357 replies to this topic

[Cluster Fox]

C64 Warrior, on 22 June 2020 - 03:40 PM, said:

Now what if you were told player A has a .87 w/l ratio and player B has a 1.31 w/l ratio would you then say player B is more skilled than player A?

I think if you put 12x player A on one side and 12x Player B on the other, Player B team is likely to win and it's likely to be a stomp. But MS numbers will likely be low for them.

If you 1v1 them, that's Solaris.

Edited by Cluster Fox, 22 June 2020 - 03:44 PM.

[Nightbird]

C64 Warrior, on 22 June 2020 - 03:40 PM, said:

If both players average MS is 300 then yes they are equally skilled...if one has a higher w/l than the other then you could either say one played with better teammates than the other OR one is a bit luckier than the other (maybe a bit of both). If one player averages 300 MS and the other averages 250 MS the better player likely is the 300 MS average player and you can say the same thing about the equal win loss ratio. If all you consider is the w/l ratio though you cant really tell anything about either player because all you know is that in scenario 1 player B wins twice as many games as player A and in the second scenario both players win as often as each other. Win rate does not equal skill.

Take a 3rd scenario into consideration: player A after 1000 matches averages 340 match score and player B after 1000 matches has an average match score of only 172...which is the more skilled player? Now what if you were told player A has a .87 w/l ratio and player B has a 1.31 w/l ratio would you then say player B is more skilled than player A?

Congratz, you give the same answers as the current match maker. It's very good at it's job

Regarding scenario 3, yes, player B is better since he is probably only playing with a group and doing actions that don't earn MS but helps his team... like running a 10 flamer piranha.

Edited by Nightbird, 22 June 2020 - 03:44 PM.

[Vindicated]

C64 Warrior, on 22 June 2020 - 03:02 PM, said:

One game doesn't matter if you are analyzing 1000s...but it matters infinitely more when you are only using 20 to establish baselines. And this becomes the point of why wins/losses doesn't accurately portray INDIVIDUAL skill/performance...if both players had a match score of 260 (and that was just to pick a middling number) their performance is EQUAL, what happens to A's PSR should also happen to B's PSR regardless of how the match turned out or what the other 22 players did even if A and B are on opposite teams. But in a w/l system even though these two players performed equally, A will go up and B will go down. If you repeat the same match 1000s of times A ends up tier 1 B ends up tier 5 and you have the same dilemma that we have with our current MM because looking at someones win rate does not give you an accurate description of their performance in each match. People can do really well consistently and can still loose more games than they win just like people can do poorly and still win more often than they loose. Neither is a true measure of skill. How a person performs match after match, on good teams and on bad teams that is a far more accurate measure of their skill than how often they win matches will ever be.

I will have to agree with you that 20 games can be few enough to misplace people in tiers. Hopefully you do not end up in T1 the odds are against you. If you land in T5, you become the turning point. Ever see high damage, high kill posts? You can do that and enjoy it while it lasts, just try not to degrade in skill.

Any player misplaced in tier will become cannon fodder and will quickly drop, so don't worry about that. Even today, I've heard of people dropping out of T2 immediately after getting there.

C64 Warrior, on 22 June 2020 - 03:21 PM, said:

You dont make any sense, on the one hand you say "its random" but on the other hand you say "its a quantifiable statistic". If I see a person had a win/loss greater than 1 all I know about that player is that they win more games than they loose...but that doesn't telly if they are a good player or not or give you any indication on how accurate a shot they are, or if they are good a placing strikes or if they play objectives...it just tells me that they win more often than they loose

One result is indeed random. The average is not (keyword statistics). If you are telling me that you are a good player but cannot win games, does that make sense, or would I be more likely to believe you are just farming MS?

Besides, as I mentioned before, the W/L system is not pure W/L, it's still factoring in MS. If you are a consistently high MS player, you are still recognized to a lesser extent. And again as I've mentioned before, high MS does not mean you necessarily contributed more to the win.

There's going to be some sort of break even point between higher W/L lower MS and lower W/L higher MS. If you don't lose EVERY game and you are indeed good enough to consistently maintain high MS this will show. More importantly, if you are doing actions not reflected in MS (such as coms) to help your team win, this will also show, where these actions would be under represented in a pure MS system (even though there is currently a small MS bonus for winning).

[C64 Warrior]

Vindicated, on 22 June 2020 - 03:45 PM, said:

I will have to agree with you that 20 games can be few enough to misplace people in tiers. Hopefully you do not end up in T1 the odds are against you. If you land in T5, you become the turning point. Ever see high damage, high kill posts? You can do that and enjoy it while it lasts, just try not to degrade in skill.

