Whats The Best Way To Messure A Players Skill?
#1
Posted 04 May 2020 - 08:54 AM
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.
#2
Posted 04 May 2020 - 08:55 AM
Elo ratings that we had, but hidden, before the PSR was implemented, work just fine.
For matchmaking, a single outlier, as a significantly better or worse player than the average, actually can throw matches into disarray harder than you'd expect just from being one of 12 players making up a team, so for evaluating groups and teams versus each other, you might want to add players who are clearly above or below the median with a multiplier >1.
For example, 6 good players together with 6 bad players will always beat 12 average players, even if both teams had equal total ratings put together.
Edited by Gagis, 04 May 2020 - 09:00 AM.
#3
Posted 04 May 2020 - 09:12 AM
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.
#4
Posted 04 May 2020 - 09:21 AM
#5
Posted 04 May 2020 - 09:27 AM
#6
Posted 04 May 2020 - 10:28 AM
The most efficient way to kill a mech is with the least damage, but, the next best thing is to farm as much damage as possible before you explode. So, the more damage, usually the better. Damage is always better than a miss.
KDR is a good one but there are always a few players that try to artificially inflate their KDR by hiding etc etc.
Win/loss is another good but sometimes players win games that contribute nothing or they play with better players in groups.
It's actually really easy to rank players in three categories and then combine them into an overall ranking. Of all of them avg damage is probably the best. Thats why people seem to look at avg matchscore first since its essentially based on damage.
Edited by Kin3ticX, 04 May 2020 - 10:31 AM.
#7
Posted 04 May 2020 - 10:39 AM
The problem there is that we are counting on PGI's formula for AMS, which probably weights things like lance in formation, missiles shot down, etc a little too highly.
Assuming we aren't changing PGI's AMS formula, I would assign players a rank score of WLR*((AMS/population avg AMS)*0.1
So, if population avg AMS is 230, a player with a WLR of 2 and an AMS of 345 would end up with an adjusted rank score of 2.15.
Just my thoughts on a simple system using existing data that should be really easy to code. Should be.
And yes, you'd want to put a multiplier on group rankings if they dropped in solo, not that I'm in favor of said merger.
#8
Posted 04 May 2020 - 10:46 AM
#9
Posted 04 May 2020 - 11:17 AM
Two equally competent players (regardless of skill level) will always have a stat. ratio of 1:1 when matched against each other.
That means a T1 player will have a K:D of 1:1 and a T5 player will also have a 1:1 if the matchmaker is only placing them against same tier players. The reason this number diverges is because players are matched against those outside their Tier, which means the matchmaker isn't doing its job.
-----
So the problem becomes, how does the matchmaker differentiate between a T1 and a T5 player given they have the same stats?
The answer is to subdivide the population into brackets (this was the original intention of the Tier system). You shuffle the players into 5 equally divided Tiers and order them 1-5. The matchmaker then creates matches within each stack.
Now we have a matchmaker creating matches within each Tier, T1 vs. T1 through T5 vs. T5. Players who are over/undertierd are pretty easy to pick up on this as their stats will diverge from the 1:1 ratio. So you pick a time frame, say every month, and what you do is you exchange an equal proportion of over performing players with the same ratio of under performers in the adjacent tier. Reset everybody's stats and start again.
Every month, good players move up and bad players move down. Eventually most players will end up in the right Tier and a few will find themselves up bouncing between them.
#10
Posted 04 May 2020 - 11:21 AM
Average match score, can be viable, the higher the score, the better your overall performance has been. Though, sometimes, good players get ****** teams, and die fast or whatever, and then there are players with great teams, and they get a fair score.
Now to messure a players skill takes alot more than mear numbers. Working as a team, with the team as this IS a TEAM based game. Do you bite the bullet for your team? This can turn into a poor rating, even if the biting the bullet worked in favore of the team. Communication with your team, as in spotting enemies, calling targets, calling positions. Do you play well with others? There are too many variables that say if a player is good or not.
Just because you can pilot a fighter, doesn't mean your a good pilot.
