1C. The counterweight high Elo player on the other team could be dropping in an Ebon Jaguar or a Timberwolf. Now in no way are these 2 chassis remotely similar in capability. This means anyone in a High Elo mech has to pretty much always take high powered meta mechs, rather than enjoying his/her mechs.

The new system Russ proposed can sort of fix the solution. Both teams battle values will be equal but the High Elo player could still be dropping in that Dragon while some low tier player is using an Ebon Jaguar. What I propose is each mech gets a battle value still. Then it gets combined in some sort of function say (Skill Level)(Battle Value)=Total Combat Value. This means if the high tier player drops in a locust he/she wouldn't be expected to carry a team against team lead by a high tier Direwolf pilot.

[Soulscour]

As a late night east coast player, I've found the matchmaker to be acceptable during that time period. Early in the morning however, the player pool seems to get smaller, and matchmaker suffers dramatically as there's not enough good/average players to fill up a game. Regardless of how they evaluate elo, if the player pool is too small, it will not work.

[Medi0cr3]

I am Medi0cr3 and I approve this message!

[PhoenixFire55]

DeathlyEyes, on 25 June 2015 - 11:17 PM, said:

It was discussed in the town hall meeting that PGI was considering a new matchmaker that used a new system to determine "skill" and factoring in some sort of battlevalue system. I am fully on board with this.

Well it can't get any worse than what we have now, because what we have now isn't any different in terms of MM quality than a completely random MM we had back in early days.

DeathlyEyes, on 25 June 2015 - 11:17 PM, said:

In the current system, a high Elo player could drop in a DRG-1C. The counterweight high Elo player on the other team could be dropping in an Ebon Jaguar or a Timberwolf. Now in no way are these 2 chassis remotely similar in capability. This means anyone in a High Elo mech has to pretty much always take high powered meta mechs, rather than enjoying his/her mechs.

The new system Russ proposed can sort of fix the solution. Both teams battle values will be equal but the High Elo player could still be dropping in that Dragon while some low tier player is using an Ebon Jaguar.

Yeah, thats what I've been saying for almost two years now. There were like hundreds of nearly identical suggestions to do a simple MM change and assign Elo scores to each variant of a mech chassis, instead of only classes of mechs. All happily ignored.

DeathlyEyes, on 25 June 2015 - 11:17 PM, said:

What I propose is each mech gets a battle value still. Then it gets combined in some sort of function say (Skill Level)(Battle Value)=Total Combat Value. This means if the high tier player drops in a locust he/she wouldn't be expected to carry a team against team lead by a high tier Direwolf pilot.

The problem is, who is going to assign BV to mechs? And depending on what? It raises about as many balancing problems as it actually solves. For example, obviousely, a brawler Atlas is a very efficient mech for a map like RiverCity, same way it is entirely inefficient mech for Alpine. If you assign a high BV to it the imbalance will occur on Alpine, if you assign a low BV it'll occur on RC, if you go for a compromise it'll be imbalanced all the time ...

[Joseph Mallan]

Should be interesting to try out.

[Duke Nedo]

Interesting stuff! Even if BV adds new balancing problems and will never be perfect, it's a sharper tool that should make BV better than no BV.

[Speedy Plysitkos]

new MM , that include loss as +gain points, = good idea. like it.

[DeathlyEyes]

PhoenixFire55, on 26 June 2015 - 12:37 AM, said:

The problem is, who is going to assign BV to mechs? And depending on what? It raises about as many balancing problems as it actually solves. For example, obviousely, a brawler Atlas is a very efficient mech for a map like RiverCity, same way it is entirely inefficient mech for Alpine. If you assign a high BV to it the imbalance will occur on Alpine, if you assign a low BV it'll occur on RC, if you go for a compromise it'll be imbalanced all the time ...

Battlevalue will have to be created based off what the top tier Elo players are successful with. It's going to have to take a commited developer involved constantly changing BVs. Certain combinations could lead to a higher BV. High weapon mounts+ certain weapons = higher bv.

[EgoSlayer]

DeathlyEyes, on 25 June 2015 - 11:17 PM, said:

It was discussed in the town hall meeting that PGI was considering a new matchmaker that used a new system to determine "skill" and factoring in some sort of battlevalue system. I am fully on board with this.

