123 replies to this topic

[blackicmenace]

Wintersdark, on 20 June 2014 - 03:09 PM, said:

Wins and losses is the ONLY thing that can be used. Nothing else works. But it's not just wins and losses, it's wins and losses relative to the opposing teams' rating. Thus, losing a match to a higher ranked team has little to no impact on your ranking, for example.

You can't use stats and such, because so much of what determines who wins a match is not quantifiable. Damage? No. You could be some meathead firing all yer guns from the back, grossly inaccurately and not actually accomplishing anything. Damage in and of itself is worthless. Kills? Kills are important, but it doesn't matter if you or your teammate gets the kill. What if you crippled a mech, and moved on to a threatening target, then some opportunistic moron chases after the crippled mech? He gets a kill, you don't, but he actually hurt your team's chances to win while you helped it.

It goes on and on with all the stats. There's only one thing that works. Wins. Maybe you're terrible at the game, can't hit a Dire Wolf's center torso from 50m. But you can lead. You're great at getting pugs to act (comparatively) cohesively. Maybe you're a great scout, or just have a good grasp of when to lead a push. There are countless actions that contribute to victory but cannot be measured.

Hell, this is why premades end up with higher Elo ratings. Not necessarily because the players are better at the game, but because they have comms (or at least just know each other's play styles well) and can act more cohesively. Because of that, they win more.


TLDR: Elo isn't necessarily a measure of skill, but rather a measure of how much a given player contributes to his teams victory on average. These are not the same thing.

It's a rough valuation, by necessity - it's working on very abstract values, after all - and it requires a LOT of matches to really get tuned well for you... And the more different mechs you use, the more "averaged" it'll be across those mechs and thus the more.

Accuracy is a measure of skill. I have no idea what the formula should be but Wins and losses feels a bit blah since its more about the team and a whole host of things than a single person that results in wins or losses. As you said people that play with other people will naturally have a higher ELO vs someone that pugs and maybe thats why I have a problem with it as a rating.

NRP, on 20 June 2014 - 01:43 PM, said:

I've been told that the lowest ELO players are grouped in Charlie lance. I always end up in Charlie lance when I drop solo.

Therefore, I'm a scrub.

I am a scrub too, its a scrub life.

[bullyj]

I am a PUG only player, most of the time I'm in Bravo now. The rest of the time I get Alpha and Charlie pretty much equal. When I first started playing if I saw a faction I was like oh I'm gonna die, Now I usually see a couple of lords and a couple of founders on each side and most of the rest are factions. I would like to think my elo is getting better because of this. Just my observations.

Edited by bullyj, 20 June 2014 - 04:26 PM.

[Wintersdark]

blackicmenace, on 20 June 2014 - 04:05 PM, said:

Accuracy is a measure of skill. I have no idea what the formula should be but Wins and losses feels a bit blah since its more about the team and a whole host of things than a single person that results in wins or losses.

Accuracy isn't. The accuracy stat is almost completely meaningless, unless you work to ensure it's valid. Maybe you blast about at random all the time. Or, maybe you always hit enemy mechs, but just wing them (slap mechs with a laser, etc) or even connect fully but spend most of your time shooting arms and such because you're too slow to figure out that DPS matters far less than Useful Damage.

So, lets assume you use Accuracy to build your own rating system. All I need to do to Smurf is spend a lot of time randomly shooting dirt. This doesn't impact my chances to win a match, but tanks my stats. Low stats = worse opponents = lots of easy wins.

With Elo, the only way to "smurf" is to deliberately lose - unless you're smurfing prior to a tourney or something (the old school kind, not the new private match kind) there's no gain in that - you'd be better off just playing normally.

Quote

As you said people that play with other people will naturally have a higher ELO vs someone that pugs and maybe thats why I have a problem with it as a rating.

Because it works? You need to step back, and rethink what Elo is. It is NOT a measure of your skill. People like to say that, and to some extent it's representative, but it's NOT your skill. It's how much you, in aggregate, contribute to your team's victory or loss - as I said before, these two things are not the same. Elo is a match making tool, not an Awesomeness Rating. And that is one major reason why Elo rankings are hidden.

If you play most matches in a premade, then overall you contribute more to your team's victory than you would have solo, and as such you tend to creep higher in Elo ranking. That's Elo functioning exactly as intended. The more often you play in a premade, the higher your Elo will be, because you contribute more to team victory more often.

It's also important to understand that Elo is something that functions over time, leading to (ideally) better match quality overall rather than perfect matches each time. As such, a single match (or even 10) is statistically insignificant.

That's why it does work, to some degree. Say you play silly troll mechs occasionally. This will result in slightly more losses against comparably rated opponents, but on average it'll work out given a particular amount of troll-build-derping. Someone who plays troll mechs more often will be rated lower. Thus, when they are slotted into a match in the matchmaking process, the matchmaker still has a reasonable idea of how likely you are to "bring it" to a match.

