112 replies to this topic

[GoManGo]

Look at the screenshot above and let me know exactly how my sucky play resulted in a loss. Better yet let me know how I could have totally won that match with 6 people vs 11.

The problem is =NO ONE CARES NOT EVEN PGI HAHA.The current MM and ELO are for the STAT PADDERS that truly think they have accomplished something in MWO?

[PhoenixFire55]

GoManGo, on 26 February 2014 - 08:17 AM, said:

The problem is =NO ONE CARES NOT EVEN PGI HAHA.The current MM and ELO are for the STAT PADDERS that truly think they have accomplished something in MWO?

I don't really understand how you can pad any stats if nobody sees any stats ...

[SI The Joker]

Reptilizer, on 26 February 2014 - 07:49 AM, said:

What a strange stat to use for measuring efficiency in battle.

I mean, imagine a guy aiming three seconds before every shot (even with medium lasers) and thus having hitrates of about 80% on everything.

Is he really more effective than the guy missing about 50% of the time firing continuously without much aiming?

Who has higher total damage numbers? Who killed more mechs? Who wins his team more games?

Besides that, i also hate e-peens:

No to show ELO! No to show ELO!

Missles are an exception to the rule as they are countered by AMS & cover.

So the stats pages are funny... multiply how many hits you have on your AC 5 x 5 and see the totals not match up. Not sure why? Crits?

Anyway...

In my view (and I'm crazy - this is known) your accuracy is what matters because any fool can run around the map and snatch up kill shots on cored mechs... heck spider/locust/light pilots do this all the time now that MGs are actually sorta useful.

That in and of itself, in my view, makes the KDR rating a broken measurement. A player can also simply be part of a killer 3-man team and contribute very little... and still get wins... thereby invalidating (again, in my view) W/L as a good measurement as well.

What's left? Accuracy. How many times can you connect your shot with your enemy. To me, that's what matters.

Please note these are my views... you can agree or not... I'm cool with that.

[Joseph Mallan]

PhoenixFire55, on 26 February 2014 - 08:08 AM, said:

This is how PUG matches look for me nowadays. Just finished this game ...



6 vs 11 ... hard to win that kind of a match. Was 9th straight loss today, all others pretty much same, 3-4 people on the team just not doing anything. Still balanced according to PGI I suppose.

PhoenixFire55, on 26 February 2014 - 08:14 AM, said:

Look at the screenshot above and let me know exactly how my sucky play resulted in a loss. Better yet let me know how I could have totally won that match with 6 people vs 11.

I'll take the challenge. you could have teamed up with the Battlemaster an Stalker, stuck together and stalked the enemy stragglers evening the Attrition rate..

[PhoenixFire55]

Joseph Mallan, on 26 February 2014 - 08:30 AM, said:

I'll take the challenge. you could have teamed up with the Battlemaster an Stalker, stuck together and stalked the enemy stragglers evening the Attrition rate..

We did stay together, we weren't even poking inside the middle, we just stayed and defended. I told everyone to just defend. Then we had 2 rambos who thought they knew better, then ran off and died. Then they came at us from 2 sides with 12 mechs. If you think that 3 assault mechs can chase stragglers all around Terra Therma, well ...

[Joseph Mallan]

Well you didn't say I had to give a sure fire way to win, just a way you could have.

The enemy stayed together and focus fired like a drill team, You woulda had your work cut out for you even if your missing team mates were there I think.

Edited by Joseph Mallan, 26 February 2014 - 08:37 AM.

[PhoenixFire55]

Joseph Mallan, on 26 February 2014 - 08:36 AM, said:

The enemy stayed together and focus fired like a drill team, You woulda had your work cut out for you even if your missing team mates were there I think.

They didn't really focus fire, but even if they did point remains same. One team is clearly a better team.

[Strum Wealh]

Nicholas Carlyle, on 26 February 2014 - 07:25 AM, said:

Like I said, Iceserpent explained the math very well. I'm not about to butcher what he said.

Then provide a link and quotation to the post in question, and actually make an argument in favor of your point.

I could locate and read said post myself, in time (and here is a link to IceSerpent's content, for anyone with a similar inclination).

However, I am neither obligated nor inclined to make your argument for you, and I will not do so. I will, however, call upon you to either provide something of substance to defend your position or to concede to mine... and if you feel that IceSerpent's position is stronger or more correct than Paul's, then the onus is on you to make and support an argument in favor of that position.

