Jump to content

A Tale Of 90 Matches (Updated 3. Oct)


136 replies to this topic

#1 Alistair Winter

    Member

  • PipPipPipPipPipPipPipPipPipPipPip
  • Storm
  • Storm
  • 10,823 posts
  • LocationBergen, Norway, FRR

Posted 25 September 2013 - 08:56 AM

I played 20 matches in Assault mode to test my double gauss Cataphracts. Since I was already gathering statistics for myself, I gathered some general statistics about those 20 random matches.

Needless to say, 20 matches isn't a very big number compared to the thousands and thousands of matches played every day. But PGI isn't sharing a lot of statistics with us lately, and it's a lot of work for fans to gather statistics, so just take it for what it is.
  • In these 20 matches, the distribution between weight classes was 22% assault mechs, 44% heavy mechs, 18% medium mechs and 16% light mechs.
  • On average, the winning team made 11,6 kills and the losing team made 4,7 kills.
  • The winning players (not counting disconnects) did 279,5 damage each on average while the losing players did 196,9 damage each on average.
  • In 45% of matches, the winning team had more heavy / assault mechs. In 30% of matches, the winning team had fewer heavy / assault mechs. In 25% of matches, the number of heavy / assault mechs was even.
  • On average, the winning team had 8,2 heavy / assault mechs, the losing team had 7,6 heavy / assault mechs.
  • Only 0,3 players disconnected per match. There was no correlation between disconnects and the chance of losing.
  • 15% of these matches ended by cap.
It should be factored in that I played a heavy mech in all 20 matches, for whatever that's worth.








Spoiler


So yeah. Only 20 matches. Not very meaningful. Interesting to me all the same, so I thought I'd share. Maybe I'll update after 40 matches :)


EDIT:
Update.
I've played 20 matches in my Raven.
  • In these 20 matches, the distribution between weight classes was 28% assault mechs, 37% heavy mechs, 18% medium mechs and 18% light mechs.
  • On average, the winning team made 11,6 kills and the losing team made 5,2 kills.
  • In 40% of matches, the winning team had more heavy / assault mechs. In 30% of matches, the winning team had fewer heavy / assault mechs. In 30% of matches, the number of heavy / assault mechs was even.
  • On average, the winning team had 7,95 heavy / assault mechs, the losing team had 7,45 heavy / assault mechs.
  • 5% of these matches ended by cap.
You can see the combined results from all 40 matches below.






Spoiler


Note: The statistics are currently flawed, because I haven't considered the effect of disconnects when looking at the correlation between heavy / assault mechs and the chance of winnin. Will fix later.


EDIT 2:

Links to other people's statistics here:

Edited by Alistair Winter, 03 October 2013 - 01:45 PM.


#2 Artgathan

    Member

  • PipPipPipPipPipPipPipPip
  • Knight Errant
  • Knight Errant
  • 1,764 posts

Posted 25 September 2013 - 09:14 AM

Good results, +1 for Math and Science at work.

Would it be possible to do this again while running a mech of each weight class yourself? I've noticed (and I am aware that there may be many biases at work here) that when I play in a Light, I tend to get into matches that have more lights.

#3 Joseph Mallan

    ForumWarrior

  • PipPipPipPipPipPipPipPipPipPipPipPipPipPipPip
  • FP Veteran - Beta 1
  • FP Veteran - Beta 1
  • 35,216 posts
  • Google+: Link
  • Facebook: Link
  • LocationMallanhold, Furillo

Posted 25 September 2013 - 09:15 AM

Nice info Alistair. Nice to see the majority of Mechs is Heavies and not Assault. And despite all the whine Only 15% Caps.

#4 Mehlan

    Member

  • PipPipPipPipPipPip
  • 481 posts
  • LocationTx

Posted 25 September 2013 - 09:16 AM

keep running a tally, while '20' matches may not be a lot...it's a number and if you continue tracking it..the #'s add up

#5 Bilbo

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • The Nimble
  • The Nimble
  • 7,864 posts
  • LocationSaline, Michigan

Posted 25 September 2013 - 09:18 AM

View PostJoseph Mallan, on 25 September 2013 - 09:15 AM, said:

....And despite all the whine Only 15% Caps.

