12 replies to this topic

[Karl Streiger]

One of the things PSR seem to change in comparison with ELO, was to exclude the team performance from the rating.

If you ask me a grave mistake. Because its like the moon and the tide. Even if you don't see it or don't think you consider it - its still there.

I would change a couple of calculations in the Ranking - including a different handling of kill/death ratio as well as a efficiency (accuracy and number of components hit). Adding damage taken too.

Damage efficiency formula is very simple:
damage accuracy * component efficiency * 10

damage accuracy = (damage dealt)² / (shots fired x damage)
component efficency = targets hit / components hit

For example 20 shots with the Clan ER Large Laser - deal 100dmg. 4 targets were hit and on average 4 components per target.
So damage efficiency is 100²/220 * 4/16 * 10 = 110 - rather than 100 as it is currently.

Think for your self - when the same guy would only have fired the ER-Large Laser 10 times - only targeting 3 targets but using more concentration:
80dmg 3 targets average of 2.5 components.
= 80^2/110 * 3 / 7.5 * 10 = 232

While the formula might need some tweaking. In the first example 100 dmg were splashed over 16 components for a total damage of 6.25 per component while in the second 10.66 damage per component was dealt. Something that is more important than raw damage.

---

The knack however, is to compare the individual performance with the average performance of your team, and to compare the performance of your team with the performance of the other team.
When made right - you can identify those that hardly participate as well as those that carried the team. It is not possible with the current PSR

compare:

• damage player /avg damage team
  
• avg damage enemy / avg damage team
  
• accuracy player / avg accuracy team
  
• avg accuracy enemy / avg accuracy team
  
• events player / avg events team (spotting, capturing, NARC Bonus, ECM Counter....)
  
• avg events enemy / avg events team
  
• looses / kills

Of course it is necessary to add some limiters. a 1:12 loose shouldn't modify your individual perfomance with 12 - so the maximum mod might be 3; while as the minimum should be 0.35



Yes it seems like you get a penalty for stomping over the enemy team, were as the losers get positive modification for their PSR.

It is important to understand that the overall individual rating for the losing team isn't big enough, to begin with. If you have an outstanding individual rating and still be part of the losing team you now get a huge buff.

So you decouple PSR from the W/L - so PSR could look like:

• Ranking <0 -----extreme huge drop (team damage - try this formula 0.1*(team damge)²
  
• Ranking 0-30 ----- huge drop
  
• Ranking 30-60 ---- medium drop
  
• Ranking 60-90 ---- small drop
  
• Ranking 90-120 --- stay
  
• ...
  
• ...

Did some calculations - average team perfomance on the winning team put you in the 100-130 bracket, while as loosing and average performance even with the team comparison hardly brings you above 90

Edited by Karl Streiger, 07 November 2016 - 06:34 AM.

[Black Phoebe]

Karl Streiger, on 07 November 2016 - 06:33 AM, said:

damage accuracy * component efficiency * 10

damage accuracy = (damage dealt)² / (shots fired x damage)
component efficency = targets hit / components hit

So fighting off some Lights who are trying to kill your Assaults will drag down your accuracy and cost you ranking. You basically punish people for not shooting on the biggest, fattest and slowest enemy in their range. Lights definitely could use a buff, but i don't think that's the right way.

I can already smell the "locusts stole my ranking and needs to be nerfed and doubled in size" threads.

[Karl Streiger]

Black Phoebe, on 07 November 2016 - 07:16 AM, said:

So fighting off some Lights who are trying to kill your Assaults will drag down your accuracy and cost you ranking. You basically punish people for not shooting on the biggest, fattest and slowest enemy in their range. Lights definitely could use a buff, but i don't think that's the right way.

I can already smell the "locusts stole my ranking and needs to be nerfed and doubled in size" threads.

Hm - indeed thanks for the input

On the other hand - preserving your stats instead shooting lights would cause you to loose the game - two edged

Edited by Karl Streiger, 07 November 2016 - 09:14 AM.

