Jump to content


Replay analysis - what does this tell us about the MM?

Lets do science statistics

  • Please log in to reply
215 replies to this topic

Zhul87 #41 Posted 08 July 2018 - 01:10 PM

    Lance-corporal

  • Player
  • 21321 battles
  • 99
  • [TEC] TEC
  • Member since:
    05-31-2013

View PostLordMuffin, on 08 July 2018 - 01:03 PM, said:

If the MM tried to get everyone in between 40-60% WR.

Why do I, with above 60% WR still get on average slightly better teams?

Shouldn't I get worse teams?

 

My mistake. When I was talking about "the system" and the 40-60% range, I didn't exclusively ment the mm, but the whole game with all mechanics and choices included. As in, if you play the game you'll certainly end up between 40 and 60% wr because of how the game works. Of course there are also outliers that are below 40 or above 60 %, but they are negligible.

 


Edited by Zhul87, 08 July 2018 - 01:11 PM.


LordMuffin #42 Posted 08 July 2018 - 01:16 PM

    Field Marshal

  • Player
  • 47395 battles
  • 10,512
  • [-GLO-] -GLO-
  • Member since:
    06-21-2011

View PostZhul87, on 08 July 2018 - 01:10 PM, said:

 

My mistake. When I was talking about "the system" and the 40-60% range, I didn't exclusively ment the mm, but the whole game with all mechanics and choices included. As in, if you play the game you'll certainly end up between 40 and 60% wr because of how the game works. Of course there are also outliers that are below 40 or above 60 %, but they are negligible.

 

How do the mechanics try to keep players within the 40-60% WR range? 

And why?



CoDiGGo #43 Posted 08 July 2018 - 01:41 PM

    Warrant Officer

  • Player
  • 14754 battles
  • 566
  • [MOARR] MOARR
  • Member since:
    05-10-2015

Block Quote

 LORD M

 

Total replays:

            19854

Green team average rating:

            4622.01

Red team average rating:

            4527.21

Percentage difference:

            +2.07%

Stronger than enemy:

            11039 battles

Weaker than enemy:

            8815 battles

Percentage Stronger:

            55.60%

 

Me the "unlucky" bastard

 

 

Soooooooo, as you see MM push me to around 47,6 WR in my last 830 battles.

 

 



ortega456 #44 Posted 08 July 2018 - 02:00 PM

    Second Lieutenant

  • Player
  • 59472 battles
  • 1,000
  • Member since:
    06-08-2012
Thanks for the work and the valuable insights!
I would be interested in more insight of the 357 template sorted by tier.
Is such a thing possible with this tool?

Dorander #45 Posted 08 July 2018 - 02:16 PM

    Captain

  • Player
  • 17854 battles
  • 2,118
  • Member since:
    05-07-2012

View PostZhul87, on 08 July 2018 - 11:53 AM, said:

 

If for LordMuffin the avg. team was 2% stronger for 55% of the time, we have to assume that there are also people out there where the teams are 2% weaker for 55% of the time, because the above can't be true for everyone, right?

Would that qualify to make people feel cheated by the system, if said system unintentionally gives them worse or better odds over a long period of time and large amount of battles, simply due to team composition?

 

Your graph, splitted into two, clearly indicates that the majority of games are won or lost if the difference in team ratings is between +-200 points and that outcome and team rating correlate. So getting a 100 points (dis-)advantage on avg. in 55% of your games seems like a big deal in terms of wr, if the system tries to force all players into 40-60% wr spectrum.

graph_1_defeat.pnggraph_1_win.png

 

 

I'd answer your question with "sort of". For your first sentence, remember that teams aren't consistent but randomly assigned every time, so there's really no method of telling what the balancing factor of that 2% stronger value ended up being. It's not the case that one person or one composition of 15 players got shafted in Muffin's favour. But yes, there would be people on the other side of the spectrum.

 

The problem isn't that these people don't have reason to feel unlucky or frustrated. The problem is that these people react to those feelings by claiming it's somehow about *them*, rather than the system, which as you say has no intentionality. Something that has no intentionality can't cheat people, because cheating a specific person requires that very thing: intending to cheat. Feelings of frustration about being unlucky in RNG is one thing, leaping to conclusions another.