Vague allusions and references without a link & quotation (especially when such links and quotes can be provided with relative ease) mean little-to-nothing, in this case.

Nicholas Carlyle, on 26 February 2014 - 07:25 AM, said:

But using the numbers Paul presented from the 4 matches he looked at, the matchmaker is not working.

How so, specifically? And by what specific measures are you defining & distinguishing "working" and "not working"?

The previously-cited Jan. 21 Update indicates that an Elo rating of 1350 is "is just above the median player of the game", where "median" was likely specifically used instead of "mean" to indicate that half of all MWO players had Elo ratings above that value (while the other half had Elo ratings below that value) at the time of the analysis (though, as a side-note, I think it would be interesting to also know what the mean and mode of the Elo ratings were at that time, or to know what all three are at any given time).

The differences in overall Elo rating between the teams in the screenshots Paul mentions are 10, 47, 101, and 65 - the largest of those is only ~8% of the given (approximate) median, and none of them are world-shattering when one considers that the average Elo of a PUG team could be fairly close to the given (approximate) median (e.g. in the 1300-1400 range), which also happens to be fairly close to the "standard player Elo" rating given in the previously-linked May 21 Patch Notes (e.g. 1300).

Though, while the teams can be fairly even with regard to the teams' Elo ratings, there can be quirks within either or both teams with regard to the distribution of players' Elo ratings; one could, for example, end up with extreme situations like this one or this one, with one team consisting of one (presumably) "very high Elo" Lance & two (presumably) "lower Elo" Lances versus a team consisting of three (presumably) "average" Lances.

Is the Matchmaker "perfect"? Certainly not & far-from-it, but there are evidently plans to improve it (such as the "tonnage matching" mentioned in the previously-linked Jan. 24 post from Paul). But, to say that is "not working" is a gross overstatement at best, and a blatant falsehood more often than not.

[Reno Blade]

If your example has two sides of equal ELO...
one Team with average elo of all the same
the other Team with few with very high elo and the rest lot with very low elo

But the higher probable situation on that game (or on top of elo) was the lights and trials on team2 beeing where they should not have been and killed by sheer numbers.
Also the amount of lights easily make you lose if they don't attack aggressively.
A few good PPC shots to these lights by a premade can leg them very fast.
Thats why they didnt get to do much damage.
Without any other information (video) there is no way to rate the two teams' perfomance.

[Joseph Mallan]

PhoenixFire55, on 26 February 2014 - 08:46 AM, said:

They didn't really focus fire, but even if they did point remains same. One team is clearly a better team.

And the better team is supposed to win.

[Gallowglas]

Joseph Mallan, on 26 February 2014 - 03:11 AM, said:

We have no need for the info except ego stroking, and hazing.

That may be true of some players, but those (perhaps few) of us who aren't reactionary adolescents with self-image issues, I think that exposing ratings and the mechanics of matchmaking could go a long way toward understanding more about personal advancement and getting additional metric data about matches in general. Granted, I understand you'd get a lot of trolling with that information, so I at least understand why they don't do it. I just personally, for completely selfish self-critique purposes, wish they did.

[Nicholas Carlyle]

Strum Wealh, on 26 February 2014 - 08:53 AM, said:

Then provide a link and quotation to the post in question, and actually make an argument in favor of your point.

I could locate and read said post myself, in time (and here is a link to IceSerpent's content, for anyone with a similar inclination).

However, I am neither obligated nor inclined to make your argument for you, and I will not do so. I will, however, call upon you to either provide something of substance to defend your position or to concede to mine... and if you feel that IceSerpent's position is stronger or more correct than Paul's, then the onus is on you to make and support an argument in favor of that position.

Vague allusions and references without a link & quotation (especially when such links and quotes can be provided with relative ease) mean little-to-nothing, in this case.

How so, specifically? And by what specific measures are you defining & distinguishing "working" and "not working"?

The previously-cited Jan. 21 Update indicates that an Elo rating of 1350 is "is just above the median player of the game", where "median" was likely specifically used instead of "mean" to indicate that half of all MWO players had Elo ratings above that value (while the other half had Elo ratings below that value) at the time of the analysis (though, as a side-note, I think it would be interesting to also know what the mean and mode of the Elo ratings were at that time, or to know what all three are at any given time).

