43 replies to this topic

[hammerreborn]

Quick and dirty because the site going down or something messed up my writeup.

http://mwomercs.com/...ost__p__1672291 for Previous Trials and data related to the new user experience and Mech Diversity

ECM/Premade effect on Win Percentage and Prevelance (40 trials):

All 40 trials were done in my Jenner D from Thusday 1/31 from roughly 6pm EST - 12 EST (Assault only). All premades were self-reported (thanks guys to all those who did!). As they were self-reported, numbers may be off.

Shocking details observed outside of data:
-Lots of non-ECM lights abound, a bit of the Spider 5V, but a lot of 2Xs and 1B commandos running around. Also, Jenners are back in style <3

-Awesomes, way more than I'm used to seeing. Inflated by current trial, but still lots of 9Ms and other running around probably in preparation for the Hero mech

-Fatlas, Fatlas everywhere. Assaults easily dominated the numbers game (probably at least around 4 a game each side).

And now for the data. If the images suck I'll reupload either in the morning before work or after.

Enjoy! And please keep discussion civil.





*Notes* the Equal Pre More ECM appears to be wrong (as it should just be the two numbers below it combined). Therefore it should be 8 wins out of 14 (dropped any games with equal ECM) which ends up with a win rate of 57%.

[color="#b27204"]http://mwomercs.com/...ost__p__1687963[/color]

In my last trials the breakdown of ECM mechs was the following:
32% are Ravens
12% are Commandos
8% are Cicadas
49% are Fatlases

In this trial the breakdown was:
24.14% are ravens
11.49% are Commandos
2.3% are Spiders (Poor Spider )
9.2% are Cicadas
52.87% are Fatlases

So Fatlases have increased overall by 3% of all ECM mechs, 2% of 3L pilots are trying their hand at the Spider, and some people have read Garth's posts and are playing the Cicada Mail Delivery System more often (also possibly due to the increased benefit of lasers with recent netcode changes). Nearly 10% of 3L pilots have disappeared.

Also, in previous trials, about 16% are driving ECM mechs, and in these 40 trials only ~14.5% were ECM equipped (though there were the 3 mentioned ECM equippable mechs that could have run ECM but weren't), so ECM use seems to be dropping if only slightly (most likely due to the loss of the lighter ECM mechs).

Summary: As I've stated before, DC/AFKs > Premades > ECM > Pug in terms of success rate of a game.

Edited by hammerreborn, 01 February 2013 - 10:55 AM.

[Humboldt]

So only 30% of the matches had no premade in it?

[Snowhawk]

Hmmm... did I read this correct? 85% win rate for the premades??

This is possible, because there is the pareto principle. With a bit effort (Teamspeak) you can gain 80% success rate....

[Snowhawk]

And where are the premades now who told us that 4 man groups have no impact on a match or can not decide it...?

[One Medic Army]

Snowhawk, on 01 February 2013 - 02:18 AM, said:

And where are the premades now who told us that 4 man groups have no impact on a match or can not decide it...?

Still trying to figure out exactly what I'm looking at with the numbers, there's quite a few entries alongside that premade 85% win ratio. Like the DC loss rate 100% for instance.
Are premades more likely to win since it reduces the chances of a DC?

[Apoc1138]

One Medic Army, on 01 February 2013 - 02:23 AM, said:

Still trying to figure out exactly what I'm looking at with the numbers, there's quite a few entries alongside that premade 85% win ratio. Like the DC loss rate 100% for instance.
Are premades more likely to win since it reduces the chances of a DC?

Valid point, there were 2 matches where there was a disco as well as a premade vs. pug win

I'm not sure how the OP got 85% as the premade vs. pug score either, because there were 18 matches where a premade won against a pug team and 3 where this wasn't the case, which I make 83.3%

but removing the 2 disconnect matches entirely, it becomes 13/16 which is still 81%

the weird thing is, I play in premades quite alot, and when we have new members on, trying to get them through the basics of moving and firing, we lose alot of matches... but then I can also understand the argument that the longer you are in the game, the more likely you are to cave in to the anti-premade mentality and actually try it out - so you've mostly got experienced players who team up vs. new/lesser experienced players who mainly PUG

hopefully matchmaking phase 3 (due out this month) will have positive effect here

also, as the OP points out, he only has a record of the premades that were good enough to admit to being so - we don't know and have no way of verifying how many of these matches had non-declared premades

[Skoll Lokeson]

hammerreborn, on 31 January 2013 - 10:48 PM, said:

Shocking details observed outside of data:
-Lots of non-ECM lights abound, a bit of the Spider 5V, but a lot of 2Xs and 1B commandos running around. Also, Jenners are back in style <3

Players forced to grind their non-ecm variants (2 for every 1 w/ ECM) to progress the ECM variant. Some have max it out and are just playing casually while grinding other mechs. Also, decrease in ECM reduces the need to bring ECM to counter ECM and so on (arms race in reverse[?]).

