Matchmaking Phase 4 Follow-Up - Feedback
#181
Posted 27 May 2013 - 06:39 AM
#182
Posted 27 May 2013 - 05:36 PM
Had a few games in the last day or two where new players were on my team. I have some 3.5k games - even with an average elo I shouldn't see anyone with less than say 50 games under their belt.
These people were asking in chat how to lock on, fire etc. They shot LRM's at targets 2k away on Alpine. The tried their medium lasers and SRM at 800m etc. Each time we were trying to help them and walk them through how to play. However it really isn't fair on them as just as they start understanding how to shoot they get killed.
I'm happy to assist them, but it shouldn't be required really. Although I know there just probably isn't a way to do it. Even big MMO's run into issues in their starting zones after people have moved on. So there just will never be enough new starters to form their own queue.
The only way around it really is to lower the threshold even further on the elo scores, then work with some kind of standard deviation rather than group totals. Not sure what the solution is but it still needs tweaking that's for sure.
#183
Posted 27 May 2013 - 09:42 PM
A graph can mean anything you want it to if you create the context to fit the numbers. I would be more interested in seeing how this worked out across the generally accepted 'primetimes'. Pewrsonally I'm not sure what my ELO score is(can I even find out?) but it should be low all things considered... that said I'm still droppping in games where a four man premade meets me and a gaggle of PUGs and Trials at least one in seven games(I do keep count) but that's playing in the 'primetime' for a GMT +10 area.
#184
Posted 29 May 2013 - 11:29 PM
Hauser, on 20 April 2013 - 11:08 AM, said:
It's called combat loss grouping.
After a bit of a back and forth a few members on each team have a red torso and it just takes a pop to make them go. But until that happens they are still contributing fire power. The one team that gets two of those to go they get a big advantage in numbers. Big enough to keep their weakened guys in the rear so the other team can't make a comeback.
Using wins and losses, not their ratio is fine for elo. It measures the one result that counts.
Why do you want to know your elo? What are you going to do with it?
Still screwed. Too many matches are 8:2 or worse. a perfect balance would by 8:6 or higher. 8:3 or higher I'd consider acceptable. As for showing it. Borrow from the MRB. Put a grade on your profile F- to A++. I just got told ELO was working as it was a perfect bell curve. That's a crock, it means it's distributed and evenly divided, it doesn't mean it provides a good experience. It doesn't.
#185
Posted 29 May 2013 - 11:40 PM
Redoxin, on 20 April 2013 - 01:56 PM, said:
.....
But ratios of 8:2 or worse should not be the MAJORITY of match results.
#186
Posted 31 May 2013 - 09:13 AM
I played about 40 rounds in Cicadas and had an average 415-420 dmg/match, 4:1 W/L and a 4+:1 K:D, averaging 1.25 kills/match. This is after going through basic+elites in BJs with similar results. Practically every round I was top or 2nd in match score/damage/kills. When I dropped, it dropped almost immediately, with a 3 second Searching pause at the most.
CICADA CDA-2A
20 matches
16 Wins
4 Losses
4 W/L
26 Kills
6 Deaths
4.33 K:D
8363 Damage
1.3 Kills/match
418 Damage/match
CDA-2B was very similar
CICADA CDA-2B
20 matches
16 Wins
4 Losses
4 W/L
24 Kills
6 Deaths
4 K:D
8324 Damage
1.2 kills/match
416 Damage/match
Then I played about 40 rounds in a 4 man. We noticed an extended time to actually find a match, usually 30 seconds to a minute.
We must have a similar ELO based on the fact that when we were all online and solo dropping or dropping in other groups we'd see each other with no sync dropping involved. Several other names on the server list are familiar from previous solo dropping as well. But who knows if ELO is really working.
In the 4 man I was barely able to average around 200dmg/match, 1.45:1 W/L and a 1:1 K:D and .5 kills/match. I was struggling to stay in the top 4 of the team. Are we just playing against teams that know not to ignore a ML boating medium?
