Jump to content

Proof That Elo/mm Is Working?


66 replies to this topic

#1 Duke Nedo

    Member

  • PipPipPipPipPipPipPipPipPip
  • CS 2023 Top 12 Qualifier
  • CS 2023 Top 12 Qualifier
  • 2,184 posts

Posted 28 January 2015 - 02:11 PM

Ok sorry, just needed a catchy title. :ph34r:

Just a friendly suggestion for everyone complaining about Elo and matchmaking in the solo queue. After having a few win-streaks and loss-streaks too many I decided to track all end results and plot them for fun. Plot below is cumulative wins and losses, -1 for a loss, +1 for a win.

Posted Image

It was a bit less cyclic and periodic so far than it subjectively felt like so I would recommend doing this to get a more objective picture of how it really looks. There was a spot where I felt pulled down towards 0 by the MM in this sample, but over all the trend actually looks much better than perceived and fits well with my global 1.133 w/l ratio after reset and 1.10 for archive.

Sample still small, growing slowly every day... Too early to say if anything about anything, except that the perceived pain is a bit worse than it actually is.

Edit: Added a graph of distribution of end results, plotted the opponents score on wins and own teams score on losses. Most common win result: 12-4, most common loss result: 5-12. Very few 12-0, 0-12 actually!

Posted Image

Edit 2: Thought I'd update the graph with another 250 matches. A bit weird, suddenly I was deadlocked at 1.0 w/l ratio. Would be interesting to know if anything was changed in the code in the end of January, the perceived effect is rather clear. I don't believe that this has anything to do with my Elo, I am seeing the same guys and before I started recordning this graph there is a baseline of some 4000 games or so, so I should already be at my Elo whatever that is.

I'll just leave it here:
Posted Image

Edit 3: I plotted the graphs by Elo-class. Nothing too significant there, all classes end up on the positive side, but it seems like it may be Lights that was mainly responsible for the steep ascent in the original plot.

Posted Image

Edit 4: Will paste the final graph here, about to close this experiment now, losing too many cbills when I have to watch every game til the bitter end to record the end result...

Posted Image

Edited by Duke Nedo, 13 March 2015 - 08:26 AM.


#2 LordKnightFandragon

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • 7,239 posts

Posted 28 January 2015 - 02:13 PM

Luckyyyyyy, you get on the side that wins more often then not, whats the secret?

#3 cSand

    Member

  • PipPipPipPipPipPipPipPipPip
  • 4,589 posts
  • LocationCanada, eh

Posted 28 January 2015 - 02:17 PM

I seen him play, there's no secret

just skillz yo

#4 DaZur

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Elite Founder
  • Elite Founder
  • 7,511 posts
  • LocationWisconsin

Posted 28 January 2015 - 02:21 PM

It's always easier to remember a negative touch than a positive one (Unless it was atypical as one might suspect)...

I've eluded to it a million times... The continue propagation of the MM-Boogieman as a default knee-jerk explanation for a series of poor performances is far more damning to the community and the actual fallout from these matches.

I appreciate you taking a very unfettered and logical assessment as well as the honest admission that you actually perceived negative draws where there really was not one..

Kudos. ;)

View PostLordKnightFandragon, on 28 January 2015 - 02:13 PM, said:

Luckyyyyyy, you get on the side that wins more often then not, whats the secret?

Some might preclude the secret is the reflection he sees in his screen and that persons contribution to the cause... :ph34r:

;)

#5 DaZur

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Elite Founder
  • Elite Founder
  • 7,511 posts
  • LocationWisconsin

Posted 28 January 2015 - 02:25 PM

As I side note on your graph...

If I were a wagering man I'd lay money that the dwells between 85 and 109 as well as 190 were Elo tier jumps. ^_^

#6 Davers

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Legendary Founder
  • Legendary Founder
  • 9,886 posts
  • Facebook: Link
  • LocationCanada

Posted 28 January 2015 - 03:03 PM

View PostDaZur, on 28 January 2015 - 02:21 PM, said:

It's always easier to remember a negative touch than a positive one



#7 Duke Nedo

    Member

  • PipPipPipPipPipPipPipPipPip
  • CS 2023 Top 12 Qualifier
  • CS 2023 Top 12 Qualifier
  • 2,184 posts

Posted 28 January 2015 - 03:07 PM

View PostDaZur, on 28 January 2015 - 02:25 PM, said:

As I side note on your graph...