The differences in overall Elo rating between the teams in the screenshots Paul mentions are 10, 47, 101, and 65 - the largest of those is only ~8% of the given (approximate) median, and none of them are world-shattering when one considers that the average Elo of a PUG team could be fairly close to the given (approximate) median (e.g. in the 1300-1400 range), which also happens to be fairly close to the "standard player Elo" rating given in the previously-linked May 21 Patch Notes (e.g. 1300).

Though, while the teams can be fairly even with regard to the teams' Elo ratings, there can be quirks within either or both teams with regard to the distribution of players' Elo ratings; one could, for example, end up with extreme situations like this one or this one, with one team consisting of one (presumably) "very high Elo" Lance & two (presumably) "lower Elo" Lances versus a team consisting of three (presumably) "average" Lances.

Is the Matchmaker "perfect"? Certainly not & far-from-it, but there are evidently plans to improve it (such as the "tonnage matching" mentioned in the previously-linked Jan. 24 post from Paul). But, to say that is "not working" is a gross overstatement at best, and a blatant falsehood more often than not.

Sorry, I've done the whole citing thing. I'm over it. I don't care about you enough to do it.

Go find the post if you want. I don't care. But the math has been done.

It doesn't work, and PGI does not give enough information end of story.

Edit: By the way, Steelers suck.

Edited by Nicholas Carlyle, 26 February 2014 - 09:06 AM.

[Joseph Mallan]

Gallowglas, on 26 February 2014 - 09:04 AM, said:

That may be true of some players, but those (perhaps few) of us who aren't reactionary adolescents with self-image issues, I think that exposing ratings and the mechanics of matchmaking could go a long way toward understanding more about personal advancement and getting additional metric data about matches in general. Granted, I understand you'd get a lot of trolling with that information, so I at least understand why they don't do it. I just personally, for completely selfish self-critique purposes, wish they did.

Teah... I know, I have those selfish reasons myself sometimes.

[Sandpit]

Zolaz, on 26 February 2014 - 03:37 AM, said:

You know who doesnt want to know their stats? Bad pilots.

Joseph Mallan, on 26 February 2014 - 04:07 AM, said:

I'm not bad, I'm Average. And that is the start of the hazing I mentioned. Going to show us your epeen now?

I think joe hit the nail on the head

[Malcolm Vordermark]

The matchmaker can only put together games based on how well it has recorded people doing. It can not predict a given players performance during a certain match. The elo is calculated based on wins but there is no way using that data to predict if someone is going to wander off and get themselves killed.

[Joseph Mallan]

Nicholas Carlyle, on 26 February 2014 - 09:05 AM, said:

Edit: By the way, Steelers suck.

Those are fightin' words round here Sir. My Parents grew up in West Newton, PA. We are Steeler folk!

[Nicholas Carlyle]

Joseph Mallan, on 26 February 2014 - 09:08 AM, said:

Those are fightin' words round here Sir. My Parents grew up in West Newton, PA. We are Steeler folk!

Trust me, I'm very calculated with my sports insults.

Sidney Crosby is a tool, for reference.

[Willard Phule]

Bah, whatever, it's not really a sport anyway. No ice, no sticks, no point....and, by the way, the Penguins DO suck.

[Screech]

I just wish one person could explain what they would do with this information that would actually improve the matchmaker.

[Apnu]

I doubt PGI will show us match maker and elo data.

They've said many times that they are against Elo scores being public because of how crappy gamers are to each others. Its that simple.

Bryan Eckman recently said in the 3 part NGNG interview (released last weekend) that the matchmaker isn't where they want it and they've got a lot of work to get it where they want it. The same for Elo calculations too. There is no way they are going to give us data on a system that's being reworked.

PUG matches, for me, are completely random. Sometimes I'm on a roll with a few good pilots and we win decisively, sometimes its a close game (those are the best), and sometimes half my team is newbies and its a crap shoot.

If PUG players are frustrated that they're carrying a team frequently (I am at times) I suggest they join a unit and drop in a 4-man with some kind of voice coms. That improves the chances of controlling the PUG factor and its more fun to experience victory or defeat with comrades in arms.

Trust me on this. Solo play is fun, and sometimes I think the matchmaker hates me. But 4-man and 12-man are also awesome.

There are plenty of units in this game, you'll find most of them in the faction forums, if you're a lone wolf, give them a try and see what you think. If you don't like one, no biggie, try out another unit.