We were usually the last 4 to die on our team as we would get stuck with pubbies that would just throw away their mechs. If we charged up to support the pubbies, we'd get wiped pretty quick as well. There were also several matches that were extremely unbalanced tonnage wise. For example 2 that stand out were a team with 4 lights, 2 mediums, 2 assaults against a team with 4 assaults, 3 heavys and 1 medium. Or 2 assaults 1 light and 5 mediums against 5 assaults, 3 heavies. THis actually seems to happen quite often when in a 4 man premade. I wish I had kept track of tonnage because it seems tonnage is off more when in a premade. Matches also seemed to be one sided more often.
4 man premade stats:
2a Delta from solo:
22 Matches
11 Wins
11 Losses
1:1
13 Kills
11 Deaths
1.18 K:D
3943 Damage
.59 Kills/match
179.2 Dmg/match
2B
21 Matches
12 Wins
9 Losses
1.3 W/L
12 Kills
11 Deahts
1.09 K:D
3582 dmage
.57 Kills/match
170.6 Dmg/match
After that I dropping solo some more, and went back to the previous solo pattern of high W/L and K:D ratios, and always being the top of the team/server scores. So if ELO is supposed to keep W/L around 1:1, it seems like it works in groups, but as a solo drop I am getting places with players where I carry the team.
What I want is to play against a more challenging team, but without playing on a derp team. I don't think a medium should constantly be at the top of the damage/kills list match after match. Medium is more of a support mech.
After dropping solo again:
Delta from previous stats:
2A
16 Matches
12 Wins
4 Losses
3 W:L
18 Kills
5 Deaths
3:6 K:D
6240 Damage
1.13 Kills/match
390 Dmg/Match
2B
15 Matches
13 Wins
2 Losses
6.5 W/L
16 Kills
5 Deaths
3.2 K/D
6075 Damage
1.2 Kills/match
380 Damage/match
Going back to a 4 man premade went back to losing frustratingly bad games of losing and dying. Its not like this just with my cicadas, its been like this for about a month with any mech I play. Solo, we all pretty much own the battlefield, but in a group we are the best on our team, but overall its like its just my 4 man premade against the other 8.
You could say that the 4 man I was in does not work well as a team, but before these matchmaking 'fixes' went into effect we routinely average at least a 3:1 W/L and similarly high K/D ratios. The 2 assaults in my premade were always at the top of the scoreboard (same is true if I dropped as an Assault and not a medium) and the way it is now and when 1/2 your team throws away their mechs there is only so much you can do. I think the pool (playerbase) the ELO is pulling from is just too small.
edit to fix some values I copied wrong. Had the wrong cells highlighted in the spreadsheet.
Edited by frag85, 31 May 2013 - 11:15 AM.
#187
Posted 31 May 2013 - 10:14 AM
frag85, on 31 May 2013 - 09:13 AM, said:
I played about 40 rounds in Cicadas and had an average 415-420 dmg/match, 4:1 W/L and a 4+:1 K:D, averaging 1.25 kills/match. This is after going through basic+elites in BJs with similar results. Practically every round I was top or 2nd in match score/damage/kills. When I dropped, it dropped almost immediately, with a 3 second Searching pause at the most.
CICADA CDA-2A
20 matches
16 Wins
4 Losses
4 W/L
26 Kills
6 Deaths
4.33 K:D
8363 Damage
1.3 Kills/match
418 Damage/match
CDA-2B was very similar
CICADA CDA-2B
20 matches
16 Wins
4 Losses
4 W/L
24 Kills
6 Deaths
4 K:D
8324 Damage
1.2 kills/match
416 Damage/match
Then I played about 40 rounds in a 4 man. We noticed an extended time to actually find a match, usually 30 seconds to a minute.
We must have a similar ELO based on the fact that when we were all online and solo dropping or dropping in other groups we'd see each other with no sync dropping involved. Several other names on the server list are familiar from previous solo dropping as well. But who knows if ELO is really working.
In the 4 man I was barely able to average around 200dmg/match, 1.45:1 W/L and a 1:1 K:D and .5 kills/match. I was struggling to stay in the top 4 of the team. Are we just playing against teams that know not to ignore a ML boating medium?
