136 replies to this topic

[Trauglodyte]

Well, Al, I'll posit a question: Do you want/prefer matches in a mech in which you've mastered? Or, are you just wanting matches? I'm working on my Atlas right now so I can provide you that data. Or, I can hop in something in which I'm really comfortable and good piloting.

Your answer will have an impact on the data, of course.

[PEEFsmash]

Alistair Winter, on 27 September 2013 - 12:24 PM, said:

So... who's going to post the results from 20 more matches?

Don't do it for me.

Do it for science.

Spoiler

Do I have to solo pug? I fail to find match about 80% of the time when solo pugging so that would be an all-day and rather miserable event.

[Shadey99]

I do have some results (20 in Spiders), but I need to input them all in...Then onto Cicadas... Then Jagermechs... Then Highlanders...

Oh And I forgot to add tonnage to my stats earlier... fixed now...

[RandomLurker]

Trauglodyte, on 27 September 2013 - 12:42 PM, said:

Well, Al, I'll posit a question: Do you want/prefer matches in a mech in which you've mastered? Or, are you just wanting matches? I'm working on my Atlas right now so I can provide you that data. Or, I can hop in something in which I'm really comfortable and good piloting.

Your answer will have an impact on the data, of course.

It will have an impact on YOUR data. Which is only one part of the overall picture.

A good sample requires roughly 1000 entries. Just do what you do and then submit the data, and the math will take care of the rest.

[Alistair Winter]

Trauglodyte, on 27 September 2013 - 12:42 PM, said:

Well, Al, I'll posit a question: Do you want/prefer matches in a mech in which you've mastered? Or, are you just wanting matches? I'm working on my Atlas right now so I can provide you that data. Or, I can hop in something in which I'm really comfortable and good piloting.
Your answer will have an impact on the data, of course.

I don't imagine that it will have a big impact on the data whether you've mastered it or not. But if it does, then that in itself is an interesting finding.

PEEFsmash, on 27 September 2013 - 01:15 PM, said:

Do I have to solo pug? I fail to find match about 80% of the time when solo pugging so that would be an all-day and rather miserable event.

Well, if you decide to run 4-man groups of Atlas D-DCs, then that will probably make the data less representative. So far, most teams have an average of 2 assault mechs.

****, I just realized it would be very useful to compare the sum of weight on each team, in tonnes, not just number of each weight class.

But anyway, it'll be interesting to have different samples. Then we can maybe discuss why they're so different, if they are.

[dustNbone]

Another interesting, but potentially more difficult to obtain statistic would be presence/absence of ECM on a team (and how many), and how that relates to win/loss and kill stats. It might just be me, but ECM seems to play a big part in things now, especially with the tendency for things to erupt into a close range brawl these days.

[Alistair Winter]

dustNbone, on 28 September 2013 - 07:07 AM, said:

Another interesting, but potentially more difficult to obtain statistic would be presence/absence of ECM on a team (and how many), and how that relates to win/loss and kill stats. It might just be me, but ECM seems to play a big part in things now, especially with the tendency for things to erupt into a close range brawl these days.

Good point. Worth checking, at least.

[RF Greywolf]

Hopefully after work today I can get my 20 matches compiled and put into a spreadsheet. Anything super specific anyone wants above the norms, at least that I can tell from an end game screenshot?

[Shadey99]

dustNbone, on 28 September 2013 - 07:07 AM, said:

Another interesting, but potentially more difficult to obtain statistic would be presence/absence of ECM on a team (and how many), and how that relates to win/loss and kill stats. It might just be me, but ECM seems to play a big part in things now, especially with the tendency for things to erupt into a close range brawl these days.

While hard to factor in if they actively have ECM or not, my stats do take into account which variant is played. So they will show potential ECM.

Btw I only got ~30 games in this weekend. But putting the results in will take a bit still. I do however have my 20 games in Spiders. Then part of my Cicada and Highlander results.

[Alistair Winter]

RF Greywolf, on 30 September 2013 - 06:51 AM, said:

Hopefully after work today I can get my 20 matches compiled and put into a spreadsheet. Anything super specific anyone wants above the norms, at least that I can tell from an end game screenshot?

Just don't repeat my mistake -

When looking at the probability of winning a match depending on the number of assault / heavy mechs on a team, make sure that you don't count any disconnected heavy / assault mech, as it makes no sense. After all, the point is to see if a team is more likely to win if it has more heavy / assault mechs in the fight.

It might also be interesting to look only at heavy mechs as a key factor, as it has already been pointed out that assault mechs may even have a negative impact on the chances of winning. So many new players in their stock Atlas, doing 50 damage.