If I were a wagering man I'd lay money that the dwells between 85 and 109 as well as 190 were Elo tier jumps. ^_^

I don't know if there is such a thing as Elo tiers (isn't each matched seeded by the Elo of a randomly picked player?), but I did possibly see the effect of a dropped Elo after the nosedive I did from match 150->190. I saw real trial mechs for a short time... and I swear a joystick or two... before returning to familiar names.

#8 Kroah

    Rookie

  • 7 posts

Posted 28 January 2015 - 04:20 PM

Thank you to share your thoughts.

Unfortunately, beside showing some nice wins or losses streaks, this does not show the second real frustration of this f*cked up MM: the mighty roflstomp.

We all know the MM is pretty balanced in terms of Win/Loss. BUT players level are so unbalanced between the teams, it finishes into a wipe. And NO, players base is way enough to have balanced team, this is the algorithm used in the MM which is way too permissive and it uses bad criteria to determine the player level.

Edited by Kroah, 28 January 2015 - 04:21 PM.


#9 LordKnightFandragon

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • 7,239 posts

Posted 28 January 2015 - 04:24 PM

View PostDaZur, on 28 January 2015 - 02:21 PM, said:

Some might preclude the secret is the reflection he sees in his screen and that persons contribution to the cause... :ph34r:
;)


But I contribute every match, ofc, these last like 4 matches have gone less then stellar, 0 dmg, 200 dmg, 200 dmg, but I did get 500 some in one match...we almost won but a clueless timberwolf decided to go for a swim in the Mordor lava pit and stare in zoom mode as a BJ blew him up....I would have gotten atleat 2 kills in that game, but the hit detect decided to have half of my 67 point alpha dmg vanish into thin air.....

#10 Colonel Fubar

    Member

  • PipPipPipPipPipPipPip
  • Survivor
  • 650 posts
  • LocationPlanet Agoge in the Mitera System

Posted 28 January 2015 - 04:30 PM

You truly are Duke Nedo!

#11 Alistair Winter

    Member

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

Posted 28 January 2015 - 04:46 PM

lol, look at this guy bringing statistics to a thread in the general forum.

You didn't even post a single post-match result screenshot where you got top damage or provide any kind of anecdotal evidence.

Posted Image

PS: I made a similar thread in 2013, except with fewer matches. Started collecting data to prove I was consistently losing despite doing top damage quite often. When I looked at the data before starting a thread about it, it turns out I was consistently winning more than losing. I just remembered the losses a lot better than the wins. Stupid bias. I was working on a great rant too.

#12 DaZur

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Elite Founder
  • Elite Founder
  • 7,511 posts
  • LocationWisconsin

Posted 28 January 2015 - 10:17 PM

View PostLordKnightFandragon, on 28 January 2015 - 04:24 PM, said:


But I contribute every match, ofc, these last like 4 matches have gone less then stellar, 0 dmg, 200 dmg, 200 dmg, but I did get 500 some in one match...we almost won but a clueless timberwolf decided to go for a swim in the Mordor lava pit and stare in zoom mode as a BJ blew him up....I would have gotten atleat 2 kills in that game, but the hit detect decided to have half of my 67 point alpha dmg vanish into thin air.....

You can only control what you can control. ;)

All the MM does is pair effectively what it considers a reasonably equitable composite teams... After that's it's all skill, chance and chaos theory...

IMHO I think your focusing too much on the "win" component and not enjoying a well played match, win or lose... To be honest for me a productive loss pays roughly the same as a a marginal win so in reality my focus is on having fun.

It's kind'a like looking for something and only finding it when your looking for something else... You seem to have a bit of tunnel vision on winning and missing out on having "fun". ;)

Edited by DaZur, 28 January 2015 - 10:20 PM.


#13 Duke Nedo

    Member

  • PipPipPipPipPipPipPipPipPip
  • CS 2023 Top 12 Qualifier
  • CS 2023 Top 12 Qualifier
  • 2,184 posts

Posted 28 January 2015 - 11:29 PM

View PostKroah, on 28 January 2015 - 04:20 PM, said:

Thank you to share your thoughts.

Unfortunately, beside showing some nice wins or losses streaks, this does not show the second real frustration of this f*cked up MM: the mighty roflstomp.

We all know the MM is pretty balanced in terms of Win/Loss. BUT players level are so unbalanced between the teams, it finishes into a wipe. And NO, players base is way enough to have balanced team, this is the algorithm used in the MM which is way too permissive and it uses bad criteria to determine the player level.