We were usually the last 4 to die on our team as we would get stuck with pubbies that would just throw away their mechs. If we charged up to support the pubbies, we'd get wiped pretty quick as well. There were also several matches that were extremely unbalanced tonnage wise. For example 2 that stand out were a team with 4 lights, 2 mediums, 2 assaults against a team with 4 assaults, 3 heavys and 1 medium. Or 2 assaults 1 light and 5 mediums against 5 assaults, 3 heavies. THis actually seems to happen quite often when in a 4 man premade. I wish I had kept track of tonnage because it seems tonnage is off more when in a premade. Matches also seemed to be one sided more often.
4 man premade stats:
2a Delta from solo:
20 Matches
11 Wins
11 Losses
1:1
13 Kills
11 Deaths
3943 Damage
.65 Kills/match
182 Dmg/match
2B
23 Matches
12 Wins
9 Losses
1.3 W/L
12 Kills
11 Deahts
1.09 K:D
3582 dmage
.52 Kills/match
155 Dmg/match
After that I dropping solo some more, and went back to the previous solo pattern of high W/L and K:D ratios, and always being the top of the team/server scores. So if ELO is supposed to keep W/L around 1:1, it seems like it works in groups, but as a solo drop I am getting places with players where I carry the team.
What I want is to play against a more challenging team, but without playing on a derp team. I don't think a medium should constantly be at the top of the damage/kills list match after match. Medium is more of a support mech.
After dropping solo again:
Delta from previous stats:
2A
16 Matches
12 Wins
4 Losses
3 W:L
18 Kills
5 Deaths
3:6 K:D
6240 Damage
1.13 Kills/match
390 Dmg/Match
2B
15 Matches
13 Wins
2 Losses
6.5 W/L
16 Kills
5 Deaths
3.2 K/D
6075 Damage
1.2 Kills/match
380 Damage/match
Going back to a 4 man premade went back to losing frustratingly bad games of losing and dying. Its not like this just with my cicadas, its been like this for about a month with any mech I play. Solo, we all pretty much own the battlefield, but in a group we are the best on our team, but overall its like its just my 4 man premade against the other 8.
You could say that the 4 man I was in does not work well as a team, but before these matchmaking 'fixes' went into effect we routinely average at least a 3:1 W/L and similarly high K/D ratios. The 2 assaults in my premade were always at the top of the scoreboard (same is true if I dropped as an Assault and not a medium) and the way it is now and when 1/2 your team throws away their mechs there is only so much you can do. I think the pool (playerbase) the ELO is pulling from is just too small.
This is exactly what I've been going through. I'm trying to lose a lot to see if it makes a difference and so far it hasn't.
#189
Posted 01 June 2013 - 04:06 AM
Hauser, on 21 April 2013 - 02:06 PM, said:
Now you're treating elo as a personal ranking system rather then a component of the match making system. The critical difference is that while the match maker does require some measure of skill to assign players to a match, it does not have to be an accurate assessment of that person. The only thing the matchmaker needs is a rating that describes how the player plays.
The situation you describe is already highly improbable in a random match making scenario. You must be the (un)luckiest player in the world to drop with such good/bad people. However if both are consistently dropping with 4 friends that are significantly better or worse, then their ratings are deserved. It accurately describes how the whole group will preform.
As I explained previously to Christof, given a random match maker good players will encounter more matches where their team has the advantage. The good players team will always have 1 good and 7 randoms, while the other team has 8 randoms and thus a lower chance of getting equal or more good players.
Elo is an improvement on random match making because it puts the players who would lose allot w/ random match making with similar players, just as it puts people who would win allot together. The result being that for every
1. A question. Does ELo take into account a mech you're playing in?
There's a simple and real example of the situation I'm currently stuck in. Last evening I played a lot of matches both alone, in 4 people group and 8 people group. Winned a lot in a row. But the trouble is, that I'm pumping up my Blackjacks.
now I've played alone like a dozen of matches in BJ-1DC with unlocked basic, and players, I'm getting into my team, are pretty much unskilled. I'm consistently killing at least 1 enemy every match, average number is 2 kills, coupla assists and over 300 damage, but my team is so incompetent, that I'm loosing like 4 of 5 matches no metter how hard I try. You know.. it's a Blackjack... not a Stalker... so I'm loosing and killing and dying again and again, but my team does not become better. I'm not trying to blame them, They're just obviously noobs, that's it. And It leads me to the second question.
