52 replies to this topic

[Khobai]

Quote

So reward people for playing like a selfish ***** or gaming stats and driving antisocial performance I a team v team game.

not really. since a high match score generally means you helped your team. there are some scoring elements that should be changed, but for the most part that holds true.

and yes its a team game. BUT matchmaking tier doesnt rank teams. it ranks individual players. and matchscore is the metric for how well an individual player does. since carrying is fairly difficult in 12v12, winning/losing has very little to do with how good a player is. a player can completely suck and still win. a player can do extremely good and still lose. matchscore doesnt lie though, if your matchscore is consistently high, youre a better player than someone whos matchscore is consistently low.

the only way to get balanced teams is if matchmaker accurately represents player skill and divides the skilled players equally among both teams equally. Player skill is currently misrepresented which is one of the reasons we get such lopsided teams.

Quote

The would absolutely make the MM worse because it bases trying to match teams odds of winning on behaviors only tangentially related to winning.

but if match score only rewards behaviors that are related to winning your concern is then a non-issue.

yes match score might need some changes, but its certainly doable.

the point is the way the matchmaker works now is completely terrible and theres room for improvement.

Edited by Khobai, 08 November 2017 - 09:46 AM.

[Tiewolf]

I guess its the event and the working Matchmaker.

If you have only tier 1-2 players in drops then lop sided wins become more frequently! Experienced players will use a percieved advantage from the poking phase and rush in as a murderball. Lower tier players have no such awareness and will stay longer in the poking phase. In extrem cases till the end of the match. The result is, that the losing team is getting some kills in the extended poking phase too so the wins do not look so lop-sided. In higher tiers that is only the case if you have a very static team or the losses in the poking phase are evenly destributed over the matchtime.

[MischiefSC]

Khobai, on 08 November 2017 - 09:37 AM, said:

not really. since a high match score generally means you helped your team. there are some scoring elements that should be changed, but for the most part that holds true.

and yes its a team game. BUT matchmaking tier doesnt rank teams. it ranks individual players. and matchscore is the metric for how well an individual player does. since carrying is fairly difficult in 12v12, winning/losing has very little to do with how good a player is. a player can completely suck and still win. a player can do extremely good and still lose. matchscore doesnt lie though, if your matchscore is consistently high, youre a better player than someone whos matchscore is consistently low.

the only way to get balanced teams is if matchmaker accurately represents player skill and divides the skilled players equally among both teams equally. Player skill is currently misrepresented which is one of the reasons we get such lopsided teams.



but if match score only rewards behaviors that are related to winning your concern is then a non-issue.

yes match score might need some changes, but its certainly doable.

the point is the way the matchmaker works now is completely terrible and theres room for improvement.

High match score just means your match score was high. There's a million ways to game that.

The only metric you can not game is win/loss.

For predicting win/loss the only useful metric is... win/loss. You can look at ancillary metrics to determine WHY and HOW someone has the W/L they do but they are ancillary metrics.

I get that you want to believe that you are a helpless leaf on the winds of fate and you have no relevant impact on your teams win/loss but that's absolutely 100% false. That's the reason statistics works. It's a fundamental facet of math. You are always 1/12th of your team. Your team is drawn from the same variable parameters as everyone else. Within about 40 matches you're close to settled, 80 matches you're settled, about 120 or so you're well seated within a couple percent.

Your match score is functionally irrelevant. If someone has an average match score of 180 but a win/loss of 1.9 over 200 matches as alan average over 3 seasons and another guy has a w/l of 1.2 but a 350 match score over the same sample size the first guy is more likely to win matches. Whatever guy 2 is doing it's not winning. Maybe he LRM camps, maybe he plays super cautious and so dies last and cleans up the damaged enemies his team wrecked before they died but whatever it is it doesnt win as much as guy 1.

Yes, people who win a lot also inflate match score because wins bonus your match score obviously. However that's a related consequence of winning - winning is not a related consequence of match score.

This is basic statistics.

Edited by MischiefSC, 08 November 2017 - 10:42 AM.

[Khobai]

Quote

High match score just means your match score was high. There's a million ways to game that.

not if you fix it. which is what I said in my post if you actually read it.

obviously some of the matchscore values need to change if you base tier ranking on match score.

but dont act like it would be difficult to fix. its easy to fix. its just changing numbers.

if you changed it so it cant be gamed it would work fine. much better than the current awful sytem.

Quote

Your match score is functionally irrelevant. If someone has an average match score of 180 but a win/loss of 1.9 over 200 matches as alan average over 3 seasons and another guy has a w/l of 1.2 but a 350 match score over the same sample size the first guy is more likely to win matches.

except the only way that could happen is if match score is awarded for things that dont help you win games.

which I already said would need to be changed.

so youre arguing in circles.

if matchscore is only awarded for things that help you win games, that can never happen. the player with the higher match score will always have the higher W/L ratio because match score would only be earned by contributing to winning. see how that works?

its just instead of all winning being treated equally regardless of contribution (or lack thereof) like the current welfare system, the amount you contributed to winning would actually matter. So if you contributed more to winning than some potato that did nothing, the ranking system would factor that in.

