48 replies to this topic

[Vassago Rain]

KinLuu, on 22 February 2013 - 02:00 AM, said:

You will be a liability to your team. You will drop in Elo. Nothing you can do.

The problem here is the flawed XP system or the lack of good variants.

Yeah, pretty much. If this is the course they're going, then I want them to overhaul the XP system. I have zero reasons to grind out new mechs with ladder rankings in place.

[Ghogiel]

Vassago Rain, on 22 February 2013 - 02:37 AM, said:

Yeah, pretty much. If this is the course they're going, then I want them to overhaul the XP system. I have zero reasons to grind out new mechs with ladder rankings in place.

But you'll miss out on all the achievements and shiny stuff mang

[Diablobo]

Dukarriope, on 22 February 2013 - 02:27 AM, said:

A problem with ELO is that it is traditionally used to match people who will be using equal items. Hence why it is successful on something with two players that have identical pieces, e.g. chess.

I don't think having an Elo system be a primary matchmaker in a game where your loadout is just as meaningful is ideal.

This is the crux of the issue. Elo is solely a skill-based ranking system. It is no coincidence that it was developed and used for chess matches. In chess, all players have the same conditions and a completely balanced starting point, apart from the white first move advantage. There is practically no luck, or variance in the games. Just as the complexity of the positions rises exponentially with each move in chess, the variables in MWO such as mech loadout and teammate skill and coordination make the Elo ranking break down with each difference in the factors. The loss of precision and accuracy fall off dramatically when the other things besides skill are taken into account.

[Diablobo]

Zaptruder, on 22 February 2013 - 12:25 AM, said:

Elo is a system that ignores the noise in favour of the signal.

A thousand factors contribute towards you winning.

Those thousand factors add to the noise. The noise drowns out the signal.

[Wingbreaker]

Battle Value: A system so awesome, it's going through its third revision because it's easily exploited with mass light units.

[Lyrik]

Diablobo, on 22 February 2013 - 03:02 AM, said:

This is the crux of the issue. Elo is solely a skill-based ranking system. It is no coincidence that it was developed and used for chess matches. In chess, all players have the same conditions and a completely balanced starting point, apart from the white first move advantage. There is practically no luck, or variance in the games. Just as the complexity of the positions rises exponentially with each move in chess, the variables in MWO such as mech loadout and teammate skill and coordination make the Elo ranking break down with each difference in the factors. The loss of precision and accuracy fall off dramatically when the other things besides skill are taken into account.

,

And still Elo gets used in Magic the Gathering, Soccer, League of Legends etc

[Cybermech]

Look, that kind of detail/tracking would be nice but you got to understand.
THAT A LOT OF FINE TUNING.

ELO system is in and not finished give it some time before thinking about other ways

[Diablobo]

Wingbreaker, on 22 February 2013 - 03:57 AM, said:

Battle Value: A system so awesome, it's going through its third revision because it's easily exploited with mass light units.

That is completely irrelevant in this game. We have a set number of units per side.

[Wingbreaker]

Diablobo, on 22 February 2013 - 04:19 AM, said:

That is completely irrelevant in this game. We have a set number of units per side.

Correction: Completely relevant in this game, as the BV of a mech includes its armament, making it possible for a heavy with expensive value weaponry to easily surpass an assault with "lesser" weapons.

IE: LOLOL, 14 SMALL LASER.

Yeah, no. BV is terrible.

[bentron]

Gaan Cathal, on 22 February 2013 - 02:33 AM, said:

Apologies for the doublepost, but my phone hates the edit function on this forum.

Equipment selection contributes to your ELO score, since it affects your chancea of victory. We even have seperate ELO ratings for the weight classes. There is no need to factor in equipment twice when it comes to the matchmaker. Especially not given the abusesit would open up for minimal-to-zero benifit.

You're making two inaccurate assumptions though. First, that all mechs and variants of a given weight class are equal given equal pilot skill (which is objectively false), and second, that all pilots pilot a single mech of a given weight class such that their Elo rating in that weight class is representative not only of their pilot and gunnery skill, but of their loadout choices.

You can argue that all mechs and variants SHOULD be equal, but as the game is now, they are not. The matchmaker needs to accurately make fair matches for the game we have, not the game you wish we had.

That said, I'm not proposing BV from TT be ported directly to MWO. In fact, I think that would be a terrible idea for many of the same reasons other people do. But what equipment a pilot chooses when they hit the ready button should be taken into account when matching players against each other if the goal is to have an even fight.

For example, take two pilots of equal skill, put them in (for the sake of argument) Jenner Fs with 6 MLs, but only one of them has DHS. That is not an even match any way you look at it. The contrast is even starker if you match a Raven 3L against a 2X, 4X, etc. And then you have the Trial Mechs vs any custom mech that was built with an ounce of thought.

I may be oversimplifying how the matchmaker is supposed to work (or even be completely off), but my point is that nobody wants to be on the team with a Raven 4X and Spider 5K when the other side has two 3Ls. Especially when all 16 pilots have roughly the same Elo ranking (as is the ideal). To avoid that scenario, you have to take mech/variant and equipment into account during the matching process.

Just to reiterate, I'm not proposing BV as a system to account for equipment. I'm not even proposing there should be large differences in "equipment rating" (for lack of a better term) based on weight classes or even tonnages within a given weight class. But if the goal is to have evenly matched teams and close fought matches, you can't discount such a huge part of this game.

[Signal27]

Diablobo, on 22 February 2013 - 12:17 AM, said:

Why does PGI think that ELO matchmaking is going to work in MWO better than the Battle Value system of Battletech?