2. should I stop killing and damaging to get a descent team? Or the only way out is to drop in a group of good players?
Edited by Rubidiy, 01 June 2013 - 04:20 AM.
#190
Posted 01 June 2013 - 04:19 AM
Rubidiy, on 01 June 2013 - 04:06 AM, said:
2. should I stop killing and damaging to get a descent team? Or the only way out is to drop in a group of good players?
You have different Elo score for each weight class, which seems odd, considering that your skill in a CDA will probably be different from your skill in a HBK, for example (or AWS vs. AS7, or DRG vs. CTF, etc.).
What happens on any particular day or gaming session may be more of a "snapshot in time" and not reflect the bigger picture. I track my stats pretty closely by the week, to try and judge if I am improving, but I try not to dwell too much on any one particular instance, as the player base at a certain time of day or day of the week might be dramatically different, or I/my team might be having a bad/good/lucky/unlucky streak.
#191
Posted 01 June 2013 - 04:54 AM
Kageru Ikazuchi, on 01 June 2013 - 04:19 AM, said:
The thing is, that I suspect, that matchmaking equals high skill, say, in a CDA and high skill in a STK. So you keep getting into nooby team against nooby team, BUT you're in CDA, while an equally skilled player is in STK. And this skilled STK demolishes my team long before I can do same thing to his team.
I think, there should be a multiplier for each weight class.
Edited by Rubidiy, 01 June 2013 - 06:30 AM.
#192
Posted 01 June 2013 - 03:33 PM
Does Elo also take into consideration the tons you are taking into that particular match?
#193
Posted 01 June 2013 - 03:41 PM
How effective both are depends on the time of day and availability of the players.
#194
Posted 01 June 2013 - 07:42 PM
One of the things I've observed it that it will fill one team before the other, so the variance in Elo scores when the player queue is small may be pretty wide.
#195
Posted 03 June 2013 - 11:15 PM
#196
Posted 23 June 2013 - 03:21 AM
Could of course be some bad luck streak i am having since 2-3 weeks now, but i´ve heard the same complaint from a lot of other people so i doubt that it´s Murphys Law... which leaves us with....hmmm maybe... could it be the matchmaker? ;-)
#197
Posted 23 June 2013 - 11:08 PM
At least from the higher side of the ELO system it appears to work, opponents are quality as well usually, never rolling over without a fight. The weight balancing could be a little better, but mostly only breaks down when a full 8 setup decides to bring something extreme during lower participation hours.
(Stability has also improved greatly; disconnects now appear to be contained to a player choice of leaving during match launch).
Thumbs up!
#198
Posted 25 June 2013 - 08:53 AM
#199
Posted 25 June 2013 - 11:06 AM
The issue with the bad weight matching and long wait times that this switch over somewhat solved did not have to do with the Elo spread of players, but with the match maker coding. It should be trivial to change the match maker to accept a larger spread of Elo for a match, rather then reducing the spread of the players Elo. Making the Elo rating less representive of actual player skills may have solved the problem, but it would have been a lot better to solve it at the root of the issue and that is the match maker code.
I wonder what the point of the current Elo is? It obviously doesn't do a good job at matching evenly good players against each other, so why do we have it? Wouldn't it be better to just go back to random games as the current system is just random anyway?
The only ones the current match maker actualy works for is the top 5-10% and the bottom 5-10% as they may be getting matched up with peers of skill more or less. The rest are just bunched up in the middle and face each other with no concern if they are noobs or veterans.
#200
Posted 30 June 2013 - 06:43 AM
1 user(s) are reading this topic
0 members, 1 guests, 0 anonymous users