I have recorded all scores as well, and my damage and kills so I can make a distribution plot when I get home from work. I have not experienced these 300 matches as extremely stompy so I think the distribution will lie within what is to be expected from the snowball effect... that's unavoidable and we'll have to live with that. As long as there are no stacked decks or unnatural pulls och pushes... I am guessing the most common score is something like 12-4 or 4-12.

#14 AssaultPig

    Member

  • PipPipPipPipPipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 907 posts

Posted 29 January 2015 - 02:23 AM

I would be extremely surprised if you can maintain a winrate higher than 55% or so in the solo public queue

am I to understand from your graph that 300 games in you are something like 180 and 120?

#15 Telmasa

    Member

  • PipPipPipPipPipPipPipPip
  • Bad Company
  • Bad Company
  • 1,548 posts

Posted 29 January 2015 - 05:28 AM

But, what if I want to tryhard and succeed ALL the time?

(the sad thing is, i'm only half-kidding here)

#16 Duke Nedo

    Member

  • PipPipPipPipPipPipPipPipPip
  • CS 2023 Top 12 Qualifier
  • CS 2023 Top 12 Qualifier
  • 2,184 posts

Posted 29 January 2015 - 05:31 AM

View PostAssaultPig, on 29 January 2015 - 02:23 AM, said:

I would be extremely surprised if you can maintain a winrate higher than 55% or so in the solo public queue

am I to understand from your graph that 300 games in you are something like 180 and 120?

I went +30 in 300 games so it would rather be 165/135 = 1.22. Hmm, that was a bit high, more than the 1.133 I have since stat reset so I may be in for a rough ride soon... :)

#17 Joseph Mallan

    ForumWarrior

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

Posted 29 January 2015 - 05:33 AM

View PostTelmasa, on 29 January 2015 - 05:28 AM, said:

But, what if I want to tryhard and succeed ALL the time?

(the sad thing is, i'm only half-kidding here)

Even if you can walk on water you will likely get crucified occasionally... or maybe just once. :unsure:

#18 Duke Nedo

    Member

  • PipPipPipPipPipPipPipPipPip
  • CS 2023 Top 12 Qualifier
  • CS 2023 Top 12 Qualifier
  • 2,184 posts

Posted 29 January 2015 - 12:18 PM

View PostKroah, on 28 January 2015 - 04:20 PM, said:

Thank you to share your thoughts.

Unfortunately, beside showing some nice wins or losses streaks, this does not show the second real frustration of this f*cked up MM: the mighty roflstomp.

We all know the MM is pretty balanced in terms of Win/Loss. BUT players level are so unbalanced between the teams, it finishes into a wipe. And NO, players base is way enough to have balanced team, this is the algorithm used in the MM which is way too permissive and it uses bad criteria to determine the player level.


Updated the OP with score distibutions, not stompy at all actually! :)

#19 FrDrake

    Member

  • PipPipPipPipPipPipPipPip
  • Legendary Founder
  • Legendary Founder
  • 1,086 posts
  • LocationTexas

Posted 29 January 2015 - 12:29 PM

I respect a man who uses numbers.

One more followup question, are you "tryharding" in these matches, using optimal mechs/loadouts or is this data from you leveling up new mechs?

#20 DEMAX51

    Member

  • PipPipPipPipPipPipPipPipPip
  • Elite Founder
  • 2,269 posts
  • LocationThe cockpit of my Jenner

Posted 29 January 2015 - 12:29 PM

View PostKroah, on 28 January 2015 - 04:20 PM, said:

Thank you to share your thoughts.

Unfortunately, beside showing some nice wins or losses streaks, this does not show the second real frustration of this f*cked up MM: the mighty roflstomp.

We all know the MM is pretty balanced in terms of Win/Loss. BUT players level are so unbalanced between the teams, it finishes into a wipe. And NO, players base is way enough to have balanced team, this is the algorithm used in the MM which is way too permissive and it uses bad criteria to determine the player level.


People need to stop making this argument. 12-0 / 0-12 matches are not at all indicative of poor matchmaking, it's just the nature of the game sometimes.

Go and watch some videos of relatively evenly matched comp teams playing against each other. Lords (or EMP now, I guess) & SJR matches can just as easily result in stomps because of the "snowball effect."

One or two people on your team make one bad mistake, they get taken out, and now you're outnumbered 12-10 making it harder for your team to dig out of the hole and easier for the enemies to focus you down, which leads to more losses on your side, and then even more... It's not always bad matchmaking.

Edited by DEMAX51, 29 January 2015 - 12:32 PM.






2 user(s) are reading this topic

0 members, 2 guests, 0 anonymous users