[Joseph Mallan]

blackicmenace, on 20 June 2014 - 02:08 PM, said:

I solo pug 99.9% of the time and 99% of the time I am in Charlie lance.

Cool So with Helen in Alpha, You in Charlie, and me in Bravo most often, we have proven... nothing?

[blackicmenace]

Valid points about accuracy although I always shake my head in wonder why at the start of a match people shoot dirt. I always take dirt shots as a bad omen of things to come.

Joseph Mallan, on 20 June 2014 - 05:20 PM, said:

Cool So with Helen in Alpha, You in Charlie, and me in Bravo most often, we have proven... nothing?

We have proven I am a scrub.

[Joseph Mallan]

blackicmenace, on 20 June 2014 - 05:23 PM, said:

Valid points about accuracy although I always shake my head in wonder why at the start of a match people shoot dirt. I always take dirt shots as a bad omen of things to come.

Sometimes its just to check heat. I will rarely shoot the ground with my energy weapons but never waste ammo.

[blackicmenace]

Joseph Mallan, on 20 June 2014 - 05:25 PM, said:

Sometimes its just to check heat. I will rarely shoot the ground with my energy weapons but never waste ammo.

I check that sort of thing in the testing ground but thats me. How can you explain the people shooting AC shots in the dirt, that I don't get at all other than the sound and effects fascinate them.

[Joseph Mallan]

blackicmenace, on 20 June 2014 - 05:29 PM, said:

I check that sort of thing in the testing ground but thats me. How can you explain the people shooting AC shots in the dirt, that I don't get at all other than the sound and effects fascinate them.

Don't get me wrong, I do that also, it is mostly boredom that makes me pull the trigger on the poor defenseless ground. As for the AC shooters my opinion is a bit harsh... They are dumb.

[Wintersdark]

Just like the people firing AC's at targets *way* out of their max range.

But yeah, I'll often doodle on the ground with my lasers, and when a match is wrapped up but not yet over (chasing down the last couple mechs, waiting for Conquest points, that sort of thing, where I know I can no longer contribute) I'll often just spew ammo about for the lols.

I'm particularly fond of firing off huge clouds of LRM's all over the place to watch the pretty explosions in the sky as they reach their max range.

I don't care at all about my stats, however, and even if they were public, I'd prefer people assume I'm just outrageously bad.

[Joseph Mallan]

I just can't waste perfectly good ammo like that Winters!

[Fatal25]

Another solo spawn in River City Charlie Lance took 25% of my armor before I could get close to the other lances thanks to missile boats. Not blaming the LRM boats, they are a viable weapon that really does take skill to play it well (regardless of what some say). Even with speed tweak on the direwolf I could not get far enough out of range before I got 4-5 volleys on me. If there was a chance that I would not spawn in Charlie it would be acceptable, but knowing when I see River City map come up that I am most likely going to start off running for my life is lame. Again, I could just leroy jenkins and run towards the enemy, but in a pug match I guarantee that the three other players are going to make a bee line to the right and the other lances. The devs do seem to actually read the boards on occasion so I hope they take note of this issue and at least give it some consideration.

[Ngamok]

I am the mightiest scrub, not gonna lie.

[Wintersdark]

Wrote to Karl Berg (he's rewriting the Matchmaker now) with regards to this subject, and this was his answer:

Quote

Hi Wintersdark,

What happens is that the matchmaker currently produces games using some metric. Elo, weight classes, how long you've been sitting in the queue, group sizes, etc.. The actual matchmaker doesn't matter for lance allocation.

The *dedicated server* assigns lances once the game launches. It does so using a fairly simple algorithm that works something like the following:

Sort the groups on the team by group size, in the case where groups are the same size, sorting is random due to the sort algorithm not being a stable sort (http://en.wikipedia....ry:Stable_sorts).

Place the largest group in lance alpha
Place the second largest group in lance beta
Place the third largest group in lance charlie
From here, the algorithm simply continues
- For each remaining group: place the groups in the smallest lance, breaking any players that don't fit into a single lance across multiple lances if required

The dedicated server doesn't consider nor care about Elo during lance allocation. It's solely concerned about trying to keep pre-made groups together within a single lance if at all possible.

Thanks for bringing this to my attention however! Feel free to repost this explanation if you think it would help; or simply let me know if you think I should clarify anywhere specifically.

-Karl.

So, yeah, your Elo has no bearing on your placement in a match. So, while us Charlie Lance folks may still be scrubs, that's not why you're in Charlie lance. You're in Charlie lance because that's statistically the most likely place for you to end up, as The Evil Premades stole Alpha and probably Bravo lance.

Edited by Wintersdark, 25 June 2014 - 06:23 AM.

[Votanin FleshRender]

Wintersdark, on 25 June 2014 - 06:11 AM, said:

Wrote to Karl Berg (he's rewriting the Matchmaker now) with regards to this subject, and this was his answer:

[/size][/font]



So, yeah, your Elo has no bearing on your placement in a match. So, while us Charlie Lance folks may still be scrubs, that's not why you're in Charlie lance. You're in Charlie lance because that's statistically the most likely place for you to end up, as The Evil Premades stole Alpha and probably Bravo lance.

