73 replies to this topic

[Jman5]

There has been a lot of discussion going on about wins and losses, both in the group queue and solo queue. One thing I like to do to better track my own information is to save all my end of round screens.

I decided to go through my last 250 medium class games (HBK-4J) and do a break down of how I am doing not just overall, but in both the solo and group queue. Just for the record, I am almost exclusively a medium pilot so going into this I probably had a pretty established Medium Elo.


TOTAL WIN LOSS TREND

Spoiler

The blue line is the trend. So if I won a game I got +1, if I lost a game I got -1. The red rectangles are the periods where I was playing in the group queue. My total win/loss ratio is 1.35.

-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

GROUP QUEUE WIN LOSS TREND

Spoiler

No surprise here. My win rate skyrockets in the group queue to 3.05 W/L. For those who don't know, I play for QQ Mercs. Aside from the occasional match against the SJR/EMP no-lube rapetrain, we usually play pretty well together.

-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

SOLO QUEUE WIN LOSS TREND

Spoiler

Honestly, I was not expecting it to be so even. 89 wins, 88 losses puts me almost exactly at a 1.0 W/L in the solo queue. I attribute the +0.01 above average to clearly show that hunchbacks are OP.
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

So here is an important question that is often debated on the forum. Was I unfairly punished in the solo queue because of my high W/L ratio in the group queue?

Sort of. I'm sorry if this next graph is a little dense, but I broke down Win/Loss Ratio between Group and Solo sessions. Green lines pointing up from green boxes are my solo play sessions. Red lines pointing down are my group queue play sessions. Red text is negative w/l ratio, green text is positive.
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

WIN LOSS RATIO BREAK DOWN

Spoiler

It seems that toward the end when I started playing more group queue, I started losing more than winning in the solo queue. However, the w/l ratio was rather gentle. It was usually just 1-2 game difference.


So there you have it. Matchmaker seems to do an admirable job in the solo queue, but struggles in the group queue. People who win a lot in the group queue seem to get punished for it in the solo queue, but it's not nearly as bad as people think.

Edited by Jman5, 23 March 2015 - 11:51 AM.

[norus]

Good god, it's almost as if... the people who complain about constantly losing aren't good! Conspiracy. Nice groups win rate though, just a good point that smaller groups stand no chance against non brain dead large drops.

[Ultimax]

Excellent data Jman, thanks for putting this out there.

[Impossible Wasabi]

The Space Pope approves!

Although, it makes him wonder why MM isn't just split up between solo drops and groups in regards to players stats.

Considering for instance that if a player usually rolls deep with an organized unit (say 6 or more players on average in a group when dropping) for Group drops, they probably have a great ELO, but for Solo queue their ELO might naturally be lower (less control of teammates/team composition/etc.).

Edited by The True Space Pope, 22 March 2015 - 04:15 PM.

[Hit the Deck]

Nice work, Jman! Thanks for the data!

[Kristov Kerensky]

Good work, thanks.

[Agent 0 Fortune]

Jman5, on 22 March 2015 - 03:55 PM, said:

So there you have it. Matchmaker seems to do an admirable job in the solo queue, but struggles in the group queue. People who win a lot in the group queue seem to get punished for it in the solo queue, but it's not nearly as bad as people think.

I was under the impression that Elo is only used and updated in the solo queue.
Which means you performance in the group queue has no bearing on your match quality in the solo queue.

[DDDs]

All I got from this was that Hunchbacks are OP. And I agree.

[Jman5]

Agent 0 Fortune, on 22 March 2015 - 04:30 PM, said:

I was under the impression that Elo is only used and updated in the solo queue.
Which means you performance in the group queue has no bearing on your match quality in the solo queue.

You get 4 Elo scores in this game. A light Elo, a medium Elo, a heavy Elo, and an Assault Elo. Matchmaker uses it in both solo and group queue. The difference in group queue is that it adds up and averages your premades individual Elo scores to figure out the general skill level of your group. It also does some fudging based on the size of your premade, but that's another story.

