45 replies to this topic

[Glythe]

I think it has a lot to do with people who have come back to the game on a patch day. If they didn't play during the Elo seed phase they are listed at 1300.

I suspect a lot of people are in the undercon category right now and Elo is completely screwed..... again.

[Serapth]

StalaggtIKE, on 06 March 2013 - 04:34 AM, said:

It probably have to do with host state rewind. More successful shots requires a change in playstyle. Even my Atlas felt more vulnerable. It's funny to imagine even he had a bit of a lagsheild.

This should effect everyone pretty much equally though. It doesn't make sense that I should consistently see 4 or more people on my team doing > 100 damage, while the other team have maybe 1 player that is ineffective.

More confusing is it seems to be only in certain weight classes for me, and my output is incredibly consistent. Here is my performance since the patch went live:



As you can see, in the medium category I'm hitting pretty damned near 50:50 win loss, the ideal goal for Elo.

But in the heavy, especially when in my Flame, it's a 2 to 1 loss ratio. I can pretty safely say I'm a non-factor in this equation, as my damage output is pretty consistent across all chassis and my KDR is in the positive. Stats are also skewed a bit low, thanks to a pair of disconnects ( one of which was ironically a win ).

It simply seems that when I take certain mechs, I get teamed up with trial mechs and complete greenies. I wish it tracked end results, as generally the loses were all stomps... 8-0 8-2, that kind of stuff. The kind of matches I know we are toast 30 seconds in, from watching your non-ECM team run into the water, or watching 3 mechs run straight over the ridgeline... stuff I wasn't seeing anymore once they implemented Elo.

[Frans83]

Silentium, on 06 March 2013 - 04:59 AM, said:

Just went 0-3 this morning, stomped twice, and one where the winners had 6 left. It felt just like old times, but may be an anomoly. Or I suck more than I thought.Edit: I was in my flame all three games.

This. Two games my team got stomped, 1 we won and it came down to 2 on 2, and the fourth the hamster came off the wheel and the game crashed.

[Serapth]

Silentium, on 06 March 2013 - 04:59 AM, said:

Just went 0-3 this morning, stomped twice, and one where the winners had 6 left. It felt just like old times, but may be an anomoly. Or I suck more than I thought.
Edit: I was in my flame all three games.

That's the really bizarre part for me... It's when I take Flame that I get teamed with drooling idiots. If I stick to other mechs Elo seems to be doing it's thing.

[Serapth]

Hmmmmm.... now Im starting to wonder if Flame is what's bugged... like they arent properly tracking elo on that particular mech?

Of course, way too little data to draw conclusions yet.

[Glythe]

And this is why I want Elo transparency so we can see if our Elo is giving us fair matches.....

I win a lot with my Atlas even after the changes. My medium was ranked nowhere near as high but today I got soul crushing teams (both mine being so bad and the enemy being so much better) as though this mech had its elo doubled or tripled.

[Silentium]

To be fair to my team mates, they weren't bad, our opponents were just much more organized. In one match, I barely got off any damage before trying to prevent a cap and reaping the LRM harvest. The other times we at least mounted something approximating a resistance.

[Serapth]

Glythe, on 06 March 2013 - 05:09 AM, said:

And this is why I want Elo transparency so we can see if our Elo is giving us fair matches.....

I win a lot with my Atlas even after the changes. My medium was ranked nowhere near as high but today I got soul crushing teams (both mine being so bad and the enemy being so much better) as though this mech had its elo doubled or tripled.

That's the part that really gets me when Elo craps the bed... I dont really find my performance changes all that much. It doesn't feel like the enemy is that much more accurate or playing all that much better positionally... It just inevitably seems like I am paired up with 3 or 4 complete morons, which generally results in the match being me and another mech or two, vs 8.

Again though, it's only in certain mechs.

Oddly though, my elo score shouldn't really be all that much higher in one category than another, I play pretty consistently across mediums-assaults and never play lights, and have pretty consistent performances.

Edited by Serapth, 06 March 2013 - 05:15 AM.

[Fooooo]

The only thing i have noticed with elo so far is it seems different based upon the weight class you are in. (they did say weight still plays a factor)

As in, if your in an atlas, your more likely to be used as a premade filler or opponent etc etc. I seem to get a lot more raven groups using my atlas than my heavys. (and my assaults started out on the default elo as I did not use them during the seeding.)

I'm guessing this is because the MM is weighting the higher ton mechs higher. So instead of giving a team who needs a low elo player to fill up the team average a person in a commando, they give them someone in an atlas, as it has more "weight" in the MM.

Thats the only thing I have somewhat noticed.....my atlas, (with supposedly lower elo rating) gets more 1 sided matches and 4mans etc (both ways, my side or theirs) than my heavys with (supposedly) higher elo.

Its only a general observation tho, I have not played enough games to have a good sample atm, so take it all with a grain of salt.

