49 replies to this topic

[phelan spider]

So I decided to test out ELO from the perspective of a new player and created this alt account. (I am a founder playing since last May).

Here are my observations:

1. majority of modes were conquest; no idea why (15 vs 6)
2. Teams often had 2-3 founders on them
3. About 20% the games clearly had 3 or 4 man premades; especially my first few
4. Very few other teammates had trial mechs; everyone had bought a mech already
5. Surprised by the number of hero mechs (2-3 each match)
6. I did not recognize any of the names (which I take as good)
7. Most games had some degree of organization; people knew what they were doing
8. I suicide heat-over-ride most matches; over 21 games, I went 6 W and 15 loss. I had two kills (I could not help myself).
9. PGI helps with pre-defined weapons groups but the trial mechs really do suck (except the player dragaon variant). The raven is particularly bad.


8.7MM CB to show for it; should buy a nice mech.

Best quote of the night "I have NARC so I will NARC and you guys LRM for the win" (Seriously, thats what he said)

When I asked players, many said they had been playing "for months" and several self characterized as "good players". Why would a new player, who should have terrible ELO, get matches like this?

My general conclusion is that ELO was matching a new player to much better players with much more experience. Either ELO is broken or there were just not many people on a Saturday night (prime time I assume).

[Royalewithcheese]

Haha I saw you suicide in one of my matches! I'm running a brand-new Jagermech which, as my first heavy, has a brand-new Elo. My guess is the founders/hero mechs/etc. in that bracket may be due to something similar. (Incidentally, how is it fair to put new players in stock RVN-4Xs capable of going 81 whole kilometers per hour up against people who are testing out their brand new Heavy Metal or AC20 boat?)

And yeah wow the stock 4X is bad. Why would anyone ever think an 81 KPH raven is a good idea for a trial mech?

Edited by Royalewithcheese, 06 April 2013 - 06:23 PM.

[Roadbeer]

Actually, you start out at the middle of the Elo range then adjust up/down from there, and I think it takes around 100 drops in the same weight class to have an 'established' Elo.

If I understand the system correctly.

[Kinilan]

21 matches isn't exactly a fair sample size and aside from the fact that Founder status is not an indication of skill most of your points can be chalked up to the fact that teams aren't filled within an Elo range. What the system appears to be doing is matching on a per player basis. so Red1 could have an Elo of 1640 and Blue1 could be 1600 Red2 could be 1200 and Blue2 could be 1210 Red3 could be 1800 and Blue 3 is a 1710 with Red and Blue 4 both being sub 1000 and so on.

It tried to match the whole team within a range to begin with but what we got was an endless string of "failed to find a match" followed by the hotfix to the matchmaker. Now it just tries to equalize the team Elo with no concern for the range.

And yes trials are garbage. New players have no in game documentation for anything. They can't even access the mechlab to see weapon stats let alone figure out what the various HUD symbols mean or how to set weapon groups. and even then Red1 with an Elo of 1300 could have their own mech built to their tastes with elited/mastered efficiencies and modules and Blue1 with an Elo of 1300 could be fresh off the Dropship with a trial AS7-K with it's four separate weapon systems and XL engine.

New players need to be segregated into their own matches for their own good.

[Royalewithcheese]

The one thing dropping with trial mechs really drives in is how little new players are shown about how to play the game. Haven't really had a chance to do that very much in a while, and it's really enlightening about the new player experience. You see stuff like someone spraying machineguns at 250 meters, or a light/medium parking in front of a target to brawl with it, or an Awesome just sitting there while ppl unload fire into its CT.

[deputydog]

So does every mech have its own ELO? no wonder do better in some. Its a slipery slope. The new mechs i get matched up with different groups so my matches arnt even close to the same...

[Belorion]

I think elo starts players off at the median level.

It should probably start people off at the 40 percentile mark.

[Royalewithcheese]

deputydog, on 06 April 2013 - 07:19 PM, said:

So does every mech have its own ELO? no wonder do better in some. Its a slipery slope. The new mechs i get matched up with different groups so my matches arnt even close to the same...

It's by weight class. So for instance if you spring for the giant pink poptart and it's your first assault, you have a new elo for the assault weight class.

[MonkeyCheese]

I was discussing this topic with a friend today. We are both founders we have both been around for ages

-more then 2000 matches played each since open beta
-kills in the 4 digits
-both have positive kdr
-I assume we both have exp totals over 1 million
-we both play all the weight classes

Not saying we are great or perfect or anything but we both do consistently decent. Yet I don't understand one bit why we constantly seem to see people on either team who are clearly completely new to the game and dont start me about all the trial mechs or the uneven number/weight matchmaking. The ELO matchmaker needs major work.

[Mechrophilia]

Belorion, on 06 April 2013 - 07:23 PM, said:

I think elo starts players off at the median level...It should probably start people off at the 40 percentile mark.

In my opinion, the Elo entry-level should be the very bottom.

[Belorion]

Mechrophilia, on 06 April 2013 - 07:35 PM, said:

In my opinion, the Elo entry-level should be the very bottom.

This would be abused by greifers that would just make a new account and reek havoc. ELO is a slow climb, so it should put you within easy reach of average.

[Wintersdark]

Yeah, Elo is by weight class not individual chassis. And starting Elo is roughly median, a little below.

