47 replies to this topic

[Sub5even]

I played around 20 games today, in 15 of them either we got stomped or we stomped the enemy. Exactly one game was close and ended with 12 to 10 kills.

To be honest...this way MWO is boring like hell!

I would definitely prefer to wait 2 more minutes to get equal opponents.

[MW Waldorf Statler]

the most Players not have any Experience or Tactical Awarness ...sadly ..fast run in the red fire ist the most used Tactics ,or stand behind and firing LRMs blind in the Terrain .

[Nesutizale]

I would say that the problem is not just the players skill. Even with balanced teams you would run into the problem that if one side gets a first kill and then pushes, they will get more ground.
Even a less skilled team can win if they see the right time to push or just follow the lead given. Thats why in FP the coordinated teams are very likely to win because they coordinate. Something general PUGs don't do very often.

Most of the time when I see a team loose in FP or QP its because people don't follow a push or think that they alone can flank the enemy.
Question is how are you going to take that into account when creating a matchmaker? How should a mere programm get the data if you are a coordinated player and see if you have learned to torso twist and follow a lead?

If all they have is KMDD, Kills and wins...that is hardly any evidance of how good you are as a teamplayer...and thats where most of the battles are won.

[EnochsBook]

Measuring winrate in terms of sessions of a couple dozen games is a mistake, there isn't enough data to tell you anything other than you were unlucky.

A big part of why "stomps" are so common in MWO is the dynamic of the average match in which one team groups tightly together and rolls over the other team that split up early. That's going to create a lot of 12-0 to 12-4 scenarios.

However, the MM doesn't account for skill and it is true that it probably should.

Nesutizale, on 10 March 2018 - 06:13 AM, said:

Most of the time when I see a team loose in FP or QP its because people don't follow a push or think that they alone can flank the enemy.
Question is how are you going to take that into account when creating a matchmaker? How should a mere programm get the data if you are a coordinated player and see if you have learned to torso twist and follow a lead?

Difficult but doable. Create an aggregate, hidden or public statistic for every player that takes into account the following parameters:

• Winrate, in my opinion paramount when trying to evaluate skill.
  
• Average match score, not necessarily the best metric, but can be adjusted to be more accurate.
  
• Weigh these stats in relation to the classes (or even the 'Mechs) the player plays the most, eg doing well in a class/'Mech a lot of people do very well in should be seen as not as good as doing well in a class/'Mech most people do not do well in.

Basically the MWO equivalent of World of Tanks' WN8.

Then you have a metric on which to base your Matchmaker. Far from perfect, but a lot better than simply going off of tiers.

Edited by EnochsBook, 10 March 2018 - 06:27 AM.

[Tarl Cabot]

What? Can not be bother to post in one of the two threads that are on the first page, one MM bonkers and the other Stomps are normal?

https://mwomercs.com...esign-features/

https://mwomercs.com...s-just-bonkers/

Look at the MWO Championship Events. Many of those matches would be considered "stomps" and they were made up of players working as a team, communication, etc, not just a bunch of SOLO pugs thrown together. WTF do you really expect? The side that is more aggressive, takes CAPITALIZES the other team's mistakes, it becomes a snowball effect.

Besides the players, there is an event which requires those participating to use specific mechs, but again if one side is NOT working together it is all for nothing. MM matches those in the same or close Tiers using PSR which is an experience bar and to separate the newbies from the higher skilled vets. I say higher skilled vets because there are vets who have been here a long time and they are still T4-T5 for many reasons, some outside their control but for others a refusal to adapt and improve. A few are actual okay where they are at because they do not have to push themselves to change.

[Nesutizale]

I wouldn't value the winrate that high in a teambased game. The best player could have a bad winrate when paired with bad players and the other way around.

I see that in MWO as well as World of Warships. While I don't count me to the good players I have friends who are pretty good in WoWs and they can't turn games around alone.
Also they are tracking stats alot and allways wonder how I, with a terrible winrate, can actually affort so many ships and I even have fun playing that game. Its mostly because I play the T7 fun devision where no one cares as much and they play high Tiers where every loss is felt very hard with your pocket

Anyway what I learned from that is that winrates in a teamgame say next to nothing about your skill.

Average match score at least tells how good you are, so putting people with similar scores together might help. Its still lacking as I find matchscores not only in MWO but other games rather confusing at times as they mostly promote only one thing you can do.
For example only giveing points for doing damage but not scouting, though in WoWs they finaly added points for that as no one was interested in scouting AKA playing a certain type of ships. MWO could learn from that.


Still these stats hardly give you something in terms of teamplay.
What might help are thing like taking into account

- How often do you follow a push
- Do you follow orders given my VoiP or commandwheel.
- How often did you successfully torso twist to migrate damage
- Do you share armor?
- When using LRMs, are you staying in the back or are you getting your own target locks
- Do you scout?
- Do you play objectives?

