The new MM as of 2/14/17- What we know summarized
Feb 15, 2017 0:15:26 GMT -5
EatStinkyTofu, SGT D00M!, and 24 more like this
Post by buzzard on Feb 15, 2017 0:15:26 GMT -5
OK, since the new MM there have been tweaks, changes, a sever crash and reversion to the old system briefly- etc. There have been a few theories about how this is all working that have focused on initially what was thought to be a black box (performance rating). These theories, which may have each been correct at some point, used a combination of hangar strength of highest bot/weapon, damage relative to others on your team, etc. Right now, there is disparate information scattered throughout all of these MM threads. Some of the info is great and spot on, some is outdated, and some just untrue. I am posting this thread to pull together what we know and wisdom and advice from players that fit the data.
During a conversation with our esteemed board mods, pix said that the performance rating, at the moment is primarily damage dealt (this thread- war-robots-forum.freeforums.net/thread/2262/pixonic-developer-news-2-2017 ) . Several of us have assumed that this means average damage (from last 50 games in our profiles). I wanted to test that. I have been looking at profiles after matches and noticed that everyone was in fact hovering right around my average damage rating. I decided to record the stats and make-up of each team for a few consecutive games and see how this held up.
After quick and dirty calculations it supports the “performance rating” as pix said it was- simple average damage output. Over 3 matches (36 pilots), I did find 2 outliers which I excluded from the calculations below- yes, to a small degree this is cherry picking, but I do not have 20 hours to generate a large data set, and even with doing this, the trend is very clear. In one case, the outlier was on my team and in that match someone bailed before the start timer hit 0. I would bet this was the last minute replacement but I can’t confirm. For each pilot in each match, I recorded their average damage, bot/weapons levels, and win% from their profiles. In one match, I include pics. I calculated average dmg rating and coefficient of variation (CV) from each team and then of the entire 12 bots from the match (with the exceptions I mention above). The CV is an easy way to look at how much, on average, a single player’s dmg rating deviated from the average of the other players in the match and is expressed as a % (std deviation/mean).
Match 1 (Blue wins)
Blue Team
Player 1: avg dmg 298k ; win 54%: Stalk 9/10.5, stalk 9/11, rog 8/10, carn 8/10, rhino 7/10
Player 2: avg dmg 357k; win 42%: Leo 7/7, grif 6/8.5, Grif 8/7.5, Rhino 7/8.75
Player 3: avg dmg 351k; win 54%: Lance 8/8, leo 8/8, grif 6/8.5, grif 8/7.5
Player 4 (me): avg dmg 325k; win 60%: Lance 8/8, gal 9/9, fury 8/8, gar 9/9, gar 8/8
Player 5: avg dmg 317k; win 56%: Grif 6/7.25, grif 7/6, coss 3/3(ecu), grif 9/6
Player 6: avg dmg 336k; win 52%: Carn 6/8, gep 6/10.33, grif 6/9, fuj 6/9, grif 6/9
Red Team
