25 replies to this topic

[Delas Ting Usee]

FIX IT PLEASE!!!

Go through the data collected so far and fix the damn thing. It's worse than before, my team who I pug with is getting steamrolled constantly and before the patch there were close fights with the occasional steamroll - not now, this is despite me doing an average of 200-250 pts of damage. Whatever you guys did on the 19th and 20th patch IS NOT WORKING. Please, go through it again. Thanks

Edited by Delas Ting Usee, 21 February 2013 - 07:47 AM.

[Redshift2k5]

Elo matchmaking was worknig great for me and my friends all night last night.

It's possible you are not yet into your 'correct' elo plateau. The more you play the more accurate your elo score will become.

[iminbagdad]

This is going to sound more mean than it is.

Unless you are piloting a 4mg spider averaging 200-250 damage puts you in a low bracket. Your elo was artificially inflated and is now going to go down to your correct rating. Until they take into account anything besides damage for score and win loss for elo you need to maximize your damage and kill abilities if you want to hold your ranking.

[Broceratops]

Edited by Broceratops, 21 February 2013 - 08:08 AM.

[Inviticus]

Since they turned it back on, I have seen nothing but pugstomp premades over and over. ELO is either not working or it's a complete failure.

[Broceratops]

I disagree. I played all night last night and didn't get any ******** teammates. Everyone handled themselves pretty well. This goes for the other team as well. No one did suicidally stupid things.

I still won all my games except 1 or 2 because the other team played too damn passive, but I take this as a symptom of them being used to camping for a bit while crappy pugs charged 1 by 1.

[Dr Killinger]

My experience has been decent thus far. Even spread of wins and losses, some very close indeed.

[Truesight]

Besides, a perfect matchmaker will give you a 50:50 Win/Loss ratio, how far are you really from that?

[Bilbo]

iminbagdad, on 21 February 2013 - 08:03 AM, said:

This is going to sound more mean than it is.

Unless you are piloting a 4mg spider averaging 200-250 damage puts you in a low bracket. Your elo was artificially inflated and is now going to go down to your correct rating. Until they take into account anything besides damage for score and win loss for elo you need to maximize your damage and kill abilities if you want to hold your ranking.

Elo has no correlation in game performance. All it cares about is whether you won or lost the match. He and his team will settle into the correct bracket over time.

[Comassion]

A simple consequence of ELO is that before ELO, you'd generally get matched against an average team.

That means if you won most of your games as 'close matches', then you were slightly above average. However, that means you won most of your games - we'll call it 90%.

If you totally stomped your opponents in most of those matches, then you may be an elite player. That might mean you won, say, 95% of your matches.

Problem is, now the ELO of the 'slightly above average' guys is very close to the ELO of the 'elite' guys, and that's rough on the 'above average' guys. It doesn't care that you had a close game, it only cares that you won.

To further complicate your problem, many of the very best were still running 8-mans pre-ELO. That meant losing a lot more than with 4-mans, because the only other guys left in the 8-man queue after a week or two were other elite teams, because everyone who wasn't on their game got smashed. We had good days and bad days, but probably ended up winning around 55-60% of our 8-man drops. Problem there is, we lost 40-45% of our matches, compared to you losing 10-15% of yours, so our ELO may very well be lower than yours even if we were better players, because we entered more challenging matches.

Keep playing the game and it'll start to even out now that you're losing matches, and we're winning matches, and as time goes on we'll both approach more accurate ELO scores.

If all you did was stomp around in 4-mans, you may be in for a rough set of matches as your ELO goes toward a more accurate value. Remember that for every match you lose, your future matches will get easier, and for every match you win they will get harder, until you're at a level where you're looking at something close to a 50/50 win/loss record.

[Broceratops]

That is an excellent analysis Comassion. Everyone's Elo is off by varying amounts to the true number and the only way to fix it is to keep playing and let the system sort itself out in the next few weeks.

[GODzillaGSPB]

Delas Ting Usee, on 21 February 2013 - 07:45 AM, said:

FIX IT PLEASE!!!