Edited by Khobai, 08 November 2017 - 10:57 AM.

[MischiefSC]

It doesn't matter what you fix. At all. Unless you just make match score a reflection of your win/loss - in which case just use win/loss.

There is no way to account for everything that wins. However winning does, very effectively, identify what wins and what doesn't. Consistent winners and people who drive wins adapt behaviors every match based on their team and the other team.

What you're doing is like saying you should determine how long it takes this one specific person to bake a cake by how many ingredients you use and what brand of oven. While those can impact it, if you just measure how long that person spends bakong a cake 100 times you have a vastly, vastly more accurate predictive model. Saying "well what if this one time X" means absolutely totally completely nothing at all for anything ever except that one specific time. One sample is one sample is one sample. An average of the same events in the same environment around the same person wash out the "but this one time in band camp" anomalies.

All that matters in what you do in a match for the matchmaker is "did it help your team win". If you are consistently doing what helps your team win it is reflected in your win/loss. Damage, kills, survival, KDR, component destruction, LRMs shot down, lance in formation, the value of that to winning varies every match.

I really really get that you want to divorce your performance from the team and equate that to winning/losing but it doesn't work that way and can't work that way because you're always on a team of 12. I all but guarantee you that's exactly why your win/loss is consistently poor while your match score bounces around.

Edited by MischiefSC, 08 November 2017 - 11:10 AM.

[scadateck]

IMHO this is a function of the game attracting fewer (if any) new players and retaining less existing players. I hate to say it, but I'm starting to agree with those that are saying these may be the end times for MWO. Sucks because I've only been playing for a year or so, didn't get the memo in 2013 apparently.

Also IMHO, "git gud" or "always been that way", etc. responses should really be interpreted as "I want this game to die ASAP, and am doing my best to hasten its demise, so GTFO and reduce the population further". I suppose we'll see if it can survive as an exclusively e-sports instrument (and a niche one at that), which seems like the ultimate conclusion of this trend.

We can be assured that financial analysis at PGI has determined certain minimum metrics to continue justifying the expense of keeping the lights on. How close are we? That's the question I want the answer for, not what the next balance-pass or iteration of the match-making algorithm will do.

[TLBFestus]

I really haven't been playing much for months now. I would have thought that that would cut down on the number of lop-sided loses experienced by my team-mates by now.

[Troa Barton]

From my observations it's which side loses an assault first usually loses the match. Which isn't hard because assault mechs melt faster than a T800 in a steel mill.

[Grus]

Maybe people have realised this is a numbers game. If you get the first kill, then another, and another, you have a huge advantage and can be aggressive. Once there victory is far more likely as long as your team just dosnt go full ******.

[Xmith]

I think most matched teams are fairly even in skill. Stomps may depend on which team make more crucial mistakes. There could be at least 4 to 5 instances of mistakes by individual players on any given team. Too many mistakes will snowball into a stomp.

[kuma8877]

I actually have lots of very close matches. 8+ kills on the losing side happens fairly regularly for my games. Even when they are less than that, they are quite often decent games where a tactical error triggers the domino effect in a much more dramatic fashion (kill ratio per team wise). I personally don't have that many outright stomp experiences (+ or -) in QP and I only solo que.

[LordNothing]

bubble up means that as soon as everyone hits t1 that the tier system and matchmaker is no longer sorting players by skill. its still there slowing up game formation, but its meaningless. for the sake of the game try changing it to a symmetrical system so that people's tier actually represents their skill. really just try it for a month and see what happens.

Edited by LordNothing, 08 November 2017 - 07:13 PM.

[Savage Wolf]

Khobai, on 08 November 2017 - 09:05 AM, said:

its like this even when there arnt events going on.

its what happens when matchmaker puts people in tiers based on win/loss instead of matchscore

because all the potatos get carried into higher tiers they dont belong in simply by playing enough games

your tier ranking should only go up based on match score. win/loss should have nothing to do with it.

matchmaker might actually work if your tier reflected your individual skill rather than your ability to get carried by others

Since it's easier to rise than falling in PSR, it becomes an XP bar, which is the problem, not what it's focused on. In the end it's actually based more on the matchscore than win/loss. I'd wish it was based more on win/loss, because then people might actually try to win.

[JadePanther]

Alcom Isst, on 08 November 2017 - 06:27 AM, said:

Because this game snowballs hard. Once one mech dies the opposing team has an immediate advantage and the match snowballs from there. Even super balanced teams with a 50/50 chance will often end up 12-0 because of this.

yet i know for a fact this can go the other way.. especially if unfortunate circumstances are advertised and become motivational to others to work harder.. Yes i've seen it a number of times.. Oh crap did our guy just DC once his mech hit the ground.. <team> YEP! and then they proceed to stomp the enemy minus one mech from the drop..

EDIT... PS..
Also alot of it is the SKILL TREE GAP of mechs that have little or no SP invested vs fully skilled monsters.. mechs with craptons of extra armor vs ones that you can flip a coin to see if they overheat or get cored out first (hint both happen really fast to a mech without SP).. And every new mech is a mech without SP..

Edited by JadePanther, 09 November 2017 - 12:08 AM.