Player 1: avg dmg 335k, win 34%: Grif 10/9.5, rhino 10/9.5, raj 10/9, fuj 10/9.66
Player 2: (OUTLIER SO EXCLUDED) avg dmg 625k, win 68%: gal 9/12, jes 9/12, fury 10/12, lance 11/10.66, gal 9/12
Player3: avg dmg 293; win 56%: Grif 8/5.75, grif 7/5, patt 4/6, gol 4/5
Player 4: avg dmg 371; win 48%: Grif 6/7.5, fury 6/6.66, lanc 3/6.66, grif 6/8
Player 5: avg dmg 319; win 54%: Rhino 5/7, grif 8/7.5, leo 9/7, grif 11/6
Player 6: avg dmg 349; win 54%: grif 6/4, grif 6/6.5, grif 7/7, grif 7/6
Average profile dmg rating of blue team- 330k (CV- 6.7%)
Average profile dmg rating of red team- 333k (CV- 8.9%)
Average of all 12 players- 332k (CV- 7.3%)
Match 2 (Blue wins)
Blue Team
Player 1: avg dmg 309k ; win 56%: Leo 9/5.75, grif 8/6.25, fuj 9/9
Player 2: avg dmg 339k; win 60%: stalk 12/12, patt 12/12, grif 12/12, coss 12/12(ecu), nat 12/12
Player 3: avg dmg 373k; win 56%: Fuj 9/8, carn 9/8.5, gep 2/10, grif 7/6.5, rhino 4/7
Player 4: avg dmg 281; win 54%: Grif 6/9, leo 8/7.25, gal 4/6, grif 6/8
Player 5: (me): avg dmg 320k; win 60%: Lance 8/8, gal 9/9, fury 8/8, gar 9/9, gar 8/8
Player 6: avg dmg 283k; win 60%: Stalk 9/8, patt 8/9, grif 8/9, gal 8/8.66
Red Team
Player 1: avg dmg 321k, win 46%: Gep 8/7.66, grif 8/6, leo 8/5.5, patt 6/6.5
Player 2: avg dmg 274k, win 52%: Grif 7/6, gep 7/7, grif 7/6, gol 5/6
Player3: avg dmg 355k; win 58%: Carn 6/8.5, rog 6/9, gep 7/9, gol 8/9, raj 3/8
Player 4: avg dmg 319k; win 50%: Gal 6/9, nat 6/7, grif 8/8.75
Player 5: avg dmg 275k; win 64%: Stalk 11/9, rog 9/8, raj 9/8, grif 8/9, leo 8/9
Player 6: avg dmg 307k; win 46%: Grif 7/7, fuj 9/10, grif 6/8, leo 6/7.75
Average profile dmg rating of blue team- 317k (CV- 11.0%)
Average profile dmg rating of red team- 309k (CV- 10.9%)
Average of all 12 players- 313k (CV- 10.1%)
Match 3 (Blue wins) - screen shots below
Blue Team
Player 1: (OUTLIER SO EXCLUDED) avg dmg 545k ; win 56%: Rog 10/11, Boa 11/10, carn 10/10.5, grif 10/10.25
Player 2: (me) avg dmg 323k; win 60%: Lance 8/8, gal 9/9, fury 8/8, gar 9/9, gar 8/8
Player 3: avg dmg 381k; win 60%: Gep 8/12, gep 9/12, gep 8/12, gep 8/12, gep 7/12
Player 4: avg dmg 345; win 46%: Leo 7/8.75, gal 8/9, rhino 8/9,gri 7/7.75
Player 5: avg dmg 315k; win 58%: Patt 6/8.75, grif 8/7.5, grif 8/7, leo 8/7.5, rhino 8/8
Player 6: avg dmg 322k; win 64%: Grif 10/7, nat 7/6, lanc 5/5.66, leo 6/5.75, gal 6/
Red Team
Player 1: avg dmg 372k; win 52%: Grif 7/7.5, grif 9/8, patt 9/7.75, rhino 8/8
Player 2: avg dmg 336k; win 32%: Gal 2/8.66, rhino 5/8, grif 6/9, leo 6/9, vit 6/8.33
Player3: avg dmg 350k; win 46%: Gal 9/10.33, lanc 9/8.66, rog 9/9, fury 8/8.66
Player 4: avg dmg 351k; win 42%: Stalk 12/12, grif 12/8.75, fury 6/9, rhino 12/11.25, fuj 7/9
Player 5: avg dmg 311k; win 48%: Fuj 11/6.33, carn 1/5, patt 4/5, grif 6/6
Player 6: avg dmg 308k; win 42%: Rhino 7/7.75, gol 8/7, grif 8/8, stalk 11/8.5
Average profile dmg rating of blue team- 337k (CV- 7.99%)
Average profile dmg rating of red team- 338k (CV- 7.37%)
Average of all 12 players- 337k (CV- 7.26%)
BLUE TEAM
RED TEAM
What does it mean from a match making standpoint?