[El Bandito]

Black Phoebe, on 07 November 2016 - 07:16 AM, said:

So fighting off some Lights who are trying to kill your Assaults will drag down your accuracy and cost you ranking. You basically punish people for not shooting on the biggest, fattest and slowest enemy in their range. Lights definitely could use a buff, but i don't think that's the right way.

I can already smell the "locusts stole my ranking and needs to be nerfed and doubled in size" threads.

I always tell my teammates to focus on enemy Assaults and Heavies first, cause many of them waste their alphas on Lights. If this change forces people to focus on the biggest threats first, then that's one less thing I have to worry about.

Edited by El Bandito, 07 November 2016 - 10:20 AM.

[Ghogiel]

El Bandito, on 07 November 2016 - 08:17 AM, said:

I always tell my teammates to focus on Assaults and Heavies first, cause many of them waste their alphas on Lights. If this change forces people to focus on the biggest threats first, then that's one less thing I have to worry about.

But that's why you should to take out the lights if you assaults are under threat.

Allowing the enemy to do just that but to your teams assaults while only commiting minimal resources to accomplish it isn't good.

When your teams biggest and most powerful mechs end up doing F all, they cannot ignore being farmed by a light, even if it's just tieing them up by being a threat they cannot ignore forcing them to not fire at enemy assaults or position in a certain way is a winning tactic.

[El Bandito]

Ghogiel, on 07 November 2016 - 09:59 AM, said:

But that's why you should to take out the lights if you assaults are under threat.

Allowing the enemy to do just that but to your teams assaults while only commiting minimal resources to accomplish it isn't good.

When your teams biggest and most powerful mechs end up doing F all, they cannot ignore being farmed by a light, even if it's just tieing them up by being a threat they cannot ignore forcing them to not fire at enemy assaults or position in a certain way is a winning tactic.

My Assaults would be under bigger threat if the enemy Assaults are staring at them. Enemy Lights can be dealt by our own Lights or Mediums. I'd rather have my team to focus on that KDK-3 than the Spider.

[FireStoat]

Ghogiel, on 07 November 2016 - 09:59 AM, said:

But that's why you should to take out the lights if you assaults are under threat.

When I pilot my huntsman I stay close to 2 Assaults at all times just for this reason. The mech only has moderate mobility but has no trouble keeping up with Assaults. It has 24 tons of clan weapons with very generous omnipod hardpoint choices. So I have zero trouble moving to a side to bracket a target the Assaults are going after, or immediately changing to a Light (or set of Lights) wanting to stir up trouble once the action starts.

Lights are only an issue when people are spread out and acting stupid, or at least that's how it appears to me when we're talking about heavier mechs being the target.

[Ghogiel]

El Bandito, on 07 November 2016 - 10:19 AM, said:

My Assaults would be under bigger threat if the enemy Assaults are staring at them. Enemy Lights can be dealt by our own Lights or Mediums. I'd rather have my team to focus on that KDK-3 than the Spider.

Your lights and meds aren't your team?If that's the case for you then your statement makes sense.

though it still doesn't reconcile the point the guy you responded to was actually making. So you want your lights and meds further relegated to tanking their ranking by spreading dmg by shooting at lights which don't give them near the better rating. This is one instance where I think the selfish nature of pugs would rightly benfit them personally at the expense of entitled assault and heavy players.

[Karl Streiger]

The ultimate goal is the win - I have a more complete calculation - but I need to translate and check it first.
We are talking about not much - spreading 50dmg over all components of a light might reduce only the part that is damage
It doesn't has effect on assist (I would like to see a sum of percentage rather absolut numbers - so 50 of total 200 dmg dealt to a light is a 25% assist)

LRM firework guys that are creating a lightshow but nothing else would be the real loosers of such a system - and again it is not a huge thing
We are talking about 20-30 points between spray and prey and average accuracy - however real snipers might get a huge bonus - but I don't think it would happen all the time

[Karl Streiger]

Ok some of the missing ideas:

first - the usual Solo Kill, KMDD, Kill, Assist even Death Rating have to go.
Now you only have two kill values the Skill and the Fkill (Solo Kill and Fill Kill) - names are terrible I know.