Any player misplaced in tier will become cannon fodder and will quickly drop, so don't worry about that. Even today, I've heard of people dropping out of T2 immediately after getting there.




One result is indeed random. The average is not (keyword statistics). If you are telling me that you are a good player but cannot win games, does that make sense, or would I be more likely to believe you are just farming MS?

Besides, as I mentioned before, the W/L system is not pure W/L, it's still factoring in MS. If you are a consistently high MS player, you are still recognized to a lesser extent. And again as I've mentioned before, high MS does not mean you necessarily contributed more to the win.

There's going to be some sort of break even point between higher W/L lower MS and lower W/L higher MS. If you don't lose EVERY game and you are indeed good enough to consistently maintain high MS this will show. More importantly, if you are doing actions not reflected in MS (such as coms) to help your team win, this will also show, where these actions would be under represented in a pure MS system (even though there is currently a small MS bonus for winning).

I read your post but I still don't see anything that indicates how w/l ratio accurately measures someone's skill or performance although what I do see is you saying that even in a w/l system there needs to be some consideration for average match performance...because as I have said over and over w/l doesn't accurately measure a persons skill by itself. And this is exactly the reason why I support 1A over all the other options

[C64 Warrior]

Nightbird, on 22 June 2020 - 03:43 PM, said:

Congratz, you give the same answers as the current match maker. It's very good at it's job

Regarding scenario 3, yes, player B is better since he is probably only playing with a group and doing actions that don't earn MS but helps his team... like running a 10 flamer piranha.

And if Player A had the 1.31 w/l ratio then suddenly player A becomes the more skilled player or are you going to try to argue that player B with a .87 w/l ratio is still the more skilled player despite having an average match score about half of Player A...at what point do you look at what is in front of you and realize that w/l ratio really doesn't tell you what you think it does about someones skill/performance?

[Vindicated]

C64 Warrior, on 22 June 2020 - 03:40 PM, said:

If both players average MS is 300 then yes they are equally skilled...if one has a higher w/l than the other then you could either say one played with better teammates than the other OR one is a bit luckier than the other (maybe a bit of both). If one player averages 300 MS and the other averages 250 MS the better player likely is the 300 MS average player and you can say the same thing about the equal win loss ratio. If all you consider is the w/l ratio though you cant really tell anything about either player because all you know is that in scenario 1 player B wins twice as many games as player A and in the second scenario both players win as often as each other. Win rate does not equal skill.

Take a 3rd scenario into consideration: player A after 1000 matches averages 340 match score and player B after 1000 matches has an average match score of only 172...which is the more skilled player? Now what if you were told player A has a .87 w/l ratio and player B has a 1.31 w/l ratio would you then say player B is more skilled than player A?

Ok fine, we can play it that way. You have no problem taking the player with 300MS average and 1 W/L right? What if said player admits they are damage farming? As already I've mentioned those are my stats this season, and I currently I tend to play a certain type of mech (6 AC2) where I can rack up large amounts of damage (and kills/KMDD so even MS adjustment is not a solution) by being one of the last mechs on the team, and I know for a fact there are better players who have lower MS that I would have a very hard time winning against (checked the leaderboards).

By contrast the other player with 300MS and 2 W/L (they do exist) is probably using low total damage output/rate of file weapons, such as gauss and PPC. Said player is also strictly superior at deleting components, rendering a good part of the opposing firepower unusable without kills. To get such good W/L said player is probably also tactically superior and is capable of leading the team (as well as playing in support of the masses if they do not care to listen)

On 1 W/L player's team you would have died early, because I can't make calls and I'm average. 2 W/L player would have given you advice to keep you alive and possibly risk his neck to support you if you do something stupid.

So go ahead. Pick the 1 W/L player. You probably don't have a choice if you're making your team because the 2 W/L is already picked, and is charismatic enough to make their own team with equal good players.

[Laser Kiwi]

C64 Warrior, on 22 June 2020 - 04:05 PM, said:

And if Player A had the 1.31 w/l ratio then suddenly player A becomes the more skilled player or are you going to try to argue that player B with a .87 w/l ratio is still the more skilled player despite having an average match score about half of Player A...at what point do you look at what is in front of you and realize that w/l ratio really doesn't tell you what you think it does about someones skill/performance?

The issue is, this is a total strawman. Except in exceptionally rare circumstances these mechwarriors do not exist. People with a match score average 340 virtually all have a 1+ win loss, people with 170 match score are all less than 1. You can't argue from a position that doesn't even exist in 99.9% of the population. Maybe you can find one, maybe 5, but in a population of several thousand the vast majority have a positive kdr when the match score is high and vice versa and are by every metric a better player than the person with half their match score.

edit: win loss, not kdr, nobody cares about kdr

Edited by Laser Kiwi, 22 June 2020 - 04:29 PM.