[Davers]

Jman5, on 22 March 2015 - 05:01 PM, said:

You get 4 Elo scores in this game. A light Elo, a medium Elo, a heavy Elo, and an Assault Elo. Matchmaker uses it in both solo and group queue. The difference in group queue is that it adds up and averages your premades individual Elo scores to figure out the general skill level of your group. It also does some fudging based on the size of your premade, but that's another story.

Well, it seems that Elo/MM is working for you. So they next question would be, what do you think of the quality of your games?

[Lynx7725]

The way I'd interpret is that your group win streaks will be followed usually by solo loss streaks, suggesting a causality. With a limited data set, it's hard to be sure though.

Was thinking whether I want to pull up my EOM screenshots to track my Hunchie record, but mine's solo queue only, so might not be worthwhile.

[Ultimax]

Lynx7725, on 22 March 2015 - 06:53 PM, said:

Was thinking whether I want to pull up my EOM screenshots to track my Hunchie record, but mine's solo queue only, so might not be worthwhile.

If yours are solo queue only, then couldn't you just use your stats page instead of EOM screenshots?

[Lynx7725]

Ultimatum X, on 22 March 2015 - 07:49 PM, said:

If yours are solo queue only, then couldn't you just use your stats page instead of EOM screenshots?

It's more about the rate of change rather than the final results; the stats page only give you the final, not the in-betweens. If the progression shows clear streaks, then there is evidence of possible causality.

[Jman5]

Lynx7725, on 22 March 2015 - 06:53 PM, said:

The way I'd interpret is that your group win streaks will be followed usually by solo loss streaks, suggesting a causality. With a limited data set, it's hard to be sure though.

Was thinking whether I want to pull up my EOM screenshots to track my Hunchie record, but mine's solo queue only, so might not be worthwhile.

It wasn't so much of a loss streak as it was slightly more losses than wins (usually). my w/l ratio didn't just fall off the face of the earth which is what some people seem to suggest.

You would probably want to do all your mediums in a given period of time instead of just hunchbacks since you have 1 Elo for all your mediums.

[Lynx7725]

Jman5, on 22 March 2015 - 07:54 PM, said:

You would probably want to do all your mediums in a given period of time instead of just hunchbacks since you have 1 Elo for all your mediums.

I made my life easier lately by just playing one medium...

[El Bandito]

Jman5, on 22 March 2015 - 03:55 PM, said:

So there you have it. Matchmaker seems to do an admirable job in the solo queue, but struggles in the group queue. People who win a lot in the group queue seem to get punished for it in the solo queue, but it's not nearly as bad as people think.

1.2 WLR in the Solo-Q over all. Take that, MM!

[Jman5]

El Bandito, on 22 March 2015 - 08:57 PM, said:

1.2 WLR in the Solo-Q over all. Take that, MM!

It's probably the mustache.

[KraftySOT]

Maybe its the booze, but uh, your experience in a single chassis has absolutely no bearing on the MM, which is completely counter intuitive to our type of the game in the first place.

I get it. Its throwing completely random levels of skill at each other. But this isnt even remotely scientific.

Edited by KraftySOT, 22 March 2015 - 11:59 PM.

[White Bear 84]

Ultimatum X, on 22 March 2015 - 04:04 PM, said:

Excellent data Jman, thanks for putting this out there.

Good info. Here is a thought to make the data more subjective though - run an equal amount of group and an equal amount of solo and see how they both compare. If you really want to go into detail can try different combinations e.g. Solo/Group/Solo etc to see what trends develop, or get community helpers to map this info out!

I think the graph would vary depending on mech too - say you are a good pilot, judging from your graphs you do alright; if you pilot an assault that does a tonne more damage, does this influence the growth of W/L - particularly in pugs. To what extent would your individual skill level & firepower influence the flow of the games.

So much potential.. ..would love it if PGI collected these metrics for public use e.g. pug w/l ratio in jager with elo > xx, or pug w/l ratio for all heavies etc etc etc...

[KraftySOT]

Lets just say, PGI has collected this metrics on a mass scale and they paint an entirely different experience than one witnesses statements.

Im just sayin.

ELO is bad. The MM is bad. But is this really scientific?