38 replies to this topic

[New Day]

If ELO works why do I keep seeing total newbs (trial and overheat) almost every game. Seems that something like this should be prevented by ELO, no?

[Syllogy]

Nope.

Since the vast majority of players fall near the same average elo rating, you are by far more likely to see noobs on your team rather than elite players.

Light Blue = 50 Matches or Less
Dark Blue = 51 Matches or More

Edited by Syllogy, 09 July 2013 - 07:39 AM.

[Stoicblitzer]

you're on an island.

[New Day]

Syllogy, on 09 July 2013 - 07:37 AM, said:

Nope.

Since the vast majority of players fall near the same average elo rating, you are by far more likely to see noobs on your team rather than elite players.

Light Blue = 50 Matches or Less
Dark Blue = 51 Matches or More

Couldn't this be solved by grouping sub 200/300 matches players together and +300 together?

[Angel of Annihilation]

ELO has never worked because you can't attach a mathematical formula to skill. I mean who has more skill, the light mech that keeps the enemy team completely off balance because it is always threatening the enemy base but does little damage or the Assault that blasts away at the enemy mech's backs who turn away to chase down the light mech threatening their base?

[AnnoyingCat]

what is this ELO you speak of?

[Kazly]

Viktor Drake, on 09 July 2013 - 09:16 AM, said:

ELO has never worked because you can't attach a mathematical formula to skill. I mean who has more skill, the light mech that keeps the enemy team completely off balance because it is always threatening the enemy base but does little damage or the Assault that blasts away at the enemy mech's backs who turn away to chase down the light mech threatening their base?

I agree with this 100%, but also think that Elo can/will work with their current formula, but for CW/8mans that drop with the same folks over and over. At this point, your team is the same folks, under the same leadership folks. But basing a win/loss on 15 other random variables will never properly work.

[Bilbo]

Viktor Drake, on 09 July 2013 - 09:16 AM, said:

ELO has never worked because you can't attach a mathematical formula to skill. I mean who has more skill, the light mech that keeps the enemy team completely off balance because it is always threatening the enemy base but does little damage or the Assault that blasts away at the enemy mech's backs who turn away to chase down the light mech threatening their base?

If they are doing it right, they are both equally skilled and the result is a win which is then reflected by your ELO.

[New Day]

Viktor Drake, on 09 July 2013 - 09:16 AM, said:

ELO has never worked because you can't attach a mathematical formula to skill. I mean who has more skill, the light mech that keeps the enemy team completely off balance because it is always threatening the enemy base but does little damage or the Assault that blasts away at the enemy mech's backs who turn away to chase down the light mech threatening their base?

Well ELO=skill. Problem is this is a team game, and that not enough factors are valued (eg base capping) for the match score

[Braggart]

Its broken. ELO was not designed for what it is currently being used for.

There are far to many factors that are not related to your personal skill that effect the outcome of the game. Each and every game.

Edited by Braggart, 09 July 2013 - 09:37 AM.

[Gen Kumon]

Not to mention Elo doesn't actually measure your personal skill. It could be said to ROUGHLY measure the aggregate skill of every team you've ever played on...but even that's not really true, since there's a lot of factors outside of personal and team skill that can effect win/loss.

So yeah, Elo doesn't actually mean anything more than your win/loss ratio does.

[General Taskeen]

Looks more like a volcanoe than an island. So an island forming?

[Para B]

I have to say I have no idea how this ELO system is supposed to work. I've seen within the same match vet players who surgically dismembered enemy mechs and nebiews who killed themselves running off the map.

I still see 8:0 PUGstomps and I still see pitched fights with even teams. I see PPC boats and lolfits. I see people who can't lead a target and see players who leg lights with a single salvo.

Can't really say I've noticed a big change in the game experience since ELO was introduced.

[Fishhawk]

The ELO is a lie. The cake is real.

[Acid Phase]

OP, you seem to forget that ELO.....

[IceSerpent]

Viktor Drake, on 09 July 2013 - 09:16 AM, said:

ELO has never worked because you can't attach a mathematical formula to skill. I mean who has more skill, the light mech that keeps the enemy team completely off balance because it is always threatening the enemy base but does little damage or the Assault that blasts away at the enemy mech's backs who turn away to chase down the light mech threatening their base?

Given that Elo rating is based on wins/losses, either one of those mechs can have their rating accurately reflect their skill. Using a mathematical model to represent skill is not only possible, it's fairly easy in most cases, as there's a direct correlation between skill level and result of applying that skill. Better results indicate better skills and result in higher rating.

[Roland]

The bigger issue isn't simply that Elo doesn't work, but rather that the playerbase isn't big enough to really support it. There aren't enough folks available to play when you hit the launch button, to make a team of equally skilled pilots.

[xDeityx]

Syllogy, on 09 July 2013 - 07:37 AM, said:

Nope.

Since the vast majority of players fall near the same average elo rating, you are by far more likely to see noobs on your team rather than elite players.

Light Blue = 50 Matches or Less
Dark Blue = 51 Matches or More

Which data point am I on this graph? jkjk

Real men label their X axes on graphs though.

Edit: I wonder how this graph accounts for the four different Elo ratings that each player has. Does the light blue include the <50 games that a player has in a certain 'mech weight category? PGI is so opaque with their information.

Edited by xDeityx, 09 July 2013 - 10:05 AM.

[Braggart]

IceSerpent, on 09 July 2013 - 09:48 AM, said:

Given that Elo rating is based on wins/losses, either one of those mechs can have their rating accurately reflect their skill. Using a mathematical model to represent skill is not only possible, it's fairly easy in most cases, as there's a direct correlation between skill level and result of applying that skill. Better results indicate better skills and result in higher rating.

It is possible, but....................................not when the other 7 people have more of an effect on whether you win or lose the game 95% of the time.

ELO is great when it is on a 1vs1 or anytime teams never change game after game. But when every game features a different mix of teammates and such. ELO is not a measure of skill at all.

[IceSerpent]

Braggart, on 09 July 2013 - 02:05 PM, said:

It is possible, but....................................not when the other 7 people have more of an effect on whether you win or lose the game 95% of the time.

ELO is great when it is on a 1vs1 or anytime teams never change game after game. But when every game features a different mix of teammates and such. ELO is not a measure of skill at all.

You pretty much nailed it - problem is not with Elo rating itself or applying it to the team game, problem is that PGI (in their infinite wisdom) decided to put players with vastly different Elo ratings on the same team, thus completely defeating the whole purpose of having Elo to begin with. Yes, strict Elo matching would have resulted in long queue times and a lot of "failed to find a match" messages, but doing it the way they did is exactly the same as having no Elo at all - either way you get good players on your team if you are lucky and bad ones if you forgot to sacrifice a goat last night.