 

If Muffin got a 2% advantage 55% of the time on average, then we can't conclude that the system is rigged if his winrate ends up being over 60% (didn't verify his stats, taking him at face value here). There's no way that deterministic factors (even randomly generated ones) could've caused that, so we have to conclude that the true determining factor in what your winrate actually ends up being is playerskill. Furthermore it doesn't support the contention that some players are favoured and some aren't, or we'd see numbers that'd suggest far more RNG-support for Muffin and other unicums/blunicums. Finally, a third seperate assumption often used to claim manipulation: if the system were rigged to limit players between the 40-60% range, we'd expect people like Muffin to be *disfavoured* by the system, rather than find a slight statistic advantage, because the system should be toning him down to make sure he doesn't rise further.



Baldrickk #46 Posted 08 July 2018 - 02:35 PM

    Field Marshal

  • Player
  • 29304 battles
  • 13,334
  • [-TAH-] -TAH-
  • Member since:
    03-03-2013

View PostZhul87, on 08 July 2018 - 12:53 PM, said:

 

If for LordMuffin the avg. team was 2% stronger for 55% of the time, we have to assume that there are also people out there where the teams are 2% weaker for 55% of the time, because the above can't be true for everyone, right?

Would that qualify to make people feel cheated by the system, if said system unintentionally gives them worse or better odds over a long period of time and large amount of battles, simply due to team composition?

 

Your graph, splitted into two, clearly indicates that the majority of games are won or lost if the difference in team ratings is between +-200 points and that outcome and team rating correlate. So getting a 100 points (dis-)advantage on avg. in 55% of your games seems like a big deal in terms of wr, if the system tries to force all players into 40-60% wr spectrum.

graph_1_defeat.pnggraph_1_win.png

 

 

Remember that bit about my removing LordMuffin from his results?  There is another way to ensure you can bring good players into battles - platooning.  If you were to platoon each battle, then we have another constant.

I know that LordMuffin toons sometimes, I'm one of his sometime toon-mates

@LordMuffin, would you say that 1/20 is a fair approximation of how much you platoon?

 

View PostCoDiGGo, on 08 July 2018 - 01:41 PM, said:

 

Me the "unlucky" bastard

 

 

Soooooooo, as you see MM push me to around 47,6 WR in my last 830 battles.

 

 

So 2.4% of times your team is weaker, with an average team difference (before we incllude you) of 1%.  that's pretty close to expected error for that number of games.

Have you tried it with more battles?

What does your trend graph look like?


 

Zhul87 #47 Posted 08 July 2018 - 03:11 PM

    Lance-corporal

  • Player
  • 21321 battles
  • 99
  • [TEC] TEC
  • Member since:
    05-31-2013

View PostLordMuffin, on 08 July 2018 - 01:16 PM, said:

How do the mechanics try to keep players within the 40-60% WR range? 

And why?

 

How? Ask WG for specific details.

Why? To keep people playing. The lower end of the playerbase wins more and the upper end wins less than they would otherwise.
 

View PostDorander, on 08 July 2018 - 02:16 PM, said:

 

I'd answer your question with "sort of". For your first sentence, remember that teams aren't consistent but randomly assigned every time, so there's really no method of telling what the balancing factor of that 2% stronger value ended up being. It's not the case that one person or one composition of 15 players got shafted in Muffin's favour. But yes, there would be people on the other side of the spectrum.

 

The problem isn't that these people don't have reason to feel unlucky or frustrated. The problem is that these people react to those feelings by claiming it's somehow about *them*, rather than the system, which as you say has no intentionality. Something that has no intentionality can't cheat people, because cheating a specific person requires that very thing: intending to cheat. Feelings of frustration about being unlucky in RNG is one thing, leaping to conclusions another.

 

If Muffin got a 2% advantage 55% of the time on average, then we can't conclude that the system is rigged if his winrate ends up being over 60% (didn't verify his stats, taking him at face value here). There's no way that deterministic factors (even randomly generated ones) could've caused that, so we have to conclude that the true determining factor in what your winrate actually ends up being is playerskill. Furthermore it doesn't support the contention that some players are favoured and some aren't, or we'd see numbers that'd suggest far more RNG-support for Muffin and other unicums/blunicums. Finally, a third seperate assumption often used to claim manipulation: if the system were rigged to limit players between the 40-60% range, we'd expect people like Muffin to be *disfavoured* by the system, rather than find a slight statistic advantage, because the system should be toning him down to make sure he doesn't rise further.