The numbers support the notion that damage rating is, in fact, the primary matching constant. 94% of pilots (34/36) in these 3 matches were within 11% of the average player dmg rating (and usually less than 8% variance). The average team damage was amazingly close in all 3 of the matches (blue/red- 330/333; 37/309k; 337/338k). If this is what pix intended (as they said), this match maker is doing it well. Does hangar strength make a difference- it sure doesn’t look like it. There were 12/12 bots in these matches, 4/5’s and everything in between. Win % doesn’t seem to matter either. Bottom line- it is your average damage over the last 50 rounds that will determine who you are matched against. There may be other factors that are layered on top of this to refine things more, but if so, they are subtle and probably depend on the number of people matching when you start a game. MAJOR CAVEAT- the individual pilot dmg rating does not work when squads are involved (obviously)
What should I expect to see from other pilots and hangars in a given match?
You can expect pilots of about the same average damage (with a few outliers), but from a hangar standpoint almost anything. You can have matches where your hangar looks way OP, and matches where you see hangars with some or even most of the bots stronger than you, because it isn’t the hangar directly, it is the damage that hangar does which is affected by:
1) Play style- Is the pilot after max damage every game? Is the pilot a beacon runner? Is the pilot a sniper/camper? Does the pilot use a mixed or balanced play style?
2) Skill level- Some say it is all about equipment but we know this isn’t the case. Some pilots play bots to their strengths and use good tactics, team work, and battlefield awareness. Others…nope.
3) Hangar make up/equipment- the mix of bots and how they are set-up (think frankenbots here) will obviously make a difference, but this is mitigated already by the fact they have the same average damage output as you
4) Tankers- depending on whether you hit them on the way down or the way up, they skew things
Because you will see a mix of bot strengths there WILL be matches that your poor 9/8 griff gets annihilated in 3 seconds by the lvl 12/12 standing in front of you. So before you freak out and curse the MM, remember that looking at a hangar, individual bot, or occasional event tells you little except a guess as to someone’s playstyle. A person can have 300k average damage with a 3 slot hangar of knife fighters, a hangar of 5 light bots, etc. and any spread of levels.
What should I be expecting as far as wins and how does my single game performance affect me?
The days of 90% win rates and finishing top 3 every match are over clubbers! Using damage as the matching constant really should even things out. All things equal, you will be on the winning team 50% of the time and in the top three of your team damage wise 50%, so in the top three AND on the winning team 25% of the time. If you are fixated on damage output, you will likely be in the top 3 more often, but win less often. If you are a beacon capping fiend, strategic distracter or dedicated beacon defender, you will probably win a little more, but maybe do less damage along the way (although the MoC may skew gold towards the cappers). All in all, I think this MM system really favors a balanced approach where you focus on winning first.
As far as how you do in a single game, form a MM stand point, it doesn’t matter if you finish 1st or 3rd or 6th. It doesn’t matter if you are on the winning team or the losing team. All that matters is if your damage output from the single match moves your running 50 game average up or down. (When the league system is implemented the can possibly be hard boundaries, but we won’t know until it is in).
What should I do about my hangar and how do adjustments affect me?
People have been saying keep you bots and weapon levels similar with the new MM. This is good advice. In fact, if avg damage is going to remain the matching constant, it will pay to level your bots higher since this won’t have a huge impact on damage output but will improve speed, hp, and survivability. (Disclaimer- maybe wait for the league system just in case!). As far as bot balance within your hangar, keeping your bot strength balanced should give you the most consistent performance across bots within a match.
Unfortunately for those of you who like to change your hangars up a lot, sudden, large changes will have a big short-term effect. As we all know, you cant run your TT and then field a hangar of lower level light bots to play around. Your matching constant is your average damage from 50 matches, so it will take time for your matches to balance out to the hangar you are running or more accurately, the damage output of the new hangar. The MM will adjust to it eventually, which is why tanking works right now.
Exploiting the MM
Tanking- it has been talked about to death. It obviously works. Pix has said they would put in some type of penalty for serial leavers. If they do this and players need to stay in the match to tank, it is going to be time consuming to drop your damage output. Don’t forget, if you keep your hangar consistent and play smart, you should win gold ~25% of your matches anyways.