We've known this bit for a while. Of course, you have to actually believe the numbers PGI give us when they choose to do so.

#6 Joseph Mallan

    ForumWarrior

  • PipPipPipPipPipPipPipPipPipPipPipPipPipPipPip
  • FP Veteran - Beta 1
  • FP Veteran - Beta 1
  • 35,216 posts
  • Google+: Link
  • Facebook: Link
  • LocationMallanhold, Furillo

Posted 25 September 2013 - 09:19 AM

He has enough to get an SPC chart going.

@ Bilbo, Touche'

Edited by Joseph Mallan, 25 September 2013 - 09:20 AM.


#7 PEEFsmash

    Member

  • PipPipPipPipPipPipPipPip
  • Philanthropist
  • Philanthropist
  • 1,280 posts
  • LocationLos Angeles

Posted 25 September 2013 - 09:23 AM

View PostAlistair Winter, on 25 September 2013 - 08:56 AM, said:

I played 20 matches in Assault mode to test my double gauss Cataphracts. Since I was already gathering statistics for myself, I gathered some general statistics about those 20 random matches.

Needless to say, 20 matches isn't a very big number compared to the thousands and thousands of matches played every day. But PGI isn't sharing a lot of statistics with us lately, and it's a lot of work for fans to gather statistics, so just take it for what it is.
  • In these 20 matches, the distribution between weight classes was 22% assault mechs, 44% heavy mechs, 18% medium mechs and 16% light mechs.
  • On average, the winning team made 11,6 kills and the losing team made 4,7 kills.
  • The winning players (not counting disconnects) did 279,5 damage each on average while the losing players did 196,9 damage each on average.
  • In 45% of matches, the winning team had more heavy / assault mechs. In 30% of matches, the winning team had fewer heavy / assault mechs. In 25% of matches, the number of heavy / assault mechs was even.
  • On average, the winning team had 8,2 heavy / assault mechs, the losing team had 7,6 heavy / assault mechs.
  • Only 0,3 players disconnected per match. There was no correlation between disconnects and the chance of losing.
  • 15% of these matches ended by cap.
It should be factored in that I played a heavy mech in all 20 matches, for whatever that's worth.




Spoiler


So yeah. Only 20 matches. Not very meaningful. Interesting to me all the same, so I thought I'd share. Maybe I'll update after 40 matches :)


AWESOME WORK. This is fantastic stuff. I'd call this the most constructive post in the forums in MONTHS.

I should do the same for 20 of my matches.

So this presents disconfirming evidence of the following forum-frenzies:
1: OMG Light Mech Swarms Everywhere!!!! (Fewer Lights than any other class.)
2: Capwarrior Online!!! (Low cap rate)
3: The more Lights, the more wins!! (Heavier team = more likely to win.)

I think for when you add to this study, or when I attempt a replication, I will switch between classes each match, which would be 5 matches in each class.

I'll also provide video of the games for people looking for details.

Edited by PEEFsmash, 25 September 2013 - 09:25 AM.


#8 Lord Perversor

    Member

  • PipPipPipPipPipPipPipPip
  • Knight Errant
  • Knight Errant
  • 1,815 posts
  • LocationSomewhere in New Aragon

Posted 25 September 2013 - 09:24 AM

It's interesting to note that know how many ppl are in premade can help to tweak a bit the numbers.

I always tought the premades are the main reason for large imbalancing numbers, since the premade force a drop of 4 players together indiferently of their weight ( 4x assault or light lance)

#9 Joseph Mallan

    ForumWarrior

  • PipPipPipPipPipPipPipPipPipPipPipPipPipPipPip
  • FP Veteran - Beta 1
  • FP Veteran - Beta 1
  • 35,216 posts
  • Google+: Link
  • Facebook: Link
  • LocationMallanhold, Furillo

Posted 25 September 2013 - 09:27 AM

View PostPEEFsmash, on 25 September 2013 - 09:23 AM, said:


AWESOME WORK. This is fantastic stuff. I'd call this the most constructive post in the forums in MONTHS.