[hammerreborn]

Humboldt, on 31 January 2013 - 11:00 PM, said:

So only 30% of the matches had no premade in it?

Yes, out of the 40 trials only 30% had no premades on either side confirmed. This could be lower, but this is a good maxima (I wouldn't be surprised if some of the enemy teams lied with a majority of the nasty comments and herp derp what's a premade, I have a premade in my pants, I'm a solo premade comments sent my way while doing this).

Also, technically one of the last Friendly premades actually was me and a friend who didn't queue up together but when we realized we were in the same game we joined the same TS. Also one of the first friendly premades ended up being a 5 man sync drop because I got randomed into a 4 man drop of fellow Robinson Rangers and joined their TS channel for the duration of the match.

Snowhawk, on 01 February 2013 - 02:18 AM, said:

And where are the premades now who told us that 4 man groups have no impact on a match or can not decide it...?

I don't think anyone has ever said that 4 man groups don't influence the match. But as I said in my other writeups, communication is really the big key to success. Out of all 40 trials only one person took the command chair, and I'm pretty sure we won that match. With integrated VOIP and ELO matchmaking, I think we'll see the premades success rate drop drastically, but still win more often than lose.

Also, the majority of premades appeared to be 2-3s rather than straight up 4s for those who also admitted to how many they had. Though in certain matches I would end up with two groups of two on my team.

Apoc1138, on 01 February 2013 - 02:47 AM, said:

Valid point, there were 2 matches where there was a disco as well as a premade vs. pug win

I'm not sure how the OP got 85% as the premade vs. pug score either, because there were 18 matches where a premade won against a pug team and 3 where this wasn't the case, which I make 83.3%

but removing the 2 disconnect matches entirely, it becomes 13/16 which is still 81%

the weird thing is, I play in premades quite alot, and when we have new members on, trying to get them through the basics of moving and firing, we lose alot of matches... but then I can also understand the argument that the longer you are in the game, the more likely you are to cave in to the anti-premade mentality and actually try it out - so you've mostly got experienced players who team up vs. new/lesser experienced players who mainly PUG

hopefully matchmaking phase 3 (due out this month) will have positive effect here

also, as the OP points out, he only has a record of the premades that were good enough to admit to being so - we don't know and have no way of verifying how many of these matches had non-declared premades

18/21 = 85% 18 wins, 3 losses is 21 games. Removing the 2 DC/AFK matches brings it to 16/19 which is still 84%.

Skoll Lokeson, on 01 February 2013 - 06:09 AM, said:

Players forced to grind their non-ecm variants (2 for every 1 w/ ECM) to progress the ECM variant. Some have max it out and are just playing casually while grinding other mechs. Also, decrease in ECM reduces the need to bring ECM to counter ECM and so on (arms race in reverse[?]).

Granted, and I touched point in that in http://mwomercs.com/...ost__p__1680369, but while I don't have any concrete numbers this time around, there was a significant difference. The raven was not anywhere close to the nearly 60% of lights. With a greater influx of commandos (effect of death's knell) and Jenners (netcode fixes) and even a few spiders here and there the Raven 3L is slowly but steadily losing it's iron grip on the light champion.

I think what I can say even without the actual data is that releasing these hero mechs are actually making people flock to the chassis, which I think is a great thing for diversity in the long run.


Something I wish I could have done but don't have the manpower to accurately do during these matches would be to record how many trials were on each team. But as I can't target my own team, and really can't stop and analyze each enemy mech and determine if its a trial or not, I couldn't use that as a way to determine trials effects on how the games turn out.


Thank you all for your replies.

Edited by hammerreborn, 01 February 2013 - 07:54 AM.

[Apoc1138]

hammerreborn, on 01 February 2013 - 07:51 AM, said:

18/21 = 85% 18 wins, 3 losses is 21 games.

doh, my bad

Edited by Apoc1138, 01 February 2013 - 07:55 AM.

[hammerreborn]

It's all good, I did do the mathy part at like 1am so I wouldn't have been surprised that I missed a win or a loss. The way I noted wins wasn't exactly as easy to digest as I originally hoped.

And either way it made me check all my numbers and noticed the Equal Pre More ECM appears to be wrong (as it should just be the two numbers below it combined). Therefore it should be 8 wins out of 14 (dropped any games with equal ECM) which ends up with a win rate of 57%. So having more ECM (with relatively few trials, granted) appears to give you just the slightest edge in a pure PUG or a premade on premade battle, in which relative skill was probably the bigger indicator of success.


