61 replies to this topic

[50 50]

We seem to get a fair number of metrics from each match but need to focus on personal stats and not team stats.
A win/loss ratio can be influenced by an individual but is ultimately a team stat so is not a great metric.
Match score might be the obvious one but does not necessarily tell the whole picture as there are ways to pad the score.

It does need to be averaged out over a number of matches.

Does using the stats we get for the match bonuses like lance in formation, saviour bonuses, KMDD and all those other things need to be taken into consideration?
Does it need to be identified according to weight class and then averaged over that as well?

[John Bronco]

W/L. As has been proven many times over solo queue W/L is the most powerful predictor of performance of the metrics that are available to us.

[Tarl Cabot]

Elo was created to rank 1 player vs 1 player, W/L.

• PGI Elo seeded the MM with a player then selected players within a specific range.
  
• We also had hard, or very rigid weight classes
  
• in an 8vs8 environment
  
• Groups 4 or less used AVERAGE Elo numbers (that sure was a big joke)
  
• Once MM was finished, the highest avg Elo was selected to be the winning team
  
• If said team won, they moved up some and the losing team went down some
  
• If said team lost though, they took a big hit and the expected losing team received big numbers, equal
  
  • With the above, it did not matter if you pulled in 1000 damage or 1 damage, you earned the same the same amount of points
• Comes along PGI PSR.. they totally fraked that up. Good intentions but what can one expect from a team which only a couple are even competitive enough to move up and able to use the tools provided.. Look at the low MS thresholds on the PSR ladder. Talk about participation awards. No wonder the devs were easy marks for Dragon bowling.

[Hellbringer]

i think if you play FP a lot, you know pretty much who are all the big hitters. I dont really bother searching them up on jarlslist or anything, you just pretty much know. Soooooo yah

[Negat1ve Nancy]

PSR would work, if it wasn't inflated because your team won.

[jjm1]

I just look at the colour of their mechs.

There are two tiers: Baby puke green and Probable Threat

[dario03]

jjm1, on 04 May 2020 - 11:54 PM, said:

I just look at the colour of their mechs.

There are two tiers: Baby puke green and Probable Threat

And they all laughed when I said nobody would suspect the newb green mech. Laughed I say, laughed!

[Vellron2005]

I think the ONLY realistic way to measure a player's skill is using match score, BUT only in a specific mech variant, and only calculated from that player's performance in that specific mech withinthe last 100 matches.

Because someone can be the boss of saaay, a Timberwolf, but completely suck in a Commando. Or even a different Timberwolf.

KDR and W/L are not good measurements of skill because they can be manipulated and the said player can get carried. Match score cannot be artificially inflated so easily, but is should still be done on a mech-by-mech basis.

Edited by Vellron2005, 05 May 2020 - 01:37 AM.

[Ignatius Audene]

Elo was terrible once u hit a certain wall. 30+ minutes wait wasn't unusual in solo mm. And that was with far bigger population. Pgi borked something with elo. They will not invest much time in rework of mm at the current state. The best we can hope for is frequent psr reset.

[Black Ivan]

W/L Ratio is one indicator

[Buster Machine 0]

By playing with them.

[Nearly Dead]

I would prefer the average of (damage / mech weight) per match from the last 100 matches. If they want to refine it further add in AMS damage, TAG damage and UAV damage so that people don't stop doing those useful but expensive (UAV in c-bills, TAG, NARC and AMS in weight) things. I assume they would need to create a new measurement at the end of the match from damage and mech weight to track.

If an Atlas pilot does 1000 damage every match he/she is probably a T1 player and gets bragging rights. If a Locust player does 200 damage every match he/she is probably completely unnoticed and doesn't even realize himself how good he is doing.

To me the important thing is to make it a rolling number so people go up and down, but remove the upward bias by ignoring W/L. We all play with players who are on our winning teams but do zero damage. I see two in particular every day or so who seem to be either insanely incompetent or are conscientious objectors who shoot rocks near opponents, but not at them; match, after match, after match (How you get 2 assists but damage =0 I don't get, but I see it pretty often).

And I would like to see a much finer graduated scale, say 24 groups. Then they could say they are going to have 4 skill classes when it is slow and they want to build matches fast. When matches are dropping every few seconds, make it 6 or 8 classes to tighten up the balance. It would just be a number they put in, Average match building time <= X, use 6 skill groups. Average match building time > X, use 4 skill groups. (Numcomplaints > X, use 2 skill groups. ) Just avoid 3 and it works pretty well.

Tremendous advantage, the more players in the queue, the better the matches get because they could dynamically adjust how close the skill grouping is. Bad news is of course that if the player population drops so that wait time goes up, matches would get sucky again. But not as bad as before because at least you would not have high seniority / low skill players in the top group.

[VonBruinwald]

ChenGGez, on 05 May 2020 - 04:25 AM, said:

By playing with them.

You mean playing against.

End of the day there's only one thing that matters: Can I kill them or will they kill me.