Something is still off then...

I've got a 1000 games under my belt, and I can count on one hand the amount of times I haven't been in Charlie as a solo pug. If Elo truly has nothing to do with it, that should be statistically nigh impossible. ( I'm certain I have a very low Elo). I didn't group much until lately, so better than 85% of those have to be solo. Less than 5 times in bravo out of 850ish games?

And every single time I did end up in bravo? Absolute roflstomps. I had assumed it was because somehow MM thought I was a better player than was (no idea why it would think that, my W/L is still a joke). If I drop in Bravo, I GUARANTEE a 2-12 game or worse.

[Wintersdark]

Votanin FleshRender, on 25 June 2014 - 06:41 AM, said:

Something is still off then...

I've got a 1000 games under my belt, and I can count on one hand the amount of times I haven't been in Charlie as a solo pug. If Elo truly has nothing to do with it, that should be statistically nigh impossible. ( I'm certain I have a very low Elo). I didn't group much until lately, so better than 85% of those have to be solo. Less than 5 times in bravo out of 850ish games?

Not so weird. If there are two premade groups in your team (note that the one-per-team rule was only around for a very brief time) then its very likely you'd be in charlie lance, with premades in Alpha and Bravo (and maybe alongside you in Charlie also).

Quote

And every single time I did end up in bravo? Absolute roflstomps. I had assumed it was because somehow MM thought I was a better player than was (no idea why it would think that, my W/L is still a joke). If I drop in Bravo, I GUARANTEE a 2-12 game or worse.

Correlation is not causation. Confirmation bias is very bad for such things as well.. Not to mention the greater likelihood that in these matches your team has only one or even no premade groups, which could have a substantial impact on team performance.

[Ngamok]

Wintersdark, on 25 June 2014 - 06:11 AM, said:

Wrote to Karl Berg (he's rewriting the Matchmaker now) with regards to this subject, and this was his answer:

[/size][/font]



So, yeah, your Elo has no bearing on your placement in a match. So, while us Charlie Lance folks may still be scrubs, that's not why you're in Charlie lance. You're in Charlie lance because that's statistically the most likely place for you to end up, as The Evil Premades stole Alpha and probably Bravo lance.

I can tell you that this is not the case. On my alt account, I have never, ever been in Alpha Lance and rarely in Bravo Lance. When I am dropping in it, 75% of the time I will be in Charlie Lance. This is Komagn. On Ngamok, I will end up in Alpha Lance if there is no 4 man there 75% of the time. How is it possible that two different accounts have such different lance placements? Something has to differentiate my two accounts?

Edit: clarification, solo.

Edited by Ngamok, 25 June 2014 - 07:15 AM.

[Ngamok]

Current stats:

Ngamok - 376 wins / 335 losses
Komagn - 24 wins / 38 losses

Archived stats:

Ngamok - 3531 wins / 3413 losses
Komagn - 53 wins / 42 losses

Edited by Ngamok, 25 June 2014 - 07:21 AM.

[wanderer]

It's not hard to understand if you bounce enough between two ELO "buckets".

Premades will be assigned to Alpha, then Bravo, then Charlie in that order (yes, all three if need be). Open slots go in order of highest ELO to lowest A-B-C - so if Alpha has a 4-man and Bravo a 3-man, the top soloer will be in Bravo,then remaining soloers Charlie.

I know when I'm in over my head when there's 8 random soloers above me- and I'm in Charlie.

[Thorqemada]

Fatal25, on 20 June 2014 - 09:06 AM, said:

I have been told that solo drops always spawn in Charlie Lance...

This is only a rumor - i drop solo since CB and most of the time i am placed in the Alpha Lance - so like others be placed in Alpha Lance also while soloing this rumor is now officially proofed wrong!

I have not found any coincidence between the quality of matches and may placing in Alpha, Bravo or Chalie Lance - albeit there is a strong coincidence with the Players in each Team!

[Votanin FleshRender]

Wintersdark, on 25 June 2014 - 06:58 AM, said:

Not so weird. If there are two premade groups in your team (note that the one-per-team rule was only around for a very brief time) then its very likely you'd be in charlie lance, with premades in Alpha and Bravo (and maybe alongside you in Charlie also).

Correlation is not causation. Confirmation bias is very bad for such things as well.. Not to mention the greater likelihood that in these matches your team has only one or even no premade groups, which could have a substantial impact on team performance.

So, in 850ish games, I've had 2 4-man premades on my side for 845ish of them? OR, the times that I don't, I get randomly put in charlie almost every time. Again, statistically very unlikely. What was your thread title again? 27 in a row pales in comparison to 845 of 850 (give or take a few).

I agree, the handful of times I've been in bravo is too small a sample size, though. But if my team has only one or no premades in those games, then neither should the red team... hence not a disadvantage to my side.

I'm just concerned there's something amiss perhaps the devs don't know about or realize.