29 replies to this topic

[Knuckles OTool]

MischiefSC, on 05 March 2018 - 02:30 PM, said:

Go to the excellent utility, The Jarls List. It's pretty much ranking people off match score.

However what you'll see is that a lot of the best players in the game are not even in the top 100. I mean people who were in MWOWC finals sort of good. You'll also see, as you go down the list, a lot of people you recognize in things like LRM Scorch, with a 1.2 or less W/L but who always do a lot of largely worthless damage and play every match from the very back so they die last but functionally help ensure their teams failure.

Match score is a terrible mechanic for sorting out who is actually good or bad at the game. If you played every match with a Deathstrike and never hit anything but CT and cockpits you could have a w/l over 3 or 4.0 and spend most matches carrying like Atlas but have a average match score of like 275; because you're killing mechs with ~100 damage each.

Sort by w/l and look down the list again. The top 100+ are going to be top performers you recognize. I know people don't like it but the reality is that good players win more often than bad players on average. That the best indicator of how good someone is at winning (not farming damage or anything else but actually winning) is.... how often they win. Crazy, I know. Inconvenient if you don't win as much as you think that you should, I know. However if you want to win more often then you should do more of what actually wins. If don't win as much as you think you should than whatever you think actually wins? Probably doesn't actually win as much as you think it does and you should look into that a bit more. Maybe watch/talk to the people who win more than you.

While this is mostly true, you will also see good players farming ST's for kill's when they arent pressured. Got the drop on someone? Eliminate one ST and then the other and get 2.5 to 3x the damage if you had just hit CT. You can watch great players do this all the time.

Opening a mech in any torso is anything but worthless damage. Taking out their weapons from range is pretty useful. Lrm boat opening a bunch of big mechs for you? Well its payday in your MG light.

Group queue skews the hell out of jarls list anyway.

And I am absolutely positive if we go by match score that I am nowhere near a top 5% player with some of the skilled guys I see around. Skewed stats are skewed no matter what.

Edited by Knuckles OTool, 05 March 2018 - 05:56 PM.

[MischiefSC]

Eisenhorne, on 05 March 2018 - 02:34 PM, said:

I realize that, but I'd rather have a guy who inflates his score to 1000 dmg of random shots than a guy that does 200 damage of CT only shots.

Also, I personally get higher damage done scores from mechs running laservomit or dakka than I do from splat mechs anyway, because of the much better weapons range.

Odd, because the first guy is largely irrelevant to the overall winning of the match, the second guy drives wins. The second guy can and will literally help win the match more often than the second guy.

[MischiefSC]

Knuckles OTool, on 05 March 2018 - 05:53 PM, said:

While this is mostly true, you will also see good players farming ST's for kill's when they arent pressured. Got the drop on someone? Eliminate one ST and then the other and get 2.5 to 3x the damage if you had just hit CT. You can watch great players do this all the time.

Opening a mech in any torso is anything but worthless damage. Taking out their weapons from range is pretty useful. Lrm boat opening a bunch of big mechs for you? Well its payday in your MG light.

Group queue skews the hell out of jarls list anyway.

And I am absolutely positive if we go by match score that I am nowhere near a top 5% player with some of the skilled guys I see around. Skewed stats are skewed no matter what.

Clean hits and effective hits are doing less total damage per match than a lot of mediocre hits and splatter damage. CT drilling with a Deathstrike may be ideal for killing but you're absolutely correct that there's value in disarming/crippling an enemy mech - the difference is that a good player does so way, way more efficiently than a guy just hiding in the back, sniping at anything/everything he can, spreading his damage due to bad aim but lasting a long time because he is always the furthest in the back of his team. That guy will have high damage and a lower win/loss than a top performer, by a good stretch.

Which is carried out very clearly in the Jarls List.

Group queue skews stats for both match score and win/loss. However there's fewer players playing consistently in group queue than there is in FW, which is saying something. Something depressing.

By match score you are in the top 3% of players. By win/loss you're a bit above average, about top 15%.

Up to you where you feel you're better placed.

Edited by MischiefSC, 05 March 2018 - 09:09 PM.

[Khobai]

its a shame that damage isnt weighed based off where you hit

doing 1000 spread damage should not get rewarded better than 500 CT damage for example

[radiv]

Zergling, on 05 March 2018 - 12:04 PM, said:

Nah, then the hordes of LRM boaters would whine because they are being set by the matchmaker at a level well above their actual skill level.

And how is this different from todays system?

[Vellron2005]

Electron Junkie, on 05 March 2018 - 09:38 AM, said:

Can we just have match maker go off average match score???
Just go right down the line and put every other player on opposite teams... Would be a hell of a lot better than the mess we currently have.

Average match score is not a good indicator of skill, cose' it's very dependent on game mode and the rest of the team..

For instance.. if you're focusing the whole match to give your team an edge by carrying batteries or capping, you are not gonna have a good matchscore, while actually being a great asset.