The Skill is granted as soon as you deal >66% damage on a target it doesn't matter if you deliver the kill blow or not.
The Fill Kill is an assist that sums up. It now trackst the amount of damage you dealt to a target (1% or 55%)
A Skill counts twice, because its automatically a Fkill as well

Another change -the damage does not count - the arms when you destroyed the ST, damage by ammo explosions, crit damage. Even if you made an educated guess where the guy stored his ammunition and blew up the zone with LBX shots.

The first part of the new PSR 2 - or let me call it UPR (Universal Pilot Ranking) - is the individual part (Skill Rating) the Prestige Rating (P) - for doing exceptional things and nice for bragging - finally the Team Rating (T) from the OP

The UPR is calculated (S+P)*T

---

S - Rating:

• The Sum of all Fkills * 10 (0.01+0.55+0.76)*10
  
• Damage efficiency rating = damage accuracy * component efficiency * 10
  
  • damage accuracy = (damage dealt)² / (shots fired x damage)
    
  • component efficency = targets hit / components hit
• Damage Taken * 10 / tonnage
  
• Kilometers traveled * 2( )
  
• Distance Jumped * 5 ( )
  
• Events:
  
  • AMS Cover x X (max Y)
    
  • ECM Conver x X (max Y)
    
  • Target Spotted x X (max Y)
    
  • Hit & Run x X (max Y)
    
  • ECM Counter x X (max Y)
    
  • NARC bonus x X (max Y)
    
  • ....
    
  • ....
    
  • ....

---

Prestige Ranking

• Win 100
  
• Skill * 20
  
• One Man Army - FKill > 3 * 1.1
  
• MarksMan - damage accuracy > 50 * 1.1 >70%
  
  • ~70% accuracy (7 Shots out of 10 PPC hit = 70²/100 = 49)
    
  • ~80% MPLAS hit (96dmg 20 Shots = 96^2/120 = 76.8)
• Team Player - Sum of Events > YYY * 1.1

The Prestige Values should be listened in leaderboard :
Could look like this
Mech, Map, Victory, Skill, Bonus, UPR
AS7-DDC, Caustic Valley, 1, 2,OMA,240

---

So how should UPR be used:

• you can add the UPRs - its a clearly positive trend = XP Bar (maybe separate for each mech)
  
• you can create the average of UPR - not very good but it might be a indicator for strange behaviour (TKs, Cheating)
  
• most important the Median of a sorted list of your UPR Ranks (this would be your real value for MM)

[DrxAbstract]

El Bandito, on 07 November 2016 - 08:17 AM, said:

I always tell my teammates to focus on enemy Assaults and Heavies first, cause many of them waste their alphas on Lights. If this change forces people to focus on the biggest threats first, then that's one less thing I have to worry about.

I secretly love you for this... I'll still try to kill you with my ground-hugging spitfire, but gently... And with love.

[Duke Nedo]

IMO we need players to think less about the personal score and more about how they can make their team win the current match. Anything else will make team play even worse than it is today, and it's not good today.

[Karl Streiger]

Duke Nedo, on 08 November 2016 - 04:06 AM, said:

IMO we need players to think less about the personal score and more about how they can make their team win the current match. Anything else will make team play even worse than it is today, and it's not good today.

Exactly - this is the issue I want to address. Try to score as much damage as possible or just to shoot as many targets as possible won't grant you anything. You need to deal concentrated damage to a single target to score points.

winning and losing have currently effect on your ranking numbers (because it is more difficult to get good ranking on the loose) and again win/loose has effect on the direction of your PSR change.
Strictly spoken a good score during a defeat is worth more a good score during a victory.

The third - team ranking part should filter the conditions of win/loose as well as compare your playstyle with those of your team. If you didn't participate on average you will loose points - and damage alone or high number of assists would not be able to change it anymore.

I could live with blowing up PSR and moving back to W/L aka ELO. Because W/L agregates all the necessary "data" you need. The current PSR however ignores team play - I know what Paul said - but contra-indicatory - playing pure Rambo grants me a better ranking.

Edited by Karl Streiger, 08 November 2016 - 05:37 AM.