In the current system, a high Elo player could drop in a DRG-1C. The counterweight high Elo player on the other team could be dropping in an Ebon Jaguar or a Timberwolf. Now in no way are these 2 chassis remotely similar in capability. This means anyone in a High Elo mech has to pretty much always take high powered meta mechs, rather than enjoying his/her mechs.

The new system Russ proposed can sort of fix the solution. Both teams battle values will be equal but the High Elo player could still be dropping in that Dragon while some low tier player is using an Ebon Jaguar. What I propose is each mech gets a battle value still. Then it gets combined in some sort of function say (Skill Level)(Battle Value)=Total Combat Value. This means if the high tier player drops in a locust he/she wouldn't be expected to carry a team against team lead by a high tier Direwolf pilot.

Russ specifically said in the town hall that the MWO "battle value" system would only be used for the new quirk system and would *not* be used in the match maker. Adding that to the match maker, or coming up with some Skill times value system over complicates the match maker that already struggles with the 3 (or 4 if you consider time) criteria it has now (Game mode, elo, chassis - and Time which triggers the release valves).

Edited by EgoSlayer, 27 June 2015 - 03:43 PM.

[LordKnightFandragon]

Ill take a MM that actually is skill based and not skill rigged, like WoT, where if you start to win it starts to drop you in progressively worse and worse teams till you achieve abouta 50% win rate......

[KodiakGW]

Anything, including turning it off, is better than what we have now. Tired of waiting for matches to see 5-8 player on my team doing sub 200 Damage and/or Match Score.

[Nik Reaper]

EgoSlayer, on 27 June 2015 - 03:38 PM, said:

Russ specifically said in the town hall that the MWO "battle value" system would only be used for the new quirk system and would *not* be used in the match maker. Adding that to the match maker, or coming up with some Skill times value system over complicates the match maker that already struggles with the 3 (or 4 if you consider time) criteria it has now (Game mode, elo, chassis - and Time which triggers the release valves).

This is pretty how I heard it , so no the only new change to MM Russ said is one that looks at you'r "effect" in the match.
So if you play well but lose you don't lose ranking score and if you play lousy but win your ranking doesn't go up so, nothing new for mech value but still a better system that he said would create buckets to protect lower tiers from the top as the score will more quickly change and more accurately ( though still no account on mech and mech build or map selected ) .

[DeathlyEyes]

Well it needs to start taking into an account that a Pretty Baby isn't a Direwolf and an orion isn't a Timberwolf. The way it works now is totally unaccceptable. If I wanted to play an Orion now it would be impossible. No quirks short of giving it +300 armor or some sillyness will actually make it as powerful as a Timberwolf.

Edited by DeathlyEyes, 28 June 2015 - 05:45 PM.

[Goose]

What it needs is an Elo-per-'Mech, a K-Factor linked to score, and a new scoring system that more closely resembles the pay-outs, the "recent" ones that value Lights'n skirmishers.

[kapusta11]

The only thing that feels disturbing is Russ saying that if you performed well but lost a game you may still gain points. This means that numerical evaluation of skill is dynamic while in reality it doesn't change much, at least not in a short period of time. Getting 5+ kills and doing 1000+ damage is not a result of player performing well, it's the result of poor matchmaking putting a veteran/s against novice players. In balanced matchups pulling your own weight alone is supposed to be a challenge.

Edited by kapusta11, 29 June 2015 - 02:49 AM.

[Goose]

Any system rolled out would work better with a larger sample size; Being so accurate things stratify would also be bad …

[Revis Volek]

DeathlyEyes, on 28 June 2015 - 05:44 PM, said:

Well it needs to start taking into an account that a Pretty Baby isn't a Direwolf and an orion isn't a Timberwolf. The way it works now is totally unaccceptable. If I wanted to play an Orion now it would be impossible. No quirks short of giving it +300 armor or some sillyness will actually make it as powerful as a Timberwolf.

I agree we need a little more...


Why couldn't we just begin with a base value (Zero) If your mech has i dunno....some criteria (3 or more positive quirks, 3 or more negative quirks) and say you take a TBR....you get a +1 to your ranking in MM where as the Onion gets a -1 to his rank im the MM since he/she needs helps potentially with the Chassis he/she has or the loadout vs other mechs in the class.

I dont see that being very hard....simple number system so the MM can SEE mech for what they are, not just as a class.