#11
Posted 04 May 2020 - 12:49 PM
There is another thing - there were many vocal advocats against Elo (and it hat objective flaws)!
There is another thing - one of the reasons was High Elo Players long wait times and competitive pressure!
There is another thing - that PGI finally succumbed and developed PSR!
Why had Elo a long wait time (for High Elo Players)?
Bcs there were not many High Elo Players outside the Primetime.
Which also made them combat the same opponents quite often...
Why had it competitive pressure?
Bcs the High Elo Player would be paired with many lower Elo Players forming a Team with quite a Skillgap that the the High Elo Player had to fill - he had to carry his Team, so allways using the best equip and utmost skill he could offer or get wrecked.
Then came the PSR experience bar...
But going back to Elo, or something along that line, will obviously not help the game...it will only bring back old Problems thrown at a way smaller Playerbase this time...
Edited by Thorqemada, 04 May 2020 - 01:35 PM.
#12
Posted 04 May 2020 - 01:07 PM
#13
Posted 04 May 2020 - 01:50 PM
Thorqemada, on 04 May 2020 - 12:49 PM, said:
Why had it competitive pressure?
Bcs the High Elo Player would be paired with many lower Elo Players forming a Team with quite a Skillgap that the the High Elo Player had to fill - he had to carry his Team, so allways using the best equip and utmost skill he could offer or get wrecked.
..
yup - but it's still the same situation right now, without ELO; say both teams have 11 pugs and 1 competent person; one of them runs an LBX-SLeipnir and the other one wants to have fun for a change, so he picks his trusty XL_LOL_LRM_Atlas.
not set in stone, but we all see in what direction the scales heavily tip.
the thing though: with ELO, you can at least assure that it's roughly equal "talent per side";
with PSR and the upward-bias, it's as likely to throw 12 pugs who got to T1 by sheer force of matchnumbers against 6Pugs plus 6 "diehard-meta-chitters" .. or whatever you call people that are actually trying to win ..
current MM is very flawed. based on ELO it won't be perfect either - but better at least, in lots of ways.
Edited by Teenage Mutant Ninja Urbie, 04 May 2020 - 01:56 PM.
#14
Posted 04 May 2020 - 02:45 PM
I mean if the number of players would be higher you could just do the World of Warships way and say "f*** it just throw everyone together. The chances of sealclubbing for an indiviual player are so low it dosn't matter"
Also they have a seperation by the ships you can use and those are mostly well balanced against each other. You can't have that with MWOs mechbay system that allows to do all kind of stuff.
#15
Posted 04 May 2020 - 02:51 PM
Nesutizale, on 04 May 2020 - 02:45 PM, said:
You must 1st define such terms. Ole Russ defines seal clubbing as 12-4 or worse. What do you define "very few players" to be? Only when actual numbers are defined can we talk actual numbers.
#17
Posted 04 May 2020 - 03:21 PM
RickySpanish, on 04 May 2020 - 09:27 AM, said:
W/L ratio can almost be farmed now with premades in the queue. So using that a measurement is as useless as PSR.
#18
Posted 04 May 2020 - 03:22 PM
secondly global tier reset to accompany this change.
i believe that is all that is required as a bare minimum. but something that enforces an even 20% of the active population in each tier would certainly be a good idea. then the match maker might then know the difference between a scrub and a pro.
i do like the use of ms as a good means to measure player ability. take the average match score over the last 100 games and use that to sort people out into their tiers. the best 20% get t1, the second best 20% gets t2 and so on.
Edited by LordNothing, 04 May 2020 - 03:39 PM.
#19
Posted 04 May 2020 - 03:45 PM
#20
Posted 04 May 2020 - 03:59 PM
Kiran Yagami, on 04 May 2020 - 03:21 PM, said:
W/L ratio can almost be farmed now with premades in the queue. So using that a measurement is as useless as PSR.
No. Really? You don't say! Quick, somebody promote this man to Captain of the HMS Obvious. I wrote two sentences mate. TWO. Go ahead and read the second one.
1 user(s) are reading this topic
0 members, 1 guests, 0 anonymous users