The bottom line
Set-up your hangar to suit your play style and make adjustments slowly. It is fine to have whatever mix of heavy, medium, and light bots you want. Your average damage will output settle down and the MM will team you up with and against pilots that on the average will result in good, close matches. Hopefully the promised new reward system will keep people satisfied with their Au income. If you are all about gold, you need to reset your expectations or open your wallet.
During a conversation with our esteemed board mods, pix said that the performance rating, at the moment is primarily damage dealt (this thread- war-robots-forum.freeforums.net/thread/2262/pixonic-developer-news-2-2017 ) . Several of us have assumed that this means average damage (from last 50 games in our profiles). I wanted to test that. I have been looking at profiles after matches and noticed that everyone was in fact hovering right around my average damage rating. I decided to record the stats and make-up of each team for a few consecutive games and see how this held up.
After quick and dirty calculations it supports the “performance rating” as pix said it was- simple average damage output. Over 3 matches (36 pilots), I did find 2 outliers which I excluded from the calculations below- yes, to a small degree this is cherry picking, but I do not have 20 hours to generate a large data set, and even with doing this, the trend is very clear. In one case, the outlier was on my team and in that match someone bailed before the start timer hit 0. I would bet this was the last minute replacement but I can’t confirm. For each pilot in each match, I recorded their average damage, bot/weapons levels, and win% from their profiles. In one match, I include pics. I calculated average dmg rating and coefficient of variation (CV) from each team and then of the entire 12 bots from the match (with the exceptions I mention above). The CV is an easy way to look at how much, on average, a single player’s dmg rating deviated from the average of the other players in the match and is expressed as a % (std deviation/mean).
Match 1 (Blue wins)
Blue Team
Player 1: avg dmg 298k ; win 54%: Stalk 9/10.5, stalk 9/11, rog 8/10, carn 8/10, rhino 7/10
Player 2: avg dmg 357k; win 42%: Leo 7/7, grif 6/8.5, Grif 8/7.5, Rhino 7/8.75
Player 3: avg dmg 351k; win 54%: Lance 8/8, leo 8/8, grif 6/8.5, grif 8/7.5
Player 4 (me): avg dmg 325k; win 60%: Lance 8/8, gal 9/9, fury 8/8, gar 9/9, gar 8/8
Player 5: avg dmg 317k; win 56%: Grif 6/7.25, grif 7/6, coss 3/3(ecu), grif 9/6
Player 6: avg dmg 336k; win 52%: Carn 6/8, gep 6/10.33, grif 6/9, fuj 6/9, grif 6/9
Red Team
Player 1: avg dmg 335k, win 34%: Grif 10/9.5, rhino 10/9.5, raj 10/9, fuj 10/9.66
Player 2: (OUTLIER SO EXCLUDED) avg dmg 625k, win 68%: gal 9/12, jes 9/12, fury 10/12, lance 11/10.66, gal 9/12
Player3: avg dmg 293; win 56%: Grif 8/5.75, grif 7/5, patt 4/6, gol 4/5
Player 4: avg dmg 371; win 48%: Grif 6/7.5, fury 6/6.66, lanc 3/6.66, grif 6/8
Player 5: avg dmg 319; win 54%: Rhino 5/7, grif 8/7.5, leo 9/7, grif 11/6
Player 6: avg dmg 349; win 54%: grif 6/4, grif 6/6.5, grif 7/7, grif 7/6
Average profile dmg rating of blue team- 330k (CV- 6.7%)
Average profile dmg rating of red team- 333k (CV- 8.9%)
Average of all 12 players- 332k (CV- 7.3%)
Match 2 (Blue wins)
Blue Team
Player 1: avg dmg 309k ; win 56%: Leo 9/5.75, grif 8/6.25, fuj 9/9