[Nightbird]

C64 Warrior, on 22 June 2020 - 04:05 PM, said:

And if Player A had the 1.31 w/l ratio then suddenly player A becomes the more skilled player or are you going to try to argue that player B with a .87 w/l ratio is still the more skilled player despite having an average match score about half of Player A...at what point do you look at what is in front of you and realize that w/l ratio really doesn't tell you what you think it does about someones skill/performance?

Never?

W/L is a stat that predicts how good you're at helping your team win. MS is a stat that predicts how good you are at doing actions that award MS, which may or may not help your team win.

Isn't it obvious which is the better individual performance metric?

[C64 Warrior]

Vindicated, on 22 June 2020 - 04:14 PM, said:

Ok fine, we can play it that way. You have no problem taking the player with 300MS average and 1 W/L right? What if said player admits they are damage farming? As already I've mentioned those are my stats this season, and I currently I tend to play a certain type of mech (6 AC2) where I can rack up large amounts of damage (and kills/KMDD so even MS adjustment is not a solution) by being one of the last mechs on the team, and I know for a fact there are better players who have lower MS that I would have a very hard time winning against (checked the leaderboards).

By contrast the other player with 300MS and 2 W/L (they do exist) is probably using low total damage output/rate of file weapons, such as gauss and PPC. Said player is also strictly superior at deleting components, rendering a good part of the opposing firepower unusable without kills. To get such good W/L said player is probably also tactically superior and is capable of leading the team (as well as playing in support of the masses if they do not care to listen)

On 1 W/L player's team you would have died early, because I can't make calls and I'm average. 2 W/L player would have given you advice to keep you alive and possibly risk his neck to support you if you do something stupid.

So go ahead. Pick the 1 W/L player. You probably don't have a choice if you're making your team because the 2 W/L is already picked, and is charismatic enough to make their own team with equal good players.

But your argument here is based on the false assumption that we can PICK who is or isn't on our team...we can't, we get who we get and we have to use our SKILL to make the best out of what we get as teammates and opponents. Again, this is why your win rate doesn't measure how skilled a player you are.

[Laser Kiwi]

Nightbird, on 22 June 2020 - 04:17 PM, said:

Never?

W/L is a stat that predicts how good you're at helping your team win. MS is a stat that predicts how good you are at doing actions that award MS, which may or may not help your team win.

Isn't it obvious which is the better individual performance metric?

Never, not 1 player over 100 games played between 335 and 345 match score has a win loss less than 1, nobody

edit i meant win loss sorry


My only objection to the above is that the scenario 3 guy as strictly read doesn't exist. Good leaders and players with decent win loss records invariably gravitate towards higher match score than average even if it isn't exceptionally high. Caveat, they also need team mates that listen, usually playing as a team. Until recently soup queue didn't even make that a thing.

And the positive win loss guy with average match score 180 or so, he won't be doing that under any circumstances if hes playing thousands of matches over a career, it just won't work out.

Edited by Laser Kiwi, 22 June 2020 - 04:35 PM.

[C64 Warrior]

Laser Kiwi, on 22 June 2020 - 04:16 PM, said:

The issue is, this is a total strawman. Except in exceptionally rare circumstances these mechwarriors do not exist. People with a match score average 340 virtually all have a 1+ kdr, people with 170 match score are all less than 1. You can't argue from a position that doesn't even exist in 99.9% of the population. Maybe you can find one, maybe 5, but in a population of several thousand the vast majority have a positive kdr when the match score is high and vice versa and are by every metric a better player than the person with half their match score.

Noone has said anything about KDR, this is purely a discussion on w/l ratio. And you would be amazed at just how many players there are out there with a high average match score but a w/l ratio under 1 as well as players with relatively low average match score but a w/l ratio over 1.

[Nightbird]

Laser Kiwi, on 22 June 2020 - 04:24 PM, said:

Never, not 1 player over 100 games played between 335 and 345 match score has a win loss less than 1, nobody

edit i meant win loss sorry

Pages 1-25 of 900 from Jarl's list sorted by MS... and you're only then sure of >1 WL?

That's proof of how bad MS is.

Compare that to page 379 when sorted by WL

[Vindicated]

C64 Warrior, on 22 June 2020 - 04:23 PM, said:

But your argument here is based on the false assumption that we can PICK who is or isn't on our team...we can't, we get who we get and we have to use our SKILL to make the best out of what we get as teammates and opponents. Again, this is why your win rate doesn't measure how skilled a player you are.