Go through the data collected so far and fix the damn thing. It's worse than before, my team who I pug with is getting steamrolled constantly and before the patch there were close fights with the occasional steamroll - not now, this is despite me doing an average of 200-250 pts of damage. Whatever you guys did on the 19th and 20th patch IS NOT WORKING. Please, go through it again. Thanks

Wait...let me get this straight:

- you're on a team
- and you pug with your team

So that means you're in a 4man premade and the game pits you against pugs. This is usually the case.

But those pugs steamroll you. Mhm...lol!

[Erik Jast]

GODzillaGSPB, on 21 February 2013 - 08:37 AM, said:

Wait...let me get this straight:

- you're on a team
- and you pug with your team

So that means you're in a 4man premade and the game pits you against pugs. This is usually the case.

But those pugs steamroll you. Mhm...lol!

There was a change he did not like. Instead of the game putting him against random pugs, it is matching him against evenly matched pugs. But he and his team are terribad so he gets rolled. He wants the old pugstomp days back.

[Joseph Mallan]

Comassion, on 21 February 2013 - 08:25 AM, said:

A simple consequence of ELO is that before ELO, you'd generally get matched against an average team.

That means if you won most of your games as 'close matches', then you were slightly above average. However, that means you won most of your games - we'll call it 90%.

If you totally stomped your opponents in most of those matches, then you may be an elite player. That might mean you won, say, 95% of your matches.

Problem is, now the ELO of the 'slightly above average' guys is very close to the ELO of the 'elite' guys, and that's rough on the 'above average' guys. It doesn't care that you had a close game, it only cares that you won.

To further complicate your problem, many of the very best were still running 8-mans pre-ELO. That meant losing a lot more than with 4-mans, because the only other guys left in the 8-man queue after a week or two were other elite teams, because everyone who wasn't on their game got smashed. We had good days and bad days, but probably ended up winning around 55-60% of our 8-man drops. Problem there is, we lost 40-45% of our matches, compared to you losing 10-15% of yours, so our ELO may very well be lower than yours even if we were better players, because we entered more challenging matches.

Keep playing the game and it'll start to even out now that you're losing matches, and we're winning matches, and as time goes on we'll both approach more accurate ELO scores.

If all you did was stomp around in 4-mans, you may be in for a rough set of matches as your ELO goes toward a more accurate value. Remember that for every match you lose, your future matches will get easier, and for every match you win they will get harder, until you're at a level where you're looking at something close to a 50/50 win/loss record.

Thanks for the perspective. This was a very helpful description of how the system may be doing things.

I used May cause I never even heard of Elo before open beta. Only ELO I knew broke up in the late 70s early 80s.

[Mechteric]

Also be aware of what mechs the 4 of you are taking into battle. Taking 4 Trebs with untested configs into battle against Hunches and Fast Centurions of equal skill mean you will be at a disadvantage and will need to work harder to overcome it.

[Vassago Rain]

Comassion, on 21 February 2013 - 08:25 AM, said:

A simple consequence of ELO is that before ELO, you'd generally get matched against an average team.

That means if you won most of your games as 'close matches', then you were slightly above average. However, that means you won most of your games - we'll call it 90%.

If you totally stomped your opponents in most of those matches, then you may be an elite player. That might mean you won, say, 95% of your matches.

Problem is, now the ELO of the 'slightly above average' guys is very close to the ELO of the 'elite' guys, and that's rough on the 'above average' guys. It doesn't care that you had a close game, it only cares that you won.

To further complicate your problem, many of the very best were still running 8-mans pre-ELO. That meant losing a lot more than with 4-mans, because the only other guys left in the 8-man queue after a week or two were other elite teams, because everyone who wasn't on their game got smashed. We had good days and bad days, but probably ended up winning around 55-60% of our 8-man drops. Problem there is, we lost 40-45% of our matches, compared to you losing 10-15% of yours, so our ELO may very well be lower than yours even if we were better players, because we entered more challenging matches.

Keep playing the game and it'll start to even out now that you're losing matches, and we're winning matches, and as time goes on we'll both approach more accurate ELO scores.

If all you did was stomp around in 4-mans, you may be in for a rough set of matches as your ELO goes toward a more accurate value. Remember that for every match you lose, your future matches will get easier, and for every match you win they will get harder, until you're at a level where you're looking at something close to a 50/50 win/loss record.

Yeah, this. I had a 62% win ratio, but have been doing heavy 8 manning since we first got it, and when I was in the ghetto, it was typically just me in my atlas.