 

View PostBaldrickk, on 08 July 2018 - 02:35 PM, said:

 

Remember that bit about my removing LordMuffin from his results?  There is another way to ensure you can bring good players into battles - platooning.  If you were to platoon each battle, then we have another constant.

I know that LordMuffin toons sometimes, I'm one of his sometime toon-mates

@LordMuffin, would you say that 1/20 is a fair approximation of how much you platoon?

 

Would be interesting to see how much the avg. teamrating would change if you factor Muffin in. 
 

NUKLEAR_SLUG #48 Posted 08 July 2018 - 03:22 PM

    Lieutenant

  • Player
  • 27124 battles
  • 1,808
  • Member since:
    06-13-2015

View PostZhul87, on 08 July 2018 - 03:11 PM, said:

 

How? Ask WG for specific details.

Why? To keep people playing. The lower end of the playerbase wins more and the upper end wins less than they would otherwise.

 

You're misunderstanding. There is no system that's intentionally designed to keep you within a specific WR. It's just that in a totally random system with all else being equal then WR will gravitate towards 50%. Any fluctuation above or below that 50% is a combination of statistical anomaly that would even out over time and personal skill influencing the result one way or the other.

 

The reason the range is roughly over 40% to 60% is because their is an upper limit to how much an individual player can contribute to the team. A good player will win more games, but if his team are potoatoes he'll lose other games no matter what and drag his WR downwards. An utterly useless player that contributes nothing to the game will have a better than expected WR because his team carrries him to wins he otherwise wouldn't get. The equilibrium range just so happens to be the roughly 40-60% range. It's very hard to be so good you can exceed 60% wins. It's also very hard to be so bad you end up sub 40%.



LordMuffin #49 Posted 08 July 2018 - 03:27 PM

    Field Marshal

  • Player
  • 47395 battles
  • 10,512
  • [-GLO-] -GLO-
  • Member since:
    06-21-2011

View PostNUKLEAR_SLUG, on 08 July 2018 - 03:22 PM, said:

 

You're misunderstanding. There is no system that's intentionally designed to keep you within a specific WR. It's just that in a totally random system with all else being equal then WR will gravitate towards 50%. Any fluctuation above or below that 50% is a combination of statistical anomaly that would even out over time and personal skill influencing the result one way or the other.

 

The reason the range is roughly over 40% to 60% is because their is an upper limit to how much an individual player can contribute to the team. A good player will win more games, but if his team are potoatoes he'll lose other games no matter what and drag his WR downwards. An utterly useless player that contributes nothing to the game will have a better than expected WR because his team carrries him to wins he otherwise wouldn't get. The equilibrium range just so happens to be the roughly 40-60% range. It's very hard to be so good you can exceed 60% wins. It's also very hard to be so bad you end up sub 40%.

He claimed it was an intentional system if I got his post right, both of them.

 

I think it is within those boundaries due to the reasons you named, randomness.

 

View PostZhul87, on 08 July 2018 - 03:11 PM, said:

 

How? Ask WG for specific details.

Why? To keep people playing. The lower end of the playerbase wins more and the upper end wins less than they would otherwise.
 

 

 

Would be interesting to see how much the avg. teamrating would change if you factor Muffin in. 
 

 

I don't think there is a need to any kind of system to get this 40-60% WR. Just randomness in MM, dmg, accuracy, pen etc is needed.

I don't think there is a need to artificially alter that randomness if a player wins to much or to little.

 

My teams avg PR would be ~4900. An increase with ~300


 

Edited by Daxeno, 08 July 2018 - 03:55 PM.


Zhul87 #50 Posted 08 July 2018 - 03:45 PM

    Lance-corporal

  • Player
  • 21321 battles
  • 99
  • [TEC] TEC
  • Member since:
    05-31-2013

View PostNUKLEAR_SLUG, on 08 July 2018 - 03:22 PM, said:

 

You're misunderstanding. There is no system that's intentionally designed to keep you within a specific WR. It's just that in a totally random system with all else being equal then WR will gravitate towards 50%. Any fluctuation above or below that 50% is a combination of statistical anomaly that would even out over time and personal skill influencing the result one way or the other.