Edit: Also, there were only 3 8-0 (one was 7-0) games total, so depending on your definition of PUG stomping that doesn't seem to be as frequent as some would like you to believe. Same with the only 2 cap games


Edit2:

Quote

Players forced to grind their non-ecm variants (2 for every 1 w/ ECM) to progress the ECM variant. Some have max it out and are just playing casually while grinding other mechs. Also, decrease in ECM reduces the need to bring ECM to counter ECM and so on (arms race in reverse[?]).

http://mwomercs.com/...ost__p__1687963

In my last trial the breakdown of ECM mechs was the following:
32% are Ravens
12% are Commandos
8% are Cicadas
49% are Fatlases

In this trial the breakdown was:
24.14% are ravens
11.49% are Commandos
2.3% are Spiders (Poor Spider )
9.2% are Cicadas
52.87% are Fatlases

So Fatlases have increased overall by 3% of all ECM mechs, 2% of 3L pilots are trying their hand at the Spider, and some people have read Garth's posts and are playing the Cicada Mail Delivery System more often (also possibly due to the increased benefit of lasers with recent netcode changes).

Also, as in previous trials, about 16% are driving ECM mechs, which didn't really change in these tests. So it's not that less people are driving ECM.

Edited by hammerreborn, 01 February 2013 - 08:59 AM.

[Mechwarrior Buddah]

a whole 40 sample group huh? Game must have less players than we thought if thats relevant

[Elkarlo]

More intrestingly in this Data:
An AFK/DC Mech got a higher Impact then a Premade.
Every Game with an AFK/DC Mech was lost, even when it was on the side of the Premade.
(But it is not impossible to win against odds even had a 8vs6 win.)
But only 5 Games with uneven players, so not realy stastical hard.

In 40 Games had been no (confessing) dual Premade Enemy.

[Bguk]

Mechwarrior Buddah, on 01 February 2013 - 08:56 AM, said:

a whole 40 sample group huh? Game must have less players than we thought if thats relevant

More than 0 samples of people stating premades have/have no effect, ECM has/has no effect, trials have/have no effect, etc. One has to start somewhere.

Edited by Bguk, 01 February 2013 - 09:02 AM.

[hammerreborn]

Elkarlo, on 01 February 2013 - 08:59 AM, said:

More intrestingly in this Data:
An AFK/DC Mech got a higher Impact then a Premade.
Every Game with an AFK/DC Mech was lost, even when it was on the side of the Premade.
(But it is not impossible to win against odds even had a 8vs6 win.)
But only 5 Games with uneven players, so not realy stastical hard.

In 40 Games had been no (confessing) dual Premade Enemy.

http://mwomercs.com/...mech-challenge/

Another 100 or so trials, and mainly the same thing with disconnects. Only two premades (off the top of my head) were able to overcome being down a man.

Being able to overcome AFKs also probably has to do with what the AFK was. Losing a spider isn't as bad as one match last night where their only DDC was the AFKer. You just can't overcome the loss of not only ECM but your biggest mech without the other team being abysmal, great coordination, or just pure luck.

And there were confessed sync droppers. Once it was me (technically a team of 5 but I randomed into their group than syncing), and another was a confessed 5 man premade. I took what people said with a heavy grain of salt. There were enough idiots saying they were an 8 man premade (and I was on their team) to really believe every single claim of sync drops.

Quote

a whole 40 sample group huh? Game must have less players than we thought if thats relevant

http://mwomercs.com/...mech-challenge/ Theres a lot more data here. Just wasn't always collecting the same thing, as trying to do too much during a match is not only a detriment to my team but also really really obnoxious to do.

Though if you think you can do better by all means DIY and post your stats in here and I'll compile it with my own.

Edited by hammerreborn, 01 February 2013 - 09:14 AM.

[Spot]

Im realy just tired of people blameing team work for there losses. Its a weapon in all of our arsenals.

[hammerreborn]

Spot, on 01 February 2013 - 09:36 AM, said:

Im realy just tired of people blameing team work for there losses. Its a weapon in all of our arsenals.

Exactly, but some people have the tools necessary to use those weapons much easier than others (voicechat), and people willing to listen (premade). But teamwork and communication are pretty much great ways to guarentee yourself a win.

[hammerreborn]

Updated and tidyed up OP.

[Turook]

Very interesting.

[Ryvucz]

Interesting indeed.

[Thirdstar]

Despite our recent disagreements I still find your zest to gather data very heartening.

I expect to continue to see Awesomes in some quantity for the entire life of MWO, due to currently being the cheapest assault, frequency as a trial choice and the 9M variant.

The shift away from Ravens is probably due to the stripping (in part) of the lagshield.

The shift TOWARDS Fatlasses is also probably because the lagshield fixes.

I wonder if a reconnect option will significantly effect metrics, my gut says yes. I also expect 12 vs 12 to reduce the effect of 1-2 DCs/AFKs.

Your summary of course is spot on.