Player 2: avg dmg 339k; win 60%: stalk 12/12, patt 12/12, grif 12/12, coss 12/12(ecu), nat 12/12
Player 3: avg dmg 373k; win 56%: Fuj 9/8, carn 9/8.5, gep 2/10, grif 7/6.5, rhino 4/7
Player 4: avg dmg 281; win 54%: Grif 6/9, leo 8/7.25, gal 4/6, grif 6/8
Player 5: (me): avg dmg 320k; win 60%: Lance 8/8, gal 9/9, fury 8/8, gar 9/9, gar 8/8
Player 6: avg dmg 283k; win 60%: Stalk 9/8, patt 8/9, grif 8/9, gal 8/8.66
Red Team
Player 1: avg dmg 321k, win 46%: Gep 8/7.66, grif 8/6, leo 8/5.5, patt 6/6.5
Player 2: avg dmg 274k, win 52%: Grif 7/6, gep 7/7, grif 7/6, gol 5/6
Player3: avg dmg 355k; win 58%: Carn 6/8.5, rog 6/9, gep 7/9, gol 8/9, raj 3/8
Player 4: avg dmg 319k; win 50%: Gal 6/9, nat 6/7, grif 8/8.75
Player 5: avg dmg 275k; win 64%: Stalk 11/9, rog 9/8, raj 9/8, grif 8/9, leo 8/9
Player 6: avg dmg 307k; win 46%: Grif 7/7, fuj 9/10, grif 6/8, leo 6/7.75
Average profile dmg rating of blue team- 317k (CV- 11.0%)
Average profile dmg rating of red team- 309k (CV- 10.9%)
Average of all 12 players- 313k (CV- 10.1%)
Match 3 (Blue wins) - screen shots below
Blue Team
Player 1: (OUTLIER SO EXCLUDED) avg dmg 545k ; win 56%: Rog 10/11, Boa 11/10, carn 10/10.5, grif 10/10.25
Player 2: (me) avg dmg 323k; win 60%: Lance 8/8, gal 9/9, fury 8/8, gar 9/9, gar 8/8
Player 3: avg dmg 381k; win 60%: Gep 8/12, gep 9/12, gep 8/12, gep 8/12, gep 7/12
Player 4: avg dmg 345; win 46%: Leo 7/8.75, gal 8/9, rhino 8/9,gri 7/7.75
Player 5: avg dmg 315k; win 58%: Patt 6/8.75, grif 8/7.5, grif 8/7, leo 8/7.5, rhino 8/8
Player 6: avg dmg 322k; win 64%: Grif 10/7, nat 7/6, lanc 5/5.66, leo 6/5.75, gal 6/
Red Team
Player 1: avg dmg 372k; win 52%: Grif 7/7.5, grif 9/8, patt 9/7.75, rhino 8/8
Player 2: avg dmg 336k; win 32%: Gal 2/8.66, rhino 5/8, grif 6/9, leo 6/9, vit 6/8.33
Player3: avg dmg 350k; win 46%: Gal 9/10.33, lanc 9/8.66, rog 9/9, fury 8/8.66
Player 4: avg dmg 351k; win 42%: Stalk 12/12, grif 12/8.75, fury 6/9, rhino 12/11.25, fuj 7/9
Player 5: avg dmg 311k; win 48%: Fuj 11/6.33, carn 1/5, patt 4/5, grif 6/6
Player 6: avg dmg 308k; win 42%: Rhino 7/7.75, gol 8/7, grif 8/8, stalk 11/8.5
Average profile dmg rating of blue team- 337k (CV- 7.99%)
Average profile dmg rating of red team- 338k (CV- 7.37%)
Average of all 12 players- 337k (CV- 7.26%)
BLUE TEAM
RED TEAM
What does it mean from a match making standpoint?
The numbers support the notion that damage rating is, in fact, the primary matching constant. 94% of pilots (34/36) in these 3 matches were within 11% of the average player dmg rating (and usually less than 8% variance). The average team damage was amazingly close in all 3 of the matches (blue/red- 330/333; 37/309k; 337/338k). If this is what pix intended (as they said), this match maker is doing it well. Does hangar strength make a difference- it sure doesn’t look like it. There were 12/12 bots in these matches, 4/5’s and everything in between. Win % doesn’t seem to matter either. Bottom line- it is your average damage over the last 50 rounds that will determine who you are matched against. There may be other factors that are layered on top of this to refine things more, but if so, they are subtle and probably depend on the number of people matching when you start a game. MAJOR CAVEAT- the individual pilot dmg rating does not work when squads are involved (obviously)
What should I expect to see from other pilots and hangars in a given match?