I should do the same for 20 of my matches.

So this presents disconfirming evidence of the following forum-frenzies:
1: OMG Light Mech Swarms Everywhere!!!! (Fewer Lights than any other class.)
2: Capwarrior Online!!! (Low cap rate)
3: The more Lights, the more wins!! (Heavier team = more likely to win.)

I think for when you add to this study, or when I attempt a replication, I will switch between classes each match, which would be 5 matches in each class.

I'll also provide video of the games for people looking for details.

Maybe you should do your first 20 in a Light though. To better replicate the test conditions Alistair set. Then switch it up in the next set.

Edited by Joseph Mallan, 25 September 2013 - 09:28 AM.


#10 PEEFsmash

    Member

  • PipPipPipPipPipPipPipPip
  • Philanthropist
  • Philanthropist
  • 1,280 posts
  • LocationLos Angeles

Posted 25 September 2013 - 09:37 AM

View PostJoseph Mallan, on 25 September 2013 - 09:27 AM, said:

Maybe you should do your first 20 in a Light though. To better replicate the test conditions Alistair set. Then switch it up in the next set.

Replicating his conditions would be me running all of mine in a heavy. Improving on it would be me running 5 of each.

#11 CrashieJ

    Member

  • PipPipPipPipPipPipPipPip
  • 1,435 posts
  • LocationGalatea (Mercenary's Star)

Posted 25 September 2013 - 09:44 AM

View PostAlistair Winter, on 25 September 2013 - 08:56 AM, said:

I played 20 matches in Assault mode to test my double gauss Cataphracts. Since I was already gathering statistics for myself, I gathered some general statistics about those 20 random matches.

Needless to say, 20 matches isn't a very big number compared to the thousands and thousands of matches played every day. But PGI isn't sharing a lot of statistics with us lately, and it's a lot of work for fans to gather statistics, so just take it for what it is.
  • In these 20 matches, the distribution between weight classes was 22% assault mechs, 44% heavy mechs, 18% medium mechs and 16% light mechs.
  • On average, the winning team made 11,6 kills and the losing team made 4,7 kills.
  • The winning players (not counting disconnects) did 279,5 damage each on average while the losing players did 196,9 damage each on average.
  • In 45% of matches, the winning team had more heavy / assault mechs. In 30% of matches, the winning team had fewer heavy / assault mechs. In 25% of matches, the number of heavy / assault mechs was even.
  • On average, the winning team had 8,2 heavy / assault mechs, the losing team had 7,6 heavy / assault mechs.
  • Only 0,3 players disconnected per match. There was no correlation between disconnects and the chance of losing.
  • 15% of these matches ended by cap.
It should be factored in that I played a heavy mech in all 20 matches, for whatever that's worth.



Spoiler


So yeah. Only 20 matches. Not very meaningful. Interesting to me all the same, so I thought I'd share. Maybe I'll update after 40 matches :)


Actual Math that I can follow (concise and logical), unlike "daemur's"

so this is what I'm going to do, I'm going to "Like" your post, and quote it IN FULL WITH NO EDITING.

#12 Agent 0 Fortune

    Member

  • PipPipPipPipPipPipPipPipPip
  • Bad Company
  • Bad Company
  • 3,403 posts

Posted 25 September 2013 - 09:44 AM

Good work, if we get enough people tracking their 20 match results we could probably get some interesting information (and I am sure varried information). A shame PGI shares so little telemetry trivia they gather. (So far all they have shared is that the catapult is the most owned mech, and the cataphract the most purchased, so the Heavy bias is understandable).