You do get to pick if you're forming a group. I am willing to team with random scrubs (I have done this from time to time) because I don't care about stats or PSR and would rather help someone have a good time win or loss even if I don't know how to really help them.

But the point is whose team would you rather be on? Proven average in wins, or proven above average in wins? If someone else is winning more, there is a reasonable explanation. Part of that has to be a proven ability to work with random players of all sorts of skill levels.

[C64 Warrior]

Nightbird, on 22 June 2020 - 04:17 PM, said:

Never?

W/L is a stat that predicts how good you're at helping your team win. MS is a stat that predicts how good you are at doing actions that award MS, which may or may not help your team win.

Isn't it obvious which is the better individual performance metric?

How can you tell what someone does per match to help their team win by w/l ratio? I mean you could be AFK for a whole match and your team still win...but did you contribute? W/L ratio doesn't tell you anything other than that you won more games than you lost which again does not measure what a persons average skill/individual contributions to a match are

[Nightbird]

C64 Warrior, on 22 June 2020 - 04:33 PM, said:

How can you tell what someone does per match to help their team win by w/l ratio? I mean you could be AFK for a whole match and your team still win...but did you contribute? W/L ratio doesn't tell you anything other than that you won more games than you lost which again does not measure what a persons average skill/individual contributions to a match are

Yes, because on average someone who AFKs every match will have a WLR <.4 because that's how much not contributing 1 average player's worth of skill hurts every team you are on. You can test this, though don't get caught.

[C64 Warrior]

Vindicated, on 22 June 2020 - 04:32 PM, said:

You do get to pick if you're forming a group. I am willing to team with random scrubs (I have done this from time to time) because I don't care about stats or PSR and would rather help someone have a good time win or loss even if I don't know how to really help them.

But the point is whose team would you rather be on? Proven average in wins, or proven above average in wins? If someone else is winning more, there is a reasonable explanation. Part of that has to be a proven ability to work with random players of all sorts of skill levels.

IF I had the chance to pick who is on my team I would pick highest average match performance long before I would choose someone with an above average win/loss ratio because win/loss ratio is meaningless in determining someone's skill (or infinitely less meaningful than average match performance)

[Vindicated]

C64 Warrior, on 22 June 2020 - 04:33 PM, said:

How can you tell what someone does per match to help their team win by w/l ratio? I mean you could be AFK for a whole match and your team still win...but did you contribute? W/L ratio doesn't tell you anything other than that you won more games than you lost which again does not measure what a persons average skill/individual contributions to a match are

If you AFK, your team is fighting 11v12, so they would have to be above average to pull that off. Also, if you came back from AFK after you died, how do you know that the advice you gave your team over coms didn't help them win? Matches can end up close. For example, cored for cored, your teammate could have overheated and shutdown in front of the last mech when staggering the shots (something you suggest to a panicking teammate) would have allowed them to win. Or even further back, if the team is in a disadvantaged position and might not be realizing that.

[L1f3H4ck]

I'm in the whatever camp as long as it happens soon. I appreciate all the theory-crafters and contributions, but the way I see it, there is still a lot of hazy territory here, and I think a lot of that will become much clearer once the rubber hits the road.

This doesn't have to be a one & done deal, does it? I'm not opposed to the idea of seasons, tweaking as we go, similar to Solaris Divisions.

[C64 Warrior]

Vindicated, on 22 June 2020 - 04:42 PM, said:

If you AFK, your team is fighting 11v12, so they would have to be above average to pull that off. Also, if you came back from AFK after you died, how do you know that the advice you gave your team over coms didn't help them win? Matches can end up close. For example, cored for cored, your teammate could have overheated and shutdown in front of the last mech when staggering the shots (something you suggest to a panicking teammate) would have allowed them to win. Or even further back, if the team is in a disadvantaged position and might not be realizing that.

You're making my point for me...the fact that the team won or lost tells you nothing about what you or anyone else may have done in the match. The only thing you can really say about someone with a w/l ratio greater than 1 is that they win more games than they lose...it tells you NOTHING about what they contributed to those matches.

[Kamikaze Viking]

Ok People, Time to stop arguing.

https://twitter.com/...221935631118336

"@Paul_Inouye
Full update tomorrow... spoiler.. 2C is up to the plate.
9:49 AM · Jun 23, 2020"


Thankyou Paul. I eagerly await this update, and am ready to test.

Edited by Kamikaze Viking, 22 June 2020 - 05:03 PM.