Psr Update And Changes - Jun 2020
#361
Posted 06 June 2020 - 02:26 PM
0-100 match score = -6 PSR - Reason: Player did poorly and did not contribute to the team.
101-250 match score = -3 PSR - Reason: Player provided mediocre contribution but team lost.
251-400 match score = 0 PSR - Reason: Player did well but team lost anyway.
>400 match socre = +3 PSR - Reason: Player did excellent but team lost anyway.
WIN
0-100 match score = -3 PSR - Reason: Team won but player did not contribute to the win.
101-250 match score = 0 PSR - Reason: Player provided mediocre contribution to win, but was not a major factor in achieving the win.
251-400 match score = +3 PSR - Reason: Player provided impactful contribution to team win.
>400 match score = +6 PSR - Reason: Player was a major factor to team win.
From the PGI example, this seems to be zero-sum. The loss category sums to -6 and the win category sums to +6, which is a total sum of 0. As other people have mentioned, "personal" skill rating needs to be based on more than a win or loss. Each player is 1/24th of the contributing factor to a win or loss. Do we want to encourage better teamwork, yes. Does that mean every player on the losing team was the cause of the loss, no.
You can change the PSR values (say multiples or 2 instead of 3) or add some other calculation into the score ranges (match score and the player's current PSR) but the point is that if you play well, you do well. Win/loss is not the most important factor when you are not solely responsible for the win or loss.
#362
Posted 06 June 2020 - 02:41 PM
AzarathRaven, on 06 June 2020 - 02:26 PM, said:
0-100 match score = -6 PSR - Reason: Player did poorly and did not contribute to the team.
101-250 match score = -3 PSR - Reason: Player provided mediocre contribution but team lost.
251-400 match score = 0 PSR - Reason: Player did well but team lost anyway.
>400 match socre = +3 PSR - Reason: Player did excellent but team lost anyway.
WIN
0-100 match score = -3 PSR - Reason: Team won but player did not contribute to the win.
101-250 match score = 0 PSR - Reason: Player provided mediocre contribution to win, but was not a major factor in achieving the win.
251-400 match score = +3 PSR - Reason: Player provided impactful contribution to team win.
>400 match score = +6 PSR - Reason: Player was a major factor to team win.
From the PGI example, this seems to be zero-sum. The loss category sums to -6 and the win category sums to +6, which is a total sum of 0. As other people have mentioned, "personal" skill rating needs to be based on more than a win or loss. Each player is 1/24th of the contributing factor to a win or loss. Do we want to encourage better teamwork, yes. Does that mean every player on the losing team was the cause of the loss, no.
You can change the PSR values (say multiples or 2 instead of 3) or add some other calculation into the score ranges (match score and the player's current PSR) but the point is that if you play well, you do well. Win/loss is not the most important factor when you are not solely responsible for the win or loss.
All the suggestions like this are decent, nothing terribly wrong with them except that it's not really a change, just a tweak of the current PSR scoring. It also is not zero sum. Zero sum means exactly balanced gainers and losers after a match. With the above scoring and all variations of it, the sum is variable because match scores will vary. You need a system where the score relative to other players is what matters, not to an arbitrary threshold number.
This is zero sum, and also a fair system for averaging out players to the tiers appropriate to their skill:
- Top 8 of all 24 players in the match gain PSR (+5 for the #1 score, +2 for the rest)
- Middle 8 don't change.
- Bottom 8 lose PSR (-5 the last place, -2 for the rest)
Edited by Teknomancer, 06 June 2020 - 02:48 PM.
#363
Posted 06 June 2020 - 02:45 PM
Teknomancer, on 06 June 2020 - 11:41 AM, said:
Top 8 of all 24 players in the match gain PSR (+5 for the #1 score, +2 for the rest)
Middle 8 don't change.
Bottom 8 lose PSR (-5 the last place, -2 for the rest)
You get a true zero sum with equal rise and fall. Because match score does get tweaked for win/loss, on the average players on winning teams will tend to rise, while the losing team will produce more PSR loss. High performers on a loss can still rise and low performers on a win can still fall. Middle performers stay in the middle.
This should give Tier rankings that reflect both individual performance over time, and team contribution over time. It would sort players into balanced tiers for the matchmaker, continually refined over the more games played to keep you dropping with players of comparable skill.
Much better than what they are doing. Still doesn't account for how, especially in stomps, your team can bring up or down your performance. This makes PSR far from an individual statistic. Which, in turn, makes it a poor matchmaking statistic.
Still again, your proposal is FAR better than what they plan to do, which is a tiny bit better than what is in place.
#364
Posted 06 June 2020 - 02:50 PM
#365
Posted 06 June 2020 - 02:54 PM
JoeCold, on 06 June 2020 - 02:45 PM, said:
Still again, your proposal is FAR better than what they plan to do, which is a tiny bit better than what is in place.
Thank you!
You do raise a good point, that a team can raise or lower an individual's score. But that may also just be a necessary factor in a team shooter. Piranha has to walk that line between satisfying team play as well as individual, so an element of both will inevitably be part of the system. Not perfect for either view, but a compromise that at least improves matchmaking. I think over many matches, the stomp factor gets smoothed out with a proper zero sum. The more you play, the less any one stomp affects your tier.