There are a lot of variables that could influence this: time of day, server population, elo, weight class, pug/premade, map, game mode, etc

#13 Joseph Mallan

    ForumWarrior

  • PipPipPipPipPipPipPipPipPipPipPipPipPipPipPip
  • FP Veteran - Beta 1
  • FP Veteran - Beta 1
  • 35,216 posts
  • Google+: Link
  • Facebook: Link
  • LocationMallanhold, Furillo

Posted 25 September 2013 - 09:46 AM

View PostPEEFsmash, on 25 September 2013 - 09:37 AM, said:

Replicating his conditions would be me running all of mine in a heavy. Improving on it would be me running 5 of each.

I was more see it as you running our first 20 in a light. The results could be interesting. As Alistair guaranteed there would always be one heavy, would you bringing one light increase the number of lights per match?

Then running 5 in each an tracking if it changes the numbers any.

Edited by Joseph Mallan, 25 September 2013 - 09:47 AM.


#14 PEEFsmash

    Member

  • PipPipPipPipPipPipPipPip
  • Philanthropist
  • Philanthropist
  • 1,280 posts
  • LocationLos Angeles

Posted 25 September 2013 - 10:08 AM

View PostAgent 0 Fortune, on 25 September 2013 - 09:44 AM, said:

Good work, if we get enough people tracking their 20 match results we could probably get some interesting information (and I am sure varried information). A shame PGI shares so little telemetry trivia they gather. (So far all they have shared is that the catapult is the most owned mech, and the cataphract the most purchased, so the Heavy bias is understandable).

There are a lot of variables that could influence this: time of day, server population, elo, weight class, pug/premade, map, game mode, etc


People should also post video of the matches or at least screenshots of each match so nobody with an agenda fools with the numbers for personal reasons.

#15 Wolfways

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Mercenary
  • 6,499 posts
  • LocationIn a shutdown overheated mech near you.

Posted 25 September 2013 - 10:22 AM

View PostPEEFsmash, on 25 September 2013 - 09:23 AM, said:

So this presents disconfirming evidence of the following forum-frenzies:
1: OMG Light Mech Swarms Everywhere!!!! (Fewer Lights than any other class.)
2: Capwarrior Online!!! (Low cap rate)
3: The more Lights, the more wins!! (Heavier team = more likely to win.)

No it doesn't. The stats are for assault mode only.

#16 Almond Brown

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Philanthropist
  • Philanthropist
  • 5,851 posts

Posted 25 September 2013 - 10:32 AM

Would providing end game screens to Alistair be of any help to increase the sample size? I can't be arsed to chart the results, but a SCRN capture is easy enough and show all the stats required right... :)

#17 Alistair Winter

    Member

  • PipPipPipPipPipPipPipPipPipPipPip
  • Storm
  • Storm
  • 10,823 posts
  • LocationBergen, Norway, FRR

Posted 25 September 2013 - 10:40 AM

View PostPEEFsmash, on 25 September 2013 - 09:23 AM, said:

AWESOME WORK. This is fantastic stuff. I'd call this the most constructive post in the forums in MONTHS.

I don't know about that, but it's quite possible that it's one of the most constructive posts coming from me personally :)

With that said, I do think it's true that people have stopped making threads like this one, lately. The forum used to be so full of optimism, people working hard to present solutions to fixing the game. Some of the spreadsheets people posted on this forum were probably the result of days of work. I don't see that very often any more.

View PostPEEFsmash, on 25 September 2013 - 09:23 AM, said:

So this presents disconfirming evidence of the following forum-frenzies:
1: OMG Light Mech Swarms Everywhere!!!! (Fewer Lights than any other class.)
2: Capwarrior Online!!! (Low cap rate)
3: The more Lights, the more wins!! (Heavier team = more likely to win.)

The problem with scientific proof is that the people who need to see it the most, tend to steer clear of it. I imagine putting the word "statistics" in the thread title already scared away 99% of the people making those 3 claims.

View PostPEEFsmash, on 25 September 2013 - 09:23 AM, said:

I think for when you add to this study, or when I attempt a replication, I will switch between classes each match, which would be 5 matches in each class.
I'll also provide video of the games for people looking for details.