You can expect pilots of about the same average damage (with a few outliers), but from a hangar standpoint almost anything. You can have matches where your hangar looks way OP, and matches where you see hangars with some or even most of the bots stronger than you, because it isn’t the hangar directly, it is the damage that hangar does which is affected by:
1) Play style- Is the pilot after max damage every game? Is the pilot a beacon runner? Is the pilot a sniper/camper? Does the pilot use a mixed or balanced play style?
2) Skill level- Some say it is all about equipment but we know this isn’t the case. Some pilots play bots to their strengths and use good tactics, team work, and battlefield awareness. Others…nope.
3) Hangar make up/equipment- the mix of bots and how they are set-up (think frankenbots here) will obviously make a difference, but this is mitigated already by the fact they have the same average damage output as you
4) Tankers- depending on whether you hit them on the way down or the way up, they skew things
Because you will see a mix of bot strengths there WILL be matches that your poor 9/8 griff gets annihilated in 3 seconds by the lvl 12/12 standing in front of you. So before you freak out and curse the MM, remember that looking at a hangar, individual bot, or occasional event tells you little except a guess as to someone’s playstyle. A person can have 300k average damage with a 3 slot hangar of knife fighters, a hangar of 5 light bots, etc. and any spread of levels.
What should I be expecting as far as wins and how does my single game performance affect me?
The days of 90% win rates and finishing top 3 every match are over clubbers! Using damage as the matching constant really should even things out. All things equal, you will be on the winning team 50% of the time and in the top three of your team damage wise 50%, so in the top three AND on the winning team 25% of the time. If you are fixated on damage output, you will likely be in the top 3 more often, but win less often. If you are a beacon capping fiend, strategic distracter or dedicated beacon defender, you will probably win a little more, but maybe do less damage along the way (although the MoC may skew gold towards the cappers). All in all, I think this MM system really favors a balanced approach where you focus on winning first.
As far as how you do in a single game, form a MM stand point, it doesn’t matter if you finish 1st or 3rd or 6th. It doesn’t matter if you are on the winning team or the losing team. All that matters is if your damage output from the single match moves your running 50 game average up or down. (When the league system is implemented the can possibly be hard boundaries, but we won’t know until it is in).
What should I do about my hangar and how do adjustments affect me?
People have been saying keep you bots and weapon levels similar with the new MM. This is good advice. In fact, if avg damage is going to remain the matching constant, it will pay to level your bots higher since this won’t have a huge impact on damage output but will improve speed, hp, and survivability. (Disclaimer- maybe wait for the league system just in case!). As far as bot balance within your hangar, keeping your bot strength balanced should give you the most consistent performance across bots within a match.
Unfortunately for those of you who like to change your hangars up a lot, sudden, large changes will have a big short-term effect. As we all know, you cant run your TT and then field a hangar of lower level light bots to play around. Your matching constant is your average damage from 50 matches, so it will take time for your matches to balance out to the hangar you are running or more accurately, the damage output of the new hangar. The MM will adjust to it eventually, which is why tanking works right now.
Exploiting the MM
Tanking- it has been talked about to death. It obviously works. Pix has said they would put in some type of penalty for serial leavers. If they do this and players need to stay in the match to tank, it is going to be time consuming to drop your damage output. Don’t forget, if you keep your hangar consistent and play smart, you should win gold ~25% of your matches anyways.
The bottom line
Set-up your hangar to suit your play style and make adjustments slowly. It is fine to have whatever mix of heavy, medium, and light bots you want. Your average damage will output settle down and the MM will team you up with and against pilots that on the average will result in good, close matches. Hopefully the promised new reward system will keep people satisfied with their Au income. If you are all about gold, you need to reset your expectations or open your wallet.