#366
Posted 06 June 2020 - 02:55 PM
ShogunKid, on 06 June 2020 - 08:48 AM, said:
If you don't like the Honda -- get out of the car and walk.
I guess you don't understand. Let me explain this using your own analogy:
You have a crappy car you got from PGI (PGI's current situation with matchmaking).
PGI says you can now trade in your old car (PGI will reset the current PSR rating).
And you can have a Jaguar in place of your old car (PGI will zero sum PSR making matchmaking much better).
You get a Honda instead of the Jaguar PGI said you were gonna get (PGI did not zero sum PSR and did not make match making better).
Do you feel cheated or grateful?
Edited by Mal Nilsum, 06 June 2020 - 03:27 PM.
#367
Posted 06 June 2020 - 02:56 PM
An example:
Grim Plexis, running a light mech, narc/marking targets behind enemy lines.
Lured 2 lights to their death.
Swung back around, pegged an Atlas, lured him alone...narc'd rain of death..dead.
Rinse repeat throughout the match. I had 4 or 5 targets narc'd and was spotting the rest.
We won 12-2.
Several people commented in chat that marking the targets made an immense difference.
My 2 buddies with LRM boats got 500 something match scores and 1200-1600 damage.
I did about 200 damage and received about a 200 match score despite lots of hit/run, spotting, targetting, etc. bonuses.
Polar highlands.
Similar play style, behind lines, marking targets and pegging enemies with narc.
Every time they'd push us I'd run and ping their base and 1/2 their team would turn around.
I'd run off, narc/mark them and rain of missiles to their back.
100 something match score despite working my *** off keeping targets marked while my team wrecked havok.
We won, 12-4 or 5.
Tourmaline Desert.
Ran behind, narc'd targets. Spotted and did the 'chase the squirrel' through their rear lines.
Of course several turned to chase me....run off....rinse repeat throughout the match.
Towards the end only 3 light mechs left on our team and 3 on theirs, including 2 heavies.
We picked them apart and won.
250 match score.
In every case my high damage friends that stood in the back and sniped/LRM'd got 400-600 match scores and 1000+ damage.
Given the changes will make this even worse...why should I support my team?
Am I forced to just spam damage and screw helping my team do better?
I feel if the changes are to make people more matched competitively then there needs to be a rework to how points are rewarded...its VERY easy to farm points/money/tiers with LRM's, so far at least.
#368
Posted 06 June 2020 - 04:40 PM
Kurdain, on 06 June 2020 - 02:56 PM, said:
An example:
Grim Plexis, running a light mech, narc/marking targets behind enemy lines.
Lured 2 lights to their death.
Swung back around, pegged an Atlas, lured him alone...narc'd rain of death..dead.
Rinse repeat throughout the match. I had 4 or 5 targets narc'd and was spotting the rest.
We won 12-2.
Several people commented in chat that marking the targets made an immense difference.
My 2 buddies with LRM boats got 500 something match scores and 1200-1600 damage.
I did about 200 damage and received about a 200 match score despite lots of hit/run, spotting, targetting, etc. bonuses.
Polar highlands.
Similar play style, behind lines, marking targets and pegging enemies with narc.
Every time they'd push us I'd run and ping their base and 1/2 their team would turn around.
I'd run off, narc/mark them and rain of missiles to their back.
100 something match score despite working my *** off keeping targets marked while my team wrecked havok.
We won, 12-4 or 5.
Tourmaline Desert.
Ran behind, narc'd targets. Spotted and did the 'chase the squirrel' through their rear lines.
Of course several turned to chase me....run off....rinse repeat throughout the match.
Towards the end only 3 light mechs left on our team and 3 on theirs, including 2 heavies.
We picked them apart and won.
250 match score.
In every case my high damage friends that stood in the back and sniped/LRM'd got 400-600 match scores and 1000+ damage.
Given the changes will make this even worse...why should I support my team?
Am I forced to just spam damage and screw helping my team do better?
I feel if the changes are to make people more matched competitively then there needs to be a rework to how points are rewarded...its VERY easy to farm points/money/tiers with LRM's, so far at least.
The impression I'm getting from all this is that they really should change it to the suggestions that remove the Win/Lose element from player skill (Win/Loss should be for C-Bills and stuff anyway) like the earlier suggestion:
Teknomancer, on 06 June 2020 - 02:41 PM, said:
- Top 8 of all 24 players in the match gain PSR (+5 for the #1 score, +2 for the rest)
- Middle 8 don't change.
- Bottom 8 lose PSR (-5 the last place, -2 for the rest)
... or, barring that, increase the match score value for non-damage actions.
Edited by Wraith of Shadow, 06 June 2020 - 07:57 PM.
#369
Posted 06 June 2020 - 06:18 PM
#370
Posted 06 June 2020 - 06:27 PM
This combined queue has not treated me well. I've losing and barely able to do anything.