All these things would give much more valuable information about you as a team player then simple Kills and wins. Those datas are more often influanced by what your team does or dosn't then your skill alone.

[TWIAFU]

Sub5even, on 10 March 2018 - 06:00 AM, said:

I played around 20 games today, in 15 of them either we got stomped or we stomped the enemy. Exactly one game was close and ended with 12 to 10 kills.

To be honest...this way MWO is boring like hell!

I would definitely prefer to wait 2 more minutes to get equal opponents.

LOL.

What did you expect to happen in a QP Event with specific mechs most don't even use?

Best thing to do is NOT QP.

[mogs01gt]

Sub5even, on 10 March 2018 - 06:00 AM, said:

I played around 20 games today, in 15 of them either we got stomped or we stomped the enemy. Exactly one game was close and ended with 12 to 10 kills.

To be honest...this way MWO is boring like hell!

I would definitely prefer to wait 2 more minutes to get equal opponents.

it was rough last night. It was either extremely passive players, nascar or people just swooping in for a kill steal.

[Dogstar]

Stomps are normal

'nuff said.

[EnochsBook]

Nesutizale, on 10 March 2018 - 06:52 AM, said:

I wouldn't value the winrate that high in a teambased game. The best player could have a bad winrate when paired with bad players and the other way around.

[ ... ]

Anyway what I learned from that is that winrates in a teamgame say next to nothing about your skill.

For a small sample of games you would be correct.
However, over a large number of matches the impact an individual player has on the game will be reflected in their overall win rate. As I said in my first post, you can be "unlucky" during a session, you might be "unlucky" over 100 matches, but no-one is consistently unlucky over several hundred matches. That's the point at which we will see individual player skill have an impact in how many games they win.
After several thousand matches the good teams and bad teams balance each other out like the results of repeatedly rolling dice, and the only consistent factor affecting any deviance from the average is the player.

Nesutizale, on 10 March 2018 - 06:52 AM, said:

Still these stats hardly give you something in terms of teamplay.
What might help are thing like taking into account

- How often do you follow a push
- Do you follow orders given my VoiP or commandwheel.
- How often did you successfully torso twist to migrate damage
- Do you share armor?
- When using LRMs, are you staying in the back or are you getting your own target locks
- Do you scout?
- Do you play objectives?

All these things would give much more valuable information about you as a team player then simple Kills and wins. Those datas are more often influanced by what your team does or dosn't then your skill alone.

In the end and overall, all these actions will affect a player's winrate. If someone has bad teamplay (they do not help their assaults against lights, they do not use voip and communicate, they don't play the role their 'Mech is supposed to play, etc...), then the team they are on is less likely to win than if he did all these things.

Winrate over a significant number of matches played is just the simplest, most reliable way of taking all these behaviours into account.

[adamts01]

Nesutizale, on 10 March 2018 - 06:52 AM, said:

I wouldn't value the winrate that high in a teambased game. The best player could have a bad winrate when paired with bad players and the other way around.

A month or two ago someone posted some insanely detailed work on team vs team player stats. Out if an extremely large sample size he found average win rate of a team to be the most dominant indicator of victory, followed by average match score. Given enough matches, I think it's the best metric to go off of. Considering match score would dial it in, but it doesn't seem necessary. Taken in to account a certain mech's win rate seems more important than complicating the MM formula with other stats, especially since we're talking PGI here, who doesn't think skill needs to be measured, on experience.....

[Nesutizale]

Okay. I wouldn't have paid to much attention to winrates but if its actualy working out so be it.

[UnofficialOperator]

I've accepted that every now and then a newbie player unaware of how mwo works would post a thread like this again. Just point them in the right direction of the stomp threads

[MW Waldorf Statler]

its like Battlefield

the own team not have Jetpilots, and nothing use the AA and the other Team has Bomberpilots =Stomp ...To many Players have other prioritys as Match win ..the most will Kills ...Event-targets ...testing new Chassies ..or learn the basics.
in War Thunder im can buy a B29 come in the High Rank Games of Later Periiods and loose...im become the B29 first of normal Way only after many many Matchs and a Long Skilltree trough other Chassies..here im can my buy fast a Assault.
in Battlefield im can me search a Server with a good Team and im have respawn to learn in a Match from my Faults and try other tactics .

Its here like Robocraft ...how many Folks im seeing in RC with not moving builds or Weaponless in matches

Edited by Old MW4 Ranger, 11 March 2018 - 10:14 AM.

[Nesutizale]

I have looked in the stats that i have for WoWs a bit closer.
My winrate is actually higher then I tought but looking at the others stats makes me question winrate as a good mesure again.