[Nik Reaper]

So I would think we need to do some experimentation, see how much the stats that are tracked are correlated to win/loss and base ranking around those, in to a ladder system wih x to y is tier 1 and so on, with easy and fluid up down movement on that ladder.

I would like to see what happens if one composed a team with high k/d and pit it against a team with higher w/l but lower k/d, and same with team with higher average match score vs higher w/l, so that every one on one team has higher " stat of interest" than there counter part on the other team but the other has higher w/l in turn.

This is basically putting a team of probably good shooters with killer mechs against a team of ... good team players? though I would expect that on the leader-board most players with high w/l are players that ether heavily use group play with a competent team or at least do a good mix of solo play with group play, so not sure if that is a fact or if that would be very influential on the end result.

I'm not saying that this is the answer but I would like to see some hard data sets with these conditions, and at the end of the day what I feel most are complaining about is not so much the I win screen is not in front of them often enough, it's the sub 100~200 dmg players who any way you look at them, there mech, build, aiming and movement have no reason to be in that match but are stuffed there anyway.

Only after that is the "we should go here coz I has ER-LL mech" while most others are brawler and medium range builds, even when you have 1 or 2 players who want to do there thing, if they are competent pilots they will do some work before they go down on there own, or they will retreat and rejoin the team after taking damage, none of what is as bad as a before mentioned low damage out of place pilot.

Edited by Nik Reaper, 09 November 2017 - 01:51 AM.

[meteorol]

Mole, on 08 November 2017 - 08:46 AM, said:

Why are so many of your acting like this is a new phenomenon? I joined this game back in early 2014 and it has ALWAYS been this way.

This.

MWO is simply has a gamedesign that often leads to snowballing matches, because it leaves little room for comebacks. Especially since we have 12v12. If one team quickly loses 3 mechs without dealing significant damage in return, armor and firepowers begins to heavily stack against them.

The huge TTK (compared to other FPS) makes comebacks pretty unlikely. I can't count the times i won 1v3 while having 25% HP against fresh players in CS. In MWO, you are almost guaranteed to lose in such a situation, because you can't kill them quickly enough.

[Asym]

There is not any "balance" to the game anymore because PGI's attempts to "balance" have done just the opposite of what they intended: they may have balanced weapons but fractured the culture... The May SkillTree drove of a larger number of people than we think; in more ways than we think. CW drove off an even larger segment and neither population losses are pemenant losses.... Many regular players simply stopped playing 5 hours a day to 1 hour a day pr less. Teams stopped dropping as teams: the team I am in is a case in point... 57 members and less than 10 play monthly and only a few of us drop as a team at any one time and only have been dropping as a small lance in the last two weeks...

It's lopsided because there is not a "balance" in the population itself; and, because there isn't a Strategic goal anymore. Faction Play and the whole MW struggle to focus on just is not there anymore. People are "just messing around" and are bored to trears... The Comp community isolates itself and their interests are not the interests of the larger niche market. They just aren't. We are seeing boredom in action and there is no symetry, because there is no order, because there is no point, no common challenge to work towards anymore.... Many players are asking themselves: why am I here? JMHO.

[Alan Hicks]

Yes and that made me quit the game. Someday I'll come back, when I feel like a masochist or something. Not today.

[Angel of Annihilation]

I have been around for a long while and will say that stomps have always existed in this game. However the amount of stomps have seemed to increased pretty substancially lately. Where is used to be maybe 1-2 stomps out of 10 matches, it now seems like I get 5-6 stomps out of 10 matches. Mind you, this has only been since I hit Tier 1 and the tightening of the Tier 1 MM so I assumed it had something to do either with the Tier 1 players or me personally.

[JadePanther]

Viktor Drake, on 09 November 2017 - 05:27 AM, said:

I have been around for a long while and will say that stomps have always existed in this game. However the amount of stomps have seemed to increased pretty substancially lately. Where is used to be maybe 1-2 stomps out of 10 matches, it now seems like I get 5-6 stomps out of 10 matches. Mind you, this has only been since I hit Tier 1 and the tightening of the Tier 1 MM so I assumed it had something to do either with the Tier 1 players or me personally.

agreed they can always happen.. long ago though it was easier and not unheard of to make big comebacks.

I'm leaning towards skill tree as the factor.. no longer is is 3-5 matches to have basic heat mgmt skills.. Most new mechs could have basic maneuverability and heat skills in a very short and quick time.. Add in the fact that now many enemy mechs are sporting 10-20% more armor or firepower than they used to have, and you can start to see how much harder it is for a factory fresh mech to earn the skills to compete.. xp earning rates are down, partly due to the disparagingly weakened nature these new mechs face... 800xp as an average was ok in the old system but in skill gap land average performances are more like 500xp..

It simply takes longer to get a new mech up and running to snuff in the new SP sytem.. Sure you dont have to climb 3 hills to master the mech, But its one bigger steeper mountain climb in a harsher season.. in the old system i'd have 3 varients finished or on the high side of topped off before the leaderbaord event got into swing.. Now it takes dedicated work to get one..

Combine all that with the factor that now mechs that didnt have weapon or armor quirks now have access to that via skill tree..