[UnkerZ]

JediPanther, on 04 May 2020 - 09:12 AM, said:

For myself if I do just 200 damage and get one kill before my light explodes then I did a normal adverage match. But most lights can do a lot more than that and I always try and out damage if not get more kills than an assault mech. Some times that happens other times not so much.

From watching the recent everyone-in-same-ques testing doing just 200 damage is a challenge for most players not using meta mechs and builds.

For me to win a match on average = min 800 damage output and 3-4 solo kills. No long range builds, no support builds. Only brawl builds because that is the only thing that deals enough damage and has enough armor destroy 3 potato sacks and hopefully have enough to fight the opponent's brawler who is probably into doing the same thing.

Last I rmbered the PSR used in Solaris was at least not too far away <1400 was an easy match, >2000 suicide in general.

Edited by UnkerZ, 06 May 2020 - 08:06 AM.

[Thorqemada]

Honestly they have many metrics like Damage Output, Damage Received, Kills, KMDD, Matchscore, W/L.

In the end you need an Equipment Based MM bcs only then the Player has/feels the freedom to play that sort of Mechs he likes to play and do some adjustments based on playtime (dont suddenly feed the Greenhorns to the Wolves but lead them step by step there) and some mechanic that prevents the Players from playing the same opponents over and over again plus a Loser vs Loser Mechanic if Players have bad streaks that last to long.

Edited by Thorqemada, 06 May 2020 - 05:18 AM.

[W4R GOD]

Nesutizale, on 04 May 2020 - 08:54 AM, said:

I think we can all agree that the PSR isn't doing its job very well. So what else could be used to messure a players skill, idealy useing something that is allready in the game, to create something that can be much easier used to create better matchmakeing.

The top 3 things that come to my mind would be K/D, W/L and average matchscore.

While K/D and W/L are used often in other games I don't find them very usefull in MWO or lets say they lack some finder details.

K/D and W/L will middle out for each players skill over time but are representing just one aspect of the whole game.
It either shows that you are good at killing stuff or that you have been good with a team or just very, very lucky that your team dragged you along.

I think the average matchscore is more detailed as it also takes into account the damage you have done and the kills and also was did you do the most damage and was it a solo kill or a group effort.
Also did you take down UAVs and help the team that way, do you support with AMS? All those small details say more about the skill and teamplay ability then just kills or wins.


Based on that I also would change the matchmaker to tighten up who is grouped together depending on your matchscore. The higher your score is the closer the matchmaker searches a fitting enemy for you.

Just using basic numbers to, in principal, show what I mean

Average matchscore -> What range the MM will look for enemys
100 -> 0 - 200
200 -> 150 - 250
300 -> 275 - 325

I think that could result in much better matches with more closely outcomes.


In regard of the group vs solo debate...I think you could do the following

The highest matchscore of the group is taken, for example 250. To that the MM add 50 points.
So for the entire group the result would be 300. Now the matchmaker tries to find a group and solo players in the range of 275 - 325 to be placed in the oposition.

The bonus 50 points would be to compansate for the advantage a group would have so they face stronger oposition in the solo players as they would also be taken from the pool of the 300 range.

If you get invited to groups who are not referred to as trash you are good.

If you are called trash then you are.

That's how you measure skill. None of the metrics, boards mean jack.

[JediPanther]

W4R GOD, on 06 May 2020 - 02:03 PM, said:

If you get invited to groups who are not referred to as trash you are good.

If you are called trash then you are.

That's how you measure skill. None of the metrics, boards mean jack.

Or you get competitive and semi-competitive cw units trying to recruit you for your mech skills...only for pgi to constantly mess up their own game and see your unit slowly die as more and more friends leave for their reasons.

[thievingmagpi]

W4R GOD, on 06 May 2020 - 02:03 PM, said:

If you get invited to groups who are not referred to as trash you are good.

If you are called trash then you are.

That's how you measure skill. None of the metrics, boards mean jack.

pretty much this.

if you're good, you and *other good* players know it. that's it. trying to split hairs over stats and niggling over rankings means you're not good. also, as we've seen, just because people have heard of you and *claim* you are good is meaningless as they're often wrong. the amount of "oh you should listen to (insert 30 percentile here) they know what they're doing" is quite common.

[Paul Meyers DEST]

W/L is the only one. Period.
2 Minute Lobby and form a match.

[VonBruinwald]

thievingmagpi, on 06 May 2020 - 08:33 PM, said:

the amount of "oh you should listen to (insert 30 percentile here) they know what they're doing" is quite common.

This. Don't listen to me on the forums, I talk trash and I have a laugh, even the advice I give to players who seek it is tenable at best and devoid of the meta-mentality. The only advise I'd really recommend people listening to is that given in game. They're normally picking up on thing's you're doing, or not, that could actively help you.

I remember one particularly fun match; just me and the last red, dude was dropping IS-lurms on me from within 100m, which was weird. So I put up in the chat that they didn't work within 200m. He stopped lurming, turned around and ran away! When he had sufficient distance he started lurming me again. Reds didn't even think to tell him, and my blues, well, I'm pretty sure I got reported for wasting time and "assisting the enemy"....