 

The reason the range is roughly over 40% to 60% is because their is an upper limit to how much an individual player can contribute to the team. A good player will win more games, but if his team are potoatoes he'll lose other games no matter what and drag his WR downwards. An utterly useless player that contributes nothing to the game will have a better than expected WR because his team carrries him to wins he otherwise wouldn't get. The equilibrium range just so happens to be the roughly 40-60% range. It's very hard to be so good you can exceed 60% wins. It's also very hard to be so bad you end up sub 40%.

 

View PostLordMuffin, on 08 July 2018 - 03:27 PM, said:

He claimed it was an intentional system if I got his post right, both of them.

 

I think it is within those boundaries due to the reasons you named, randomness.

 

View PostLordMuffin, on 08 July 2018 - 03:29 PM, said:

 

I don't think there is a need to any kind of system to get this 40-60% WR. Just randomness in MM, dmg, accuracy, pen etc is needed.

I don't think there is a need to artificially alter that randomness if a player wins to much or to little.

 

 

That randomnes is intentionally and by design in the game and because of that we see that 40-60% range, as you both have pointed out. Without that you'd see a different wr distribution over a wider spectrum among players. I've made it clear in the other post that I was talking about the whole game when I mentioned "the system" and not a "special system" within the game.

jabster #51 Posted 08 July 2018 - 03:55 PM

    Field Marshal

  • Beta Tester
  • 12526 battles
  • 22,406
  • [WSAT] WSAT
  • Member since:
    12-30-2010

View PostNUKLEAR_SLUG, on 08 July 2018 - 02:22 PM, said:

 

You're misunderstanding. There is no system that's intentionally designed to keep you within a specific WR. It's just that in a totally random system with all else being equal then WR will gravitate towards 50%. Any fluctuation above or below that 50% is a combination of statistical anomaly that would even out over time and personal skill influencing the result one way or the other.

 

The reason the range is roughly over 40% to 60% is because their is an upper limit to how much an individual player can contribute to the team. A good player will win more games, but if his team are potoatoes he'll lose other games no matter what and drag his WR downwards. An utterly useless player that contributes nothing to the game will have a better than expected WR because his team carrries him to wins he otherwise wouldn't get. The equilibrium range just so happens to be the roughly 40-60% range. It's very hard to be so good you can exceed 60% wins. It's also very hard to be so bad you end up sub 40%.

 

It’s hard to know exactly what WG’s thoughts were when they designed the game but my assumption has always been that it was intentional that the win-rates are ‘pushed’ to a fairly narrow band. That doesn’t mean it’s rigged just that’s what you get with 15 vs. 15 and no respawns plus a game where even good players have to trade shots with bad players.

 

The whole game screams of allow all but the worse to have good games and don’t allow good players to reliable dominate games.



NUKLEAR_SLUG #52 Posted 08 July 2018 - 03:56 PM

    Lieutenant

  • Player
  • 27124 battles
  • 1,808
  • Member since:
    06-13-2015

View PostZhul87, on 08 July 2018 - 03:45 PM, said:

That randomnes is intentionally and by design in the game and because of that we see that 40-60% range, as you both have pointed out. Without that you'd see a different wr distribution over a wider spectrum among players. I've made it clear in the other post that I was talking about the whole game when I mentioned "the system" and not a "special system" within the game.

 

Taking 15 random players and throwing them up against 15 other random players why would you expect to see anything other than what we see currently? You appear to think 40-60%  is 'wrong' somehow so why/how are you expecting to see a wider WR distribution?

Gkirmathal #53 Posted 08 July 2018 - 04:08 PM

    Second Lieutenant

  • Player
  • 8125 battles
  • 1,487
  • Member since:
    01-14-2013

@Baldrick as well.

As average figures 'even out' when the sample sizes increase, that will IMO yield somewhat skewed results. As no human can play such big amount of consecutive games and say "hey my session went great but averagely".

 

So average figures over x thousand games, are IMO indeed a good measure vs the large sample size in the analysed batch. But IMO herein lies sort of the the problem, players experience team hiatuses in the small sample sizes (sub 20 games).
Thus show a graph with n average compiled of 1000*x games evens out these hiatuses players experience (an moan about) in small sample sizes.

 

So I would like, if it is not all but impossible, if someone could make an analyse of 1000 games and show the results on a consecutive 10 (or 20) games session basis. Results, I guess a dual bar chart would be best. One bar whowing the average WN8/PR/what of your own team and the second bar showing the average of the enemy team (over 10 or 20 games).