It's been eye opening for me. I hadn't piloted a light since December, picked up a spider a couple days ago and wow, it's definitely a different place. I'd kind of forgotten the good things Elo has done for me.

Just the other day, I was spectating a dragon loaded with an lb-x ac10, machine gun, flamer, and lrm5. With a top speed of 58kph.

I haven't seen a build like that in a long time.

Since then, I've been doing my best in my stock ballistic spider scouting and relaying info (but not pursuing cap wins in assault matches) and thus losing a lot. It grows hairier all the time.

Now, I regularly see NARC equipped and used. It's shocking and weird, makes you wonder just how deep the rabid hole goes.

The problem though is that I won't deliberately lose and even with a basic spider you get to a point where you can't lose anymore.

[MischiefSC]

In addition to what everyone else has already stated, here's some other important factors:

Entry level is Elo 1300 - people who haven't won 30,50,100 games more than they've lost in the same weight class won't have moved enough in Elo to be out of range of 1300 completely.

[Primetimex]

Trial mechs suck full stop because they do not have double heatsinks - that fact alone is going to get you wiped vs experienced players - for newbies to have an even playing field they need good equipment, not gimped ones.

I do agree with Phelan Spider, newbies should be paired with other newbies so they are on same level playing field and not wait and get wiped countless times before they are where they should be - that is at the beginning with others with trial mechs!

[MonkeyCheese]

Primetimex, on 06 April 2013 - 09:27 PM, said:

Trial mechs suck full stop because they do not have double heatsinks - that fact alone is going to get you wiped vs experienced players - for newbies to have an even playing field they need good equipment, not gimped ones.

^ This is especially true for the current "Awesome" trial mechs without double heat sinks.

[phelan spider]

To the comments about starting at 1300, that was my point. Why would ELO match a brand new player to an average player who has many hundreds of games played? That just ensures the new play gets ROFL stomped in their first 20 games.

First, match all new players together when you can. Then match them to the bottom of the barrel.

I would not be worried about people creating accounts just to stomp newbs. Anyone who can endure those trials mechs deverse a hero biscuit (i.e. trolls would lose interest)

I did miss the idea that a newer player might have just bought into that weight class.

Phase 2: I will buy a mech and actually play some games and gauge the skill level. (The problem is a decent light with XL, endo, DHS, FF is gonna cost more than I have....maybe time to dust off a hunchie?) BTW I have 3000+ games and a 6 K/D ratio across lights and assaults.

PS Not to bring up an old topic, but given how bad trial mechs are, the only reason newbs WONT suicide every game for the first 20 cadet bonus games is that they do not know they can (or should). I was trying to get a lot of games to see match ups and simply could not bring myself to play anything other than the dragon. But why? Why subject someone to that as their entry into the game? Just stupid. I would tell ANY new player "Suicide every game until you get to 10MM CB; then buy a mech and see if you like the game"....nice message to build the community isnt it?

PS Sorry Royalewithcheese, hope you won despite me ;-)

Edited by phelan spider, 07 April 2013 - 04:05 AM.

[phelan spider]

OK. I decided to drop in dragon so I do not raise my L/M ELO for now.

First drop had 3 legendary and I recognized two names. These guys should NOT be dropping against a new player if ELO worked in any way shape or form. They were premade, in a hunchie pack, and tore things up (glad on my side). I am convinced ELO, at low period times like now, is simply pairing the empty queue up without regard for player skill.

[Aaren Kai]

I don't see why the match maker doesn't just use a multi tier approach. To keep it short, if your a new player with under x games AND a trail mech, you get placed in this bucket. The second you buy a customizable mech (probably at end of cadet bonus) you get placed into the regular Match making system.

To keep wait times down, fill in with bottom ELO players. Somewhat of a win-win for both groups.

[Anton Shiningstar]

Wintersdark, on 06 April 2013 - 07:45 PM, said:

Yeah, Elo is by weight class not individual chassis. And starting Elo is roughly median, a little below.

It's been eye opening for me. I hadn't piloted a light since December, picked up a spider a couple days ago and wow, it's definitely a different place. I'd kind of forgotten the good things Elo has done for me.

Just the other day, I was spectating a dragon loaded with an lb-x ac10, machine gun, flamer, and lrm5. With a top speed of 58kph.

I haven't seen a build like that in a long time.

Since then, I've been doing my best in my stock ballistic spider scouting and relaying info (but not pursuing cap wins in assault matches) and thus losing a lot. It grows hairier all the time.

Now, I regularly see NARC equipped and used. It's shocking and weird, makes you wonder just how deep the rabid hole goes.

The problem though is that I won't deliberately lose and even with a basic spider you get to a point where you can't lose anymore.

I am for Elo by Chassis not by class. I could have gained experience in a Cataphract that will not help in a Dragon. Now going from one Dragon to the next should lower my Elo by say 75 points? I have skills gained for the chassis over all but am new to the design I am trying.

[jeffsw6]

I imagine that the pool of players needs to be much larger for the match-maker to become more effective. I don't know how many people play at any time (I imagine PGI would want to keep this a secret) but there can't be more than a few dozen newbies on at any time, can there?

The "searching for game" thing usually takes just a few seconds. It seems like it places too much emphasis on quick drops and not enough on creating teams of similar skill. Would people be outraged if it took average a minute to find a match instead of the few seconds that is typical now? I am sure that is a question PGI spends time thinking about.