Shadey99, on 30 September 2013 - 08:51 AM, said:

While hard to factor in if they actively have ECM or not, my stats do take into account which variant is played. So they will show potential ECM.
Btw I only got ~30 games in this weekend. But putting the results in will take a bit still. I do however have my 20 games in Spiders. Then part of my Cicada and Highlander results.

Awesome. Awesome to the max.

[Urdnot Mau]

I logged in just to like your post

[Agent 0 Fortune]

I collected results from 10 matches in my Cicada.
I was going to post a all screens and match evaluations (and may eventually), but for now I’ll post some of the highlights.

Game Mode: Assault (all matches ended in enemy destruction, none ended in base cap)
Mech: Medium, Cicada CDA-3M (ECM, 2x ERLL, 2xML, 300XL engine) – I am not particularly good with this mech (probably average Medium Elo pilot), but I will be driving one in an future event and figured this would be a good opportunity to get some practice, and document 10 matches.
Maps: Tourmaline desert was by far the most popular 4/10 matches on that map

Weight Distribution:
90% of the time the team with the most Assault/Heavy mechs won (the underdog won once).
0% success rate on Matchmaker Weight Class balance (not one match was evenly balanced by weight class)
40% a single deviation occurred (only a single weight class on each side was shifted)
40% of the time at least 3 weight classes were unbalanced
20% of the time none of the weight classes matched
(of those last 60%, 4 of the matches were unbalanced by 3 or more in a single weight class)
On average (rounding to the nearest whole number) there were 6 Assault mechs per match, 8 Heavies, 6 Medium, and 4 Lights.

Elo Distribution:
Considering that Elo is based on weight-class, and all weight-classes are not created equal (demonstrated by 90% victory rate for teams with the most heavy/assault). This information, when combined with the matchmakers 100% failure rate on weight class balancing, make any attempt to quantify the Elo matchmaker impossible. However there were:
Only 3 matches did not end in a complete route, where the losing team was able to destroy at least 6 enemy mechs.
1 match in which the team with fewest heavy/assault mechs, plus one disconnect, and one missing player annihilated the opposing team.
1 match in which the a team that inflicted 25% more damage (600 points) on the enemy and still lost the game via destruction (separate match from the above).

ECM Distribution:
100% of matches had 3 or more ECM mechs (including myself). There were 3 matches were I was the only mech on my team with ECM. However having more ECM did not guarantee victory. Only 50% of the time did that team with more ECM win (of course you could also say that 100% of ECM teams won their match)

Match Creation:
It is worth noting that the Matchmaker never failed to find match, and all matches were setup in instantly (no wait), which makes me wonder why there was an average of 1 disconnect/AFK player per match, and 2 instances where the Matchmaker could not count to 12, and just set up a 12v11 match. I think with a little more effort better matches could be found with closer weight and Elo boundaries.

[Krivvan]

When I have time, I may be able to do the same thing in my Jenner for Conquest since I'm one of the few that prefers Conquest over Assault.

[RF Greywolf]

Ok here are the ending stats from my case study of 20 matches running a Centurion PUG. I'll try to get my spreadsheet up when I get some more free time.

Winning Team AVG Breakdown

• Assaults-24% (2.85)
  
• Heavies-38% (4.35)
  
• Mediums-20% (2.3)
  
• Lights- 18% (2.15)

AVG Damage-272.61 per active player

AVG Kills-10.2 per active player

AVG Disconnects- 0.1 per match

Losing Team AVG Breakdown

• Assaults-24% (2.75)
  
• Heavies-35% (4.05)
  
• Mediums-30% (3.5)
  
• Lights- 11% (1.35)

AVG Damage-209.77 per active player

AVG Kills-5.65 per active player

AVG Disconnects- 0.15 per match

MOST COMMON MECHS

• Assault----Atlas (2.5 avg per game)
  
• Heavy----Jagermech (2.45 avg per game)
  
• Medium---Centurion (2.3 avg per game)
  
• Light---Jenner (1.8 avg per game)

RAREST MECHS

• Assault----Awesome (0.65 avg per game)
  
• Heavy----Quickdraw (0.4 avg per game)
  
• Medium---Trebuchet (0.3 avg per game)
  
• Light---Commando (0.1 avg per game)

Sorry if I missed anything, I'll try to get more stats up when I can.

Edited by RF Greywolf, 30 September 2013 - 07:28 PM.

[Alistair Winter]

Thanks for the stats, guys. I really hope you find the time to include a spreadsheet, so I can combine all our findings and look for patterns or irregularities. Right now, it's hard to comment on some of the stats.

It would also be very useful to present the information in the same way (e.g. compare the average percentages of different weight classes, instead of saying that the average match had 4 lights)

Some observations

• RF Greywolf's finding seem to support the idea that the number of heavy mechs (not assault mechs) is often a good indication of which team is going to win.
  