Likewise, you can be the best diewolf pilot in the game, and can still find yourself left behind with a school of piranhas chewing off your rear torso cose' your team didn't come back to help.

Simply not a good indicator of skill..

[Asym]

You can't use the data collected because it is so corrupted at this point.... Garbage in garbage out, yes?!

"Average" is a dangerous word because it infers "a just outcome" and many assume that numbers in the middle value are good....

Median numbers sometime are a better representation of the middel values but even medians are dangerous... Especially, in skewed demograhpics and experience pools....

NO. Turn it off and run with totally random teams and the game would be a lot more interesting and, if I am right, nothing would change in gameplay. Sad, but true.

[Eisenhorne]

Vellron2005, on 06 March 2018 - 01:45 AM, said:

For instance.. if you're focusing the whole match to give your team an edge by carrying batteries or capping, you are not gonna have a good matchscore, while actually being a great asset.

I've seen more games lost by mechs spending the entire game "capping" when their firepower was needed at the front. 99% of my wins in conquest are when we're down on caps, kill the entire enemy team, then de-cap. Lights should get us to 3 caps, then rejoin the main body to pick off weakened enemies / defend assaults vs lights. Similarly, on Incursion, the lights should grab the batteries, and while they're doing that the main body of our force engages the enemy. When they come back to the front, they should then focus on killing open enemies, not on getting more batteries. The lighter mechs should play the mission at the start, then once the actual fight begins with the heavies and assaults get back to help. This will result in a decent damage done / match score.

Vellron2005, on 06 March 2018 - 01:45 AM, said:

Likewise, you can be the best diewolf pilot in the game, and can still find yourself left behind with a school of piranhas chewing off your rear torso cose' your team didn't come back to help.

First, don't bring direwolves, they're terrible mechs. Slow as an Annihilator without the armor to back it up. But you need to be aware that in quick play your team will never support you. Ever. You need to be 100% on top of the way your team is rotating and never stop moving. You go slow as hell, but you can PROBABLY keep up kinda if you never stop moving. Most of the dire whales I see die are the ones that stop to peek a ridge or something, while the rest of the team runs away. If you never stop moving, and fire only on the move, you can keep up because your heavies / faster assaults will generally stop every so often to poke the enemy. Yes, it means you get fewer shots in, but you don't get isolated and die, so you live longer, so you do more damage (and get a higher match score).

[Ronan]

Just using match score isn't very helpful, as it is a "shared" score -- meaning there's only so many kills to be had in a match, only so much damage that can be inflicted.

Now, if a match is made up of very differing skill-level pilots, then the low-skill pilots will get low match score, and high-skill will get higher scores. That's good.

And if the match maker actually builds a team of 12 equal-skill-level players, and they all actually try to contribute, you're going to see a set of match scores that are all about the same. If the match maker is really on a roll, there will be 24 equally skilled players, all doing their best, and all getting round-about the same score. Could be a full-potato game, could be an ace-filled skill-fest. But they all just got similar scores. Which then puts potatos with aces as having the same score.

Potentially.

A new ranking/matching algorithm might help, just don't think matchscore alone is it.

[N0MAD]

MischiefSC, on 05 March 2018 - 02:30 PM, said:

Go to the excellent utility, The Jarls List. It's pretty much ranking people off match score.

However what you'll see is that a lot of the best players in the game are not even in the top 100. I mean people who were in MWOWC finals sort of good. You'll also see, as you go down the list, a lot of people you recognize in things like LRM Scorch, with a 1.2 or less W/L but who always do a lot of largely worthless damage and play every match from the very back so they die last but functionally help ensure their teams failure.

Match score is a terrible mechanic for sorting out who is actually good or bad at the game. If you played every match with a Deathstrike and never hit anything but CT and cockpits you could have a w/l over 3 or 4.0 and spend most matches carrying like Atlas but have a average match score of like 275; because you're killing mechs with ~100 damage each.

Sort by w/l and look down the list again. The top 100+ are going to be top performers you recognize. I know people don't like it but the reality is that good players win more often than bad players on average. That the best indicator of how good someone is at winning (not farming damage or anything else but actually winning) is.... how often they win. Crazy, I know. Inconvenient if you don't win as much as you think that you should, I know. However if you want to win more often then you should do more of what actually wins. If don't win as much as you think you should than whatever you think actually wins? Probably doesn't actually win as much as you think it does and you should look into that a bit more. Maybe watch/talk to the people who win more than you.

Reason most top players have beter W/L ratios is that they usually play as part of a team and usually against pug groups, this will ensure a better win ratio, their lower match score rating shows they are generally being carried by the team.
Most people that have long term experience with any game computer or not knows that most teams are not made up with everyone being an Ace, you have Aces down to very average players, that why you see a few in a team being payed millions and others half or less, if they were all top aces you would think they get same pay rate.
Until the stats are broken up into Solo / Group Q Jarls or any ranking stats are just incorrect and useless.