 

In short: I would like a clearer picture of the deviation (hiatuses) between teams over a small sample size.

 

Would this be possible?

(I can't help regretfully, cause I'm replay less since last year.)


Edited by Gkirmathal, 08 July 2018 - 04:21 PM.


SilentGaze #54 Posted 08 July 2018 - 04:10 PM

    Sergeant

  • Player
  • 657 battles
  • 234
  • Member since:
    03-22-2017

View PostDosjer007, on 08 July 2018 - 09:33 AM, said:

 

 

Do all u guys keep old replays, that many old replays?

 

I guess you have all the client versions of the game installed in your pc as well.

 

View PostCmdRatScabies, on 08 July 2018 - 09:55 AM, said:

 

I have 32k replays saved but I don't have old client versions to view them.

 

You don't have to have all versions in your computer to view old replay files.  You can download them when needed from this side. Seems that there are all versions starting ver. 07.0  2011_12_19

Baldrickk #55 Posted 08 July 2018 - 04:17 PM

    Field Marshal

  • Player
  • 29304 battles
  • 13,334
  • [-TAH-] -TAH-
  • Member since:
    03-03-2013

View PostZhul87, on 08 July 2018 - 03:11 PM, said:

Would be interesting to see how much the avg. teamrating would change if you factor Muffin in. 

I remember for me, it was that my teams were about +2.5% better than the enemy on average.

Muffin would be a little higher. 



SilentGaze #56 Posted 08 July 2018 - 04:27 PM

    Sergeant

  • Player
  • 657 battles
  • 234
  • Member since:
    03-22-2017

@Baldrickk

One thing I have pondered is that there is no means using players ratings from the time game was played.  So analysing LordMuffings replays behind several years back doesn't give picture what the teams ratings were in that time.  

 

To see if there is some kind of inaccuracy, I used last Aprill replay_analyzer to an small group of files which vere less than 2 months old.  Planning to analyze those files after an year or so.



Baldrickk #57 Posted 08 July 2018 - 04:47 PM

    Field Marshal

  • Player
  • 29304 battles
  • 13,334
  • [-TAH-] -TAH-
  • Member since:
    03-03-2013

View PostSilentGaze, on 08 July 2018 - 04:27 PM, said:

@Baldrickk

One thing I have pondered is that there is no means using players ratings from the time game was played.  So analysing LordMuffings replays behind several years back doesn't give picture what the teams ratings were in that time.  

 

To see if there is some kind of inaccuracy, I used last Aprill replay_analyzer to an small group of files which vere less than 2 months old.  Planning to analyze those files after an year or so.

People can get better over time, so yes, the numbers are not going to be perfect.

But lets say there are a number of people in a battle who improve.

What are the chances that they will be on one side or the other?

It's not like the MM can look into the future



azakow #58 Posted 08 July 2018 - 05:03 PM

    Brigadier

  • Player
  • 74789 battles
  • 4,691
  • Member since:
    05-23-2011

Interessting.

I got 53GB of replays, 70k+ matches

Oldest replay was recorded on December, 19th in 2011.

 

Where to download this tool?


Edited by azakow, 08 July 2018 - 05:12 PM.


Dorander #59 Posted 08 July 2018 - 05:58 PM

    Captain

  • Player
  • 17854 battles
  • 2,118
  • Member since:
    05-07-2012

View PostBaldrickk, on 08 July 2018 - 03:47 PM, said:

 

It's not like the MM can look into the future

 

It can if it picked you to be a unicum! :B

 

Sorry, that conspiracy-theory-thinking I did this morning is tenacious. :hiding:



DangerMouse #60 Posted 08 July 2018 - 07:39 PM

    Warrant Officer

  • Beta Tester
  • 24578 battles
  • 584
  • [FLOG] FLOG
  • Member since:
    10-28-2010

Quick question about the below Baldrickk.

 

The mean value for his team's rating is a little higher than that for the enemy, and for about 5% of his battles, his team is stronger. This most likely due to platooning, but needs to be verified.

 

Is this not likely due to the one constant in all 20k replays being a player significantly better than the average player, I would have expected a 60% WR player on average to have a better team due to the presence of that player?

 

Great work BTW.

 

DM






1 user(s) are reading this topic

0 members, 1 guests, 0 anonymous users