Edited by Fooooo, 06 March 2013 - 05:22 AM.

[csutkakoma]

Serapth, on 05 March 2013 - 09:40 PM, said:

I am not sure if anyone else experienced this, or if I just had a bad night, but there seems to be something up with Elo.

One thing I noticed after Elo was put in was damage was spread pretty well across the board. Rarely did you see people scoring < 100 damage for example in a given round, nor did you generally see people walk away with > 1000+ damage rounds.

Tonight though, after the patch, Im regularly seeing 4-6 teammates do < 100 damage during a loss. I can honestly say, this was a complete rarity between the point Elo was put in and now. Perhaps oddest of all, it all seems to be when I pilot a heavy ( Ilya or Flame ).


Anyone else experiencing this? During losses are you seeing lots of really awful scores on your side of the fence?

Well in my experience this kind of things happens when I buy a new mech. To me it seems that the Elo is getting better if you play more with the new mech. But everytime I start with a new machine the first 10-20 match is a disaster, no matter how well I am doing.

Edited by Forsaken Ghost, 06 March 2013 - 05:22 AM.

[Glythe]

Serapth, on 06 March 2013 - 05:14 AM, said:

That's the part that really gets me when Elo craps the bed... I dont really find my performance changes all that much. It doesn't feel like the enemy is that much more accurate or playing all that much better positionally... It just inevitably seems like I am paired up with 3 or 4 complete morons, which generally results in the match being me and another mech or two, vs 8.

Again though, it's only in certain mechs.

Oddly though, my elo score shouldn't really be all that much higher in one category than another, I play pretty consistently across mediums-assaults and never play lights, and have pretty consistent performances.

I lost 10 games in a row today with my Cicada. I was often top damage or second place. I often get rear kills on people not paying attention but my performance overall did not vary much. It's pretty much as you describe it for my team.

I think everyone is bound to have one mech weight that suits them better than the rest. It makes sense that one weight class would be more noticeable than another if that is your best class.

Fooooo, on 06 March 2013 - 05:19 AM, said:

The only thing i have noticed with elo so far is it seems different based upon the weight class you are in. (they did say weight still plays a factor)

As in, if your in an atlas, your more likely to be used as a premade filler or opponent etc etc. I seem to get a lot more raven groups using my atlas than my heavys. (and my assaults started out on the default elo as I did not use them during the seeding.)

I'm guessing this is because the MM is weighting the higher ton mechs higher. So instead of giving a team who needs a low elo player to fill up the team average a person in a commando, they give them someone in an atlas, as it has more "weight" in the MM.

Unless you have a source then thats' just misinformation. Currently the matchmaker does not consider weight as part of Elo (currently disabled). It matches you based on your numeric Elo rating and nothing else. The only time the different weight classes become a factor is when you switch weights. You have 4 Elo scores corresponding to the light, medium, heavy and assault classes. Later weight will have some factor but we don't know what. Right now it is not a factor at all.

Edited by Glythe, 06 March 2013 - 05:43 AM.

[Serapth]

Glythe, on 06 March 2013 - 05:38 AM, said:

I lost 10 games in a row today with my Cicada. I was often top damage or second place. I often get rear kills on people not paying attention but my performance overall did not vary much. It's pretty much as you describe it for my team.

I think everyone is bound to have one mech weight that suits them better than the rest. It makes sense that one weight class would be more noticeable than another if that is your best class.

Hmmm, one thing I suppose would be a factor is Ilya and Flame are both CBill bonus mechs for me, so I do use them more often. This should equally effect Ilya though if that were the case.

[Star Captain Obvious Kerensky]

Serapth, on 05 March 2013 - 09:40 PM, said:

Tonight though, after the patch, Im regularly seeing 4-6 teammates do < 100 damage during a loss. I can honestly say, this was a complete rarity between the point Elo was put in and now. Perhaps oddest of all, it all seems to be when I pilot a heavy ( Ilya or Flame ).

The has been my experience with Elo since day one. The Matchmaker will pair 4 high Elo players (Read: Experts) with 4 low Elo players (Read: Newbies), and expect the expert players to carry the team.

You are probably noticing it because you have an Elo score per chassis, and your Elo score on heavies is much higher, so the matchmaker pairs you with a newbie to average out the team Elo score.

[Hayashi]

The system keeps throwing me into teams where I do either #1 or #2 highest damage in a Craven. Win or loss depends on whether or not we can do enough to carry the rest of the team. The team on the other side, however, also usually seems to be carried by one or two pilots.

Then I switched to a Commando (because I mastered the Craven at last and I'm itching to get out of the Raven chassis) and got murdered nonstop, because I can barely pull my own weight (~400 damage) when the thing dies in two hits from a half-competent pilot, leaving only one good pilot in the team against two on the other. The other players are kind of cannon fodder for both sides, honestly.