#371
Posted 06 June 2020 - 06:40 PM
Paul Meyers DEST, on 04 June 2020 - 10:40 AM, said:
Exactly. With public drops you should honestly have a 50/50 win rate if the system works so basing it on won/loss vs actual performance is nonsense.
#372
Posted 06 June 2020 - 07:10 PM
Who is going to break away and defend the base or do captures when it means their PSR takes a hit? Are most players going to play big mechs with large alpha strikes to maximize damage output?
A low player score doesn't always indicate that someone's slacking. Some people are scouting and doing captures and not encountering other players. Their services are vital and their PSR should reflect that somehow.
MWO may have just jumped the shark.
#373
Posted 06 June 2020 - 07:53 PM
Had almost the entire enemy team Narc'd though and 2 on my team had over 1000 damage.
We won 12-1.
I get good EXP and CREDITS from all the bonuses for spotting/hit-run/etc. but those seem to matter very little to match score.
I have decided I don't care and if helping my team wreck means I am rated as a horrible player then so be it...I'll have fun.
It's both nice and sad to see that leaving for 6 or 7 years and then returning that there the same problems persist but maybe in slightly different forms.
**edit for spelling**
Edited by Kurdain, 06 June 2020 - 07:55 PM.
#374
Posted 06 June 2020 - 08:21 PM
#375
Posted 06 June 2020 - 10:15 PM
Edited by Capt Deadpool, 07 June 2020 - 03:44 AM.
#376
Posted 06 June 2020 - 11:02 PM
Kurdain, on 06 June 2020 - 07:53 PM, said:
Had almost the entire enemy team Narc'd though and 2 on my team had over 1000 damage.
We won 12-1.
I get good EXP and CREDITS from all the bonuses for spotting/hit-run/etc. but those seem to matter very little to match score.
I have decided I don't care and if helping my team wreck means I am rated as a horrible player then so be it...I'll have fun.
It's both nice and sad to see that leaving for 6 or 7 years and then returning that there the same problems persist but maybe in slightly different forms.
**edit for spelling**
While it is true that the strategic contributions of light mechs are underrepresented in the match score, being consistently beneficial to the team, and thus maintaining a positive W/L ratio will compensate somewhat.
#377
Posted 06 June 2020 - 11:47 PM
Anomalocaris, on 04 June 2020 - 10:07 AM, said:
I'd be on board with this if the game adequately rewarded players who are engaging in team-beneficial gameplay other than just racking up damage/component/kill numbers.
I'm sure it's a lost cause at this point, but gaining capture points should be a significant boost to match score. Likewise engaging in secondary match mechanics per the Invasion mode should actually generate worthwhile match score bonuses.
As it stands I could do a bang up job of setting my team up for a victory, bag the win, and still get the lowest match score of anyone but that guy over there who DC'd because the game froze-on-load and he couldn't be bothered to restart and/or reload that match/mech. In fact I actually have done that before, where I'd run a stealth/harasser, keep the enemy off guard, call targets, and finish with a pathetically low score because even though I kept the enemy distracted and looking over their shoulders I only did 200 damage. In point of fact I had one match in a capture Spider where I did 243 damage, got two kills and two assists, and only got 143 match score in Domination. I've had a few Conquest matches in that mech with lower scores despite doing more to secure the victory.
It also doesn't help that game modes are far, far closer to "suggestions" than requirements. If you try to play the mode rather than the meta in Conquest you'll probably get yelled at, at least until one side or the other is half dead. I'm also reasonably certain that the last time I played Invasion neither side bothered with radar or jamming towers, and I've only seen air strike towers used 3 times in the last 2 years.
I didn't exactly love the mode myself, but that sort of thing makes me kinda miss V.I.P. since the defending team was practically forced into it while the offensive team was significantly rewarded for playing the mode and not the (kill everything until it stops moving) meta.
Edited by Marikhen, 06 June 2020 - 11:51 PM.
#378
Posted 07 June 2020 - 12:09 AM
Metalgod69, on 06 June 2020 - 11:00 PM, said:
Ding ding ding... it's the latter.
Edited by Horseman, 07 June 2020 - 12:10 AM.
#380
Posted 07 June 2020 - 01:02 AM
Lets go down to 3 tiers, the average and the left and right outliers.
If you are average you are tier 2 if you are an outlier you are tier 1 or 3.
The only question is where is the threshold to seperate the players but still give acceptable waittimes?
151-299 for average?
For psr, it should be performanced based:
A win allready gives some matchscore. (?)
(Maybe some changes on matchscore numbers must be done for supporting or objective playing.)
So just take the matchscore for calcutaling, the 6 best get +1, the 6 worst -1 and for the rest nothing changes.
But it seems we will, again, get a overcomplicated pgi solution thats not realy solving the problem.
Metalgod69, on 06 June 2020 - 11:00 PM, said:
And drinking sanitizer cures corvid19!
Never heared of the gaussian distribution?
Edited by Kroete, 07 June 2020 - 01:13 AM.
1 user(s) are reading this topic
0 members, 1 guests, 0 anonymous users