• Interesting to see that the Centurion is one of the most common mechs, I will check my own data.
  
• Interesting to see that weight was so much more important in Agent 0 Fortune's matches. There was only a minor correlation between the heavier team and winning team in my 40 matches. May have something to do with ELO, or may be a random difference due to a smaller sample (10 matches)

Agent 0 Fortune, on 30 September 2013 - 05:17 PM, said:

Match Creation:
It is worth noting that the Matchmaker never failed to find match, and all matches were setup in instantly (no wait), which makes me wonder why there was an average of 1 disconnect/AFK player per match, and 2 instances where the Matchmaker could not count to 12, and just set up a 12v11 match. I think with a little more effort better matches could be found with closer weight and Elo boundaries.

You were probably the last one added to the match, after the matchmaker had already been looking for a while. Which means you probably often had a different ELO level than the other players. If you have an average ELO level, that means you'll rarely have trouble finding a match, because you can be included in a match with a surplus of either low or high ELO levels, right?

[Blacksoul1987]

I haven't tracked any stats of my games just kinda took note but it seems like which team has most heavies wins is too vague. from what I've noticed it seems like which team has the most Jagermechs wins. this is kinda reflected by damage scores as well you can see cataphracts and Jagermechs running nearly identical builds and the Jager's version tends to consistently score several hundred points higher than cataphract running the same thing. It is not uncommon at all to see jagers running 700-1300 pts in a game. Jagers tend to score high regardless of their weapon loadouts , though AC5's are most common right now.

[Joseph Mallan]

My AC40 Jager averages 313 damage per game. Then again I am a bit too aggressive for my own good as Anton... and Joe... and Xando...

[Shadey99]

Blacksoul1987, on 01 October 2013 - 05:01 AM, said:

I haven't tracked any stats of my games just kinda took note but it seems like which team has most heavies wins is too vague. from what I've noticed it seems like which team has the most Jagermechs wins. this is kinda reflected by damage scores as well you can see cataphracts and Jagermechs running nearly identical builds and the Jager's version tends to consistently score several hundred points higher than cataphract running the same thing. It is not uncommon at all to see jagers running 700-1300 pts in a game. Jagers tend to score high regardless of their weapon loadouts , though AC5's are most common right now.

As someone who mastered Jagermechs and has nearly 200 games in the Firebrand I bought during the sale a couple weeks ago... Good Jager plots score well against average or bad players. Good players know that to carry a heavy weapons layout we tend to require XL engines and aim there instead of CT. However I've seen games where my team had 6 Jagers (including mine) and I was the only one who did well.

It's kinda funny that all the heavies tend to be support mechs... Orion, Jager, and Cataphracts are direct fire support (Or indirect for some Orions) and the Catapult is a indirect fire support using lrms.

[Blacksoul1987]

Shadey99, on 01 October 2013 - 05:23 AM, said:

As someone who mastered Jagermechs and has nearly 200 games in the Firebrand I bought during the sale a couple weeks ago... Good Jager plots score well against average or bad players. Good players know that to carry a heavy weapons layout we tend to require XL engines and aim there instead of CT. However I've seen games where my team had 6 Jagers (including mine) and I was the only one who did well.

It's kinda funny that all the heavies tend to be support mechs... Orion, Jager, and Cataphracts are direct fire support (Or indirect for some Orions) and the Catapult is a indirect fire support using lrms.

its just a general observation I wouldn't get too offended by it. it just works well in pugs cause players love ridge humping and the jager can do it without getting cored as easily.

[Trauglodyte]

Here is the data that I collected from 20 matches played on Friday night. Now, I mentioned when I played the games because it is my observation that game play for M-Th is slightly different than that of F-Su (more caps happen on the Weekends, imo).

• In the 20 matches played, the weight class distribution was 16% Light, 18% Medium, 35% Heavies, and 30% Assaults.
  
• On average for games that DID NOT END IN CAPS, the winning team had 11.7 kills per game and the losing team had 4.5 kills per game.
  
• For non-CAP games, the winning team averaged 302.4 damage versus 208.0 damage for the losing team.
  
• For all games played, 40% of the time the winning team had more Heavies and Assaults, 45% the winning team had less Heavies and Assaults, and 15% of the time the weight class distribution was even.
  
• On average, the winning team had 7.8 Heavies and Assaults while the losing team had 7.9 Heavies and Assaults.
  
• 25% of all games played ended by a Cap.

I'm going to go back through and do a weight calculation along with mech distribution.

Edited by Trauglodyte, 01 October 2013 - 11:12 AM.