On a side note I think this kind of shows just how overpowered the Craven is, and how useless the Commando is.



ELO doesn't seem to match players of equal skill levels within the teams, but rather only players of equal skill levels between teams, so it's almost always the case that the team's getting carried by somebody - unless it's a premade drop.

Since ELO, I've also seen a drastic change in the influence of being outnumbered. Prior, 7-8 or 8-7 wasn't actually very conclusive, since chances could swing either way. But now, if you start one mech short, chances are the one that's short is the player that would have carried your team. There is no shortage of players with terribad ELO scores, but a very small number of players with high ELO scores.

Edited by Hayashi, 06 March 2013 - 06:12 AM.

[NKAc Street]

From my own personal observations in the matches I have been in that elo doesn't match each team up player for player by elo, it appears to group people who launch near the same time up and group average the teams. So that is why if you believe or know you are a good player you still find trial mech players on your team or the other team.

[Serapth]

Eldragon, on 06 March 2013 - 05:55 AM, said:

The has been my experience with Elo since day one. The Matchmaker will pair 4 high Elo players (Read: Experts) with 4 low Elo players (Read: Newbies), and expect the expert players to carry the team.

You are probably noticing it because you have an Elo score per chassis, and your Elo score on heavies is much higher, so the matchmaker pairs you with a newbie to average out the team Elo score.

What I dont get is, I'm not an exceptional player. Elo should be stacking my team with experts! I mean, I pug exclusively and as a result my win/loss is in the red. Given that this is the only metric for Elo, the fact I top the board on the losing side should mean absolutely nothing.

So, shouldn't I be reaping the benefits of Elo? I would say to a degree I was, until this patch.

[Glythe]

Eldragon, on 06 March 2013 - 05:55 AM, said:

The has been my experience with Elo since day one. The Matchmaker will pair 4 high Elo players (Read: Experts) with 4 low Elo players (Read: Newbies), and expect the expert players to carry the team.

You are probably noticing it because you have an Elo score per chassis, and your Elo score on heavies is much higher, so the matchmaker pairs you with a newbie to average out the team Elo score.

This has also been my experience.

And this is an incredibly lame way to match players. New players don't want to be killed in 1 volley by a 6 ppc stalker at 800m. Experienced players don't want anyone on their team to be running a trial mech.

Both groups of players have better matches if we keep them separate.

Edited by Glythe, 06 March 2013 - 07:02 AM.

[StalaggtIKE]

Serapth, on 06 March 2013 - 05:01 AM, said:

This should effect everyone pretty much equally though. It doesn't make sense that I should consistently see 4 or more people on my team doing > 100 damage, while the other team have maybe 1 player that is ineffective.

More confusing is it seems to be only in certain weight classes for me, and my output is incredibly consistent. Here is my performance since the patch went live:



As you can see, in the medium category I'm hitting pretty damned near 50:50 win loss, the ideal goal for Elo.

But in the heavy, especially when in my Flame, it's a 2 to 1 loss ratio. I can pretty safely say I'm a non-factor in this equation, as my damage output is pretty consistent across all chassis and my KDR is in the positive. Stats are also skewed a bit low, thanks to a pair of disconnects ( one of which was ironically a win ).

It simply seems that when I take certain mechs, I get teamed up with trial mechs and complete greenies. I wish it tracked end results, as generally the loses were all stomps... 8-0 8-2, that kind of stuff. The kind of matches I know we are toast 30 seconds in, from watching your non-ECM team run into the water, or watching 3 mechs run straight over the ridgeline... stuff I wasn't seeing anymore once they implemented Elo.

Keep in mind each weight class has its separate Elo score. My guess your heavy Elo is lower.

[Padic]

Sampling error. You've got one night's worth of data. It's only like 30 matches.

I had a rough night last night, but I attribute that to driving a Cent-D with an XL300. I'm finally getting side-cored with the laser changes.

I *think* I can still make it work, but the simple fact is that I might need to reconsider running XL engines.

[Star Captain Obvious Kerensky]

Serapth, on 06 March 2013 - 06:50 AM, said:

What I dont get is, I'm not an exceptional player. Elo should be stacking my team with experts! I mean, I pug exclusively and as a result my win/loss is in the red. Given that this is the only metric for Elo, the fact I top the board on the losing side should mean absolutely nothing.

So, shouldn't I be reaping the benefits of Elo? I would say to a degree I was, until this patch.

But the Elo system doesn't stack your team with experts, first it tries to stack your team with similarly skilled players. The matchmaker starts mixing skill levels only after it can't find a perfect match.

But this is wild speculation on my part, since the Elo system is kept secret, I would say keep playing and see what happens.

Because the Elo system is willing to loosen the matchmaking restrictions as it builds the teams, you're still going to encounter a wide variety of player skill sets. So you can expect to continue to see players in Atlas DDCs die while doing less than 100 damage.