For example my friend and I have nearly the same winrate. Still he does nearly twice the damage, has better K/D and basicly all other stats are better. Still when just looking at the winrate one could think that we are on the same level while I mostly contribute far less to each battle then he does.
(We both have about 5500 matches so there should be enough data to have good average values)

So winrate mght be good to find a group of average same level while there starts to be a big gap when lots of my skill would fight a lot of people of his skill.
Theirfore to get groups of similar skill you would have to take more then just winrate into account.

Will that prevent stomps? No, the game is not designed that way as you can read in the other topic that has been mentioned here before. Would it result in games beeing closer more often...I think so.

Something that crossed my mind is close matches result mostly in longer playtimes...or better said longer waittimes until you get your mech back.
Do we want that?

[UnofficialOperator]

Nesutizale, on 10 March 2018 - 06:52 AM, said:

Anyway what I learned from that is that winrates in a teamgame say next to nothing about your skill.

Truly one of the most naive things I've heard all day. Explain why there are players with consistent positive WLR and KDR (winning players) and why there are players with consistent negative WLR and KDR like yourself then? Explain why there are players with very high WLR who seem to be perennial winners?

Edited by UnofficialOperator, 11 March 2018 - 05:30 AM.

[adamts01]

Nesutizale, on 11 March 2018 - 02:30 AM, said:

I have looked in the stats that i have for WoWs a bit closer.
My winrate is actually higher then I tought but looking at the others stats makes me question winrate as a good mesure again.

For example my friend and I have nearly the same winrate......

A few things to consider:

W/L does require a very large sample size, as PSR/match msker are near-worthless, and we all have winning and losing streaks.

W/L is the most important stat, so you can take all others with a grain of salt. For example, a good player would make the winning decision to not chase a stick off the side of the mountain and instead focus on a threat, while a bad player would bump his KDR by leaving the fight and securing a kill on a weaponless target.

Tier matters. A tier 1 player with a (-) W/L might be considerably better than a tier 3 player with a 2:1 W/L.

Computers can't accurately measure your contribution. You might be a terrible shot but an incredible leader, which earns you a good W/L ratio. I don't know what it is that sets you at the same level as your friend, but you're doing something right that other stats aren't able to track.

[EnochsBook]

adamts01, on 11 March 2018 - 05:48 AM, said:

I don't know what it is that sets you at the same level as your friend, but you're doing something right that other stats aren't able to track.

Most likely he is playing with his friend a lot in a group, do this for long enough and the winrates for both players will become very close.
It could also be that his friend does good damage and has a comparatively low winrate because he's not much of a team player, whereas Nesutizale is the opposite.

The point is winrate is the final, most important metric when trying to evaluate skill, because it's literally how likely you are to win a game, which is how a good player is defined.

adamts01, on 11 March 2018 - 05:48 AM, said:

Computers can't accurately measure your contribution.

I would say that they can, provided that we can come up with the right stats and equations.

[Asym]

UnofficialOperator, on 11 March 2018 - 05:30 AM, said:

Truly one of the most naive things I've heard all day. Explain why there are players with consistent positive WLR and KDR (winning players) and why there are players with consistent negative WLR and KDR like yourself then? Explain why there are players with very high WLR who seem to be perennial winners?

OK, this is simple to explain: there is so much 'variation' in play that some players find a niche for themselves that allows them excellent statisitcs.... Take for example a noob who is adopted into a successful competitive team.....his/her statistics will be generally higher when competing with those comp level players as compared to another player who is constantly dropping with a team of "friends" who are not competitive at all !!! Skill? The lesser statitical pilot could be greatly better at the game and just doesn't have the numbers to prove it, especially, in a game that so un-balanced and messed up as MWO is....

I played MWO with a team "for the fun of it" and quite happy being a potato.... In another game, I play with a team of ex-professional FPS comp players....want to bet where my stats in the games reside? MWO = Potato and "Other game" = comp player...... It's not a question of skill; it's a question of environment....

Anyway, it's suppose to be "fun", a place to relax and goof around and chase "stompy robots", Space ships, Sailing ships, Armored vehicles, Alien creatures and of all things Dinosaurers around maps....... Statistics are for those who are un-able to, incapable of or compelled to "prove something" to find value for their time.......the rest of us, we just play for the fun of it !

And that's why, many of us have retired from MWO for the moment....... Good hunting and forget where you are and enjoy what you have.

Edited by Asym, 11 March 2018 - 07:01 AM.

[adamts01]

EnochsBook, on 11 March 2018 - 06:40 AM, said:

I would say that they can, provided that we can come up with the right stats and equations.

Not until we have some seriously powerful AI. Just take scouting for instance. Did that scout a mech just performed help? Or did it give away your team's position? There's no argument that charisma and leadership goes a long way in solo que, and good luck getting a program to recognize that. How about a light sacrificing his mech early game to stall a base cap? Or a light dealing zero damage as an effective squirrel? Match score is in a much better place than 2 years ago, but I don't see it accounting for the above mentioned circumstances anytime soon. It's a decent addition to W/L, but that should be about it.