Awesome. I do have the screen shot for the results of these 20 matches, but haven't uploaded them yet.

View PostLord Perversor, on 25 September 2013 - 09:24 AM, said:

It's interesting to note that know how many ppl are in premade can help to tweak a bit the numbers.
I always tought the premades are the main reason for large imbalancing numbers, since the premade force a drop of 4 players together indiferently of their weight ( 4x assault or light lance)

Definitely. Impossible to tell though, which is a shame.

View Postgavilatius, on 25 September 2013 - 09:44 AM, said:

Actual Math that I can follow (concise and logical), unlike "daemur's"
so this is what I'm going to do, I'm going to "Like" your post, and quote it IN FULL WITH NO EDITING.

I lol'd.

View PostAgent 0 Fortune, on 25 September 2013 - 09:44 AM, said:

Good work, if we get enough people tracking their 20 match results we could probably get some interesting information (and I am sure varried information). A shame PGI shares so little telemetry trivia they gather. (So far all they have shared is that the catapult is the most owned mech, and the cataphract the most purchased, so the Heavy bias is understandable).
There are a lot of variables that could influence this: time of day, server population, elo, weight class, pug/premade, map, game mode, etc

So many variables, it's easy to go completely overboard and start looking for correlations between all kinds of crazy factors. I was tempted to look at average damage for different types of mechs, for example. That would have taken so much time.

Actually, while PGI probably has a very effective way of looking at their telemetry, I wouldn't be surprised to learn that there's a lot of important stuff they just don't look at, because they don't have the time. Considering how few people they have working on the game, and how many of those are working on future features rather than the status quo, it would be more surprising if they were on top of everything.

View PostAlmond Brown, on 25 September 2013 - 10:32 AM, said:

Would providing end game screens to Alistair be of any help to increase the sample size? I can't be arsed to chart the results, but a SCRN capture is easy enough and show all the stats required right... ^_^

I can push the 'print screen' button myself. What takes time is moving all the numbers over to a spreadsheet and then actually using the spreadsheet to come up with interesting numbers.

I wasn't expecting anyone to help me, because it's probably half an hour of work. And it's not like you'll be rewarded a Nobel's prize for participating.

Edited by Alistair Winter, 25 September 2013 - 10:44 AM.


#18 Corpsecandle

    Member

  • PipPipPipPipPipPip
  • 396 posts
  • LocationDenver, CO

Posted 25 September 2013 - 10:44 AM

Would you be comfortable posting your win/loss ratio to get an idea of your ELO bracket?

Also where these all pug drops?

#19 PEEFsmash

    Member

  • PipPipPipPipPipPipPipPip
  • Philanthropist
  • Philanthropist
  • 1,280 posts
  • LocationLos Angeles

Posted 25 September 2013 - 10:57 AM

View PostCorpsecandle, on 25 September 2013 - 10:44 AM, said:

Would you be comfortable posting your win/loss ratio to get an idea of your ELO bracket?

Also where these all pug drops?


Win/loss will only give us an idea if it is high (1.6/1 or higher) or low (.8 or less). Plus what we really need is w/l post-Elo implementation.

#20 Alistair Winter

    Member

  • PipPipPipPipPipPipPipPipPipPipPip
  • Storm
  • Storm
  • 10,823 posts
  • LocationBergen, Norway, FRR

Posted 25 September 2013 - 11:05 AM

View PostCorpsecandle, on 25 September 2013 - 10:44 AM, said:

Would you be comfortable posting your win/loss ratio to get an idea of your ELO bracket?
Also where these all pug drops?

My base stats here and my mech stats for all mechs with 50+ matches here.

I'm usually pug'ing, or playing with a friend of mine. All these 20 matches were pug drops.

EDIT: Of all my most played mechs, my YLW has the best KDR and second best W/L ratio. What's up with that?

Edited by Alistair Winter, 25 September 2013 - 11:11 AM.






14 user(s) are reading this topic

0 members, 14 guests, 0 anonymous users