Because you can't put a Battle Value on the skills of a pilot in MWO. We're not rolling dice, we're actually aiming the damn guns.

[Joseph Mallan]

I am a long time TT player. I do not want to see BV in the video game. BV Is so broken its ridiculous! it may be contained a bit better cause of Hard Points, but it is a easily exploitable system. Battle Value would be bad.

[FrupertApricot]

BV also heavily weighs PILOT SKILL. a 3/4 pilot is worth like 30 percent more than the same machine iwth a regular pilot in tabletop lol

Make ELO part of BV. simple.

[Diablobo]

It doesn't have to be either or. It isn't really Elo versus BV like I titled the thread. Instead of Elo versus Battle Value, I was thinking that PGI should use some sort of Battle Value factor along with the Elo rankings in the matchmaking system. It won't be either or, but a combination of the two. To leave out the differences in the mechs seems to be a pretty big problem, because not all mediums are created equal, and not even all variants of the same chassis are.

[Joseph Mallan]

BV is still to easy to manipulate. I still say no to any inclusion of BV to the Video game.

[Diablobo]

Joseph Mallan, on 22 February 2013 - 05:47 AM, said:

BV is still to easy to manipulate. I still say no to any inclusion of BV to the Video game.

There is no way to avoid it and have any reasonably fair matchmaking. Given two pilots of equal Elo rating, the one in the better mech is going to win more often than would be reflected by their ratings. Instead of the same rated players having a 50/50 shot at victory, the one in the better mech is going to win more than that. That will skew the ratings. Skewed ratings equals bad matchmaking. Bad matchmaking equals a bad time.

[Crawfield]

Anyone know why they didnt took Glicko-Rating?
It is an improvement of Elo.

Edited by Crawfield, 22 February 2013 - 06:13 AM.

[Gaan Cathal]

bentron, on 22 February 2013 - 04:39 AM, said:

You're making two inaccurate assumptions though. First, that all mechs and variants of a given weight class are equal given equal pilot skill (which is objectively false), and second, that all pilots pilot a single mech of a given weight class such that their Elo rating in that weight class is representative not only of their pilot and gunnery skill, but of their loadout choices.

I can see how you might get that, didn't clarify particularly well. I'm not saying that pilot ELO is representative of a given loadout choice, I'm saying it is representative of a player's loadout choices. A player of ELO score X will no doubt be better when he drops in his Catapult than his Dragon, but his actual ELO will reflect the odds of you facing him in either (in this two-mech example, it would be an average based more or less on number of matches in each mech). If he decides to play in his Catapult for a week his ELO will probably trend up slightly, if he decides to play in a Dragon it will probably trend down. However your exact ELO doesn't actually matter, since the bracket the matchmaker draws opponents from is fairly wide (and gets wider as the queue gets longer), and unless the aforementioned pilot takes to running a 4SLAS, 2MG K2 for a week he'll probably continue to see the same opponents. A pilot who consistently plays higher-end loadouts will, over time, accrue more wins and ergo have a higher ELO. Working as intended, essentially.

bentron, on 22 February 2013 - 04:39 AM, said:

I may be oversimplifying how the matchmaker is supposed to work (or even be completely off), but my point is that nobody wants to be on the team with a Raven 4X and Spider 5K when the other side has two 3Ls. Especially when all 16 pilots have roughly the same Elo ranking (as is the ideal). To avoid that scenario, you have to take mech/variant and equipment into account during the matching process.

Presuming you wished to (see above)...how? It's impossible to do fairly from outside the system. Tonnage doesn't work, since the 4X and 3L weigh the same, while the former is manifestly inferior, even with both pimped out (hell, even with the 3L not taking ECM). So if actual tonnage doesn't work, we just assign a 'BV' to chassis/equipment, right? The problem is, we don't, PGI do - and game designers are never particularly good at admitting when certain of their designs are crap. Even if we did, we'd not agree - see any other the umpteen weapon balance threads. That imperfection would leave the system open to gaming. It would also increase the viability of boating, since 6 of weapon X is generally superior to 3 of X, 2 of Y and 1 of Z. Relying on the summed-into-ELO effect might not be perfect, but an administrated alternative is worse. The point of ELO is that it is purely win/loss dependent and ergo takes into account all factors (admittedly with poor resolution, but that's the price).

Diablobo, on 22 February 2013 - 05:55 AM, said:

Given two pilots of equal Elo rating, the one in the better mech is going to win more often than would be reflected by their ratings.

And therefore have a higher ELO.

Edited by Gaan Cathal, 22 February 2013 - 08:31 AM.

[borisof007]

Crawfield, on 22 February 2013 - 06:09 AM, said:

Anyone know why they didnt took Glicko-Rating?
It is an improvement of Elo.

Jesus, someone's first post was in a thread regarding Elo and Battle Value, nice.

I honestly think that if equipment is going to play a huge factor in battles, then Battle Value AND Elo Rating need to both be implemented. Battle Value will determine what you can drop with (e.g. a battle value limit on drops across the 8 man premades, or battle value range requirements), Elo will determine who you fight against (Once your team's set to drop, MM searches for teams of equivalent skill).

The only problem I have is that this would be a little harder to coordinate for solo/4 mans, unless it takes ranges of battle values into account.

I understand that you're involving two systems instead of one, but I don't think Elo alone is a good enough solution.

[jakucha]

Trying to use a balancing system used in TT, in MWO, is a silly idea when you consider how comparing pretty much any TT weapon stat and so on just doesn't work.