My K/D has skyrocketed since the ladder hit. It's now slightly above 1, where it was 0.80 previously. This while I'm up against actual HARD opponents in the ghetto.

[Zaptruder]

Keep playing. Pre-seeding wasn't an accurate assessment for Elo; it was just there to have something to start the ladder with.

The beauty of the system is that it doesn't care about the noise; the particulars of the mechs you build or what you use. It will find your equilibrium for your general play style.

If you're an ok player that sticks to a top tier mech, your Elo will be adjusted to reflect that. If you only run crappy mechs, your Elo will be adjusted to reflect that.

If you gain a competitive advantage from playing 4 mans all the time, your Elo will be adjusted to reflect that.

If you run ECM on your mechs can gain a competitive advantage... your Elo will be adjusted to reflect that.

If you try everything, your Elo will average out, and you'll gain competitive spikes and despikes when you get into better mechs and use worse mechs.

[Sug]

Delas Ting Usee, on 21 February 2013 - 07:45 AM, said:

this is despite me doing an average of 200-250 pts of damage.

Sadly that would put you in the top 3 of 95% of the pug teams i get put on.

Truesight, on 21 February 2013 - 08:18 AM, said:

Besides, a perfect matchmaker will give you a 50:50 Win/Loss ratio, how far are you really from that?

God I hate that ratio. Why even play if you lose 50% of the time? Just sit at your desk, flip a ******* coin and smile.

Hopefully when CW comes out there'll be some actual thought involved with playing this game and not just random matches.



Also Elo only cares about wins/losses.

Edited by Sug, 21 February 2013 - 09:05 AM.

[Padic]

Comassion, on 21 February 2013 - 08:25 AM, said:

A simple consequence of ELO is that before ELO, you'd generally get matched against an average team.

That means if you won most of your games as 'close matches', then you were slightly above average. However, that means you won most of your games - we'll call it 90%.

If you totally stomped your opponents in most of those matches, then you may be an elite player. That might mean you won, say, 95% of your matches.

Problem is, now the ELO of the 'slightly above average' guys is very close to the ELO of the 'elite' guys, and that's rough on the 'above average' guys. It doesn't care that you had a close game, it only cares that you won.

To further complicate your problem, many of the very best were still running 8-mans pre-ELO. That meant losing a lot more than with 4-mans, because the only other guys left in the 8-man queue after a week or two were other elite teams, because everyone who wasn't on their game got smashed. We had good days and bad days, but probably ended up winning around 55-60% of our 8-man drops. Problem there is, we lost 40-45% of our matches, compared to you losing 10-15% of yours, so our ELO may very well be lower than yours even if we were better players, because we entered more challenging matches.

Keep playing the game and it'll start to even out now that you're losing matches, and we're winning matches, and as time goes on we'll both approach more accurate ELO scores.

If all you did was stomp around in 4-mans, you may be in for a rough set of matches as your ELO goes toward a more accurate value. Remember that for every match you lose, your future matches will get easier, and for every match you win they will get harder, until you're at a level where you're looking at something close to a 50/50 win/loss record.

Don't forget that Elo ranking cares not just whether you won or lost, but also the Elo of your opponents.

I'm pretty sure that a hypothetical team who only played 8v8 and who had a 50/50 win/loss ratio will still have a very high Elo because you're winning (well, half the time, anyway) against people who have very high Elo rankings.

[Rakashan]

Guys, you're overthinking the math behind Comassion's post.

If you are making no changes to your gameplay (not getting better, etc.) and your ELO has stabilized then you should be winning 50% of your matches. Once your ELO is stable you should be matched against people of your general level and that means that it's going to be a wash. If your play is improving then you'll win more than 50% as you get better than the people in your tier.

There's more going on than that, of course. Other players getting better or rustier around you. People having good or bad days. Random fluctuations in the "balance" of the mechs/weights dropped together. But once your ELO is stable when you look at your stats over 100-200 matches you'd expect to be close to 50% in reasonably well contested matches.

It *is* true that since the seeding was done using matchmaker 2.5 that some people will be artificially inflated based more on the number of matches played against "average" teams instead of the degree of success against those teams. That's why you see people saying that you need to spend some more time establishing your real ELO ranking.