Player ratings (just a thought experiment)

[blp.3489]

Just as an interesting (to me anyway) thought experiment: I wonder if it would be possible to come up with a rating system that would give a somewhat accurate approximation of a player's combat ability/history based on a few calculations each time a player is killed.

If each player had a rating that went up and down based on the kills they make and the kills they experience, it might be possible to produce a somewhat meaningful rating.

When a player is killed you could add up the ratings of all the players that have damaged that player and form a ratio to the rating of the player that was killed.  You could then add some amount of rating to each individual "killer" based on the ratio of the killing group total rating to the killed player rating and adjusted by the percent of the damage done by that individual killer. 

If a group of highly rated players gank a low rating player their rating is only increased by a small amount.

If you only tagged the killed player for a small percent of the total damage then your rating is only increased by a small amount.

The killed player would have their rating lowered based on the same ratio of total killer ratings to their own rating so being ganked by a large group or by a group with much higher ratings than you would only result in a small loss of rating while a highly rated player killed by a low rated player or group of very low rated players would take a larger hit to their rating.

I'm wondering if such a system could be the basis for some prestige with the best players acquiring the highest ratings.

If it turned out to be reasonably accurate it could be factored into the WR system so that a guild full of skilled players wouldn't be considered the same as a guild of equal size full of not so great players, or even for distribution of individual players without guilds.

[Hesione.9412]

We have this system already. It's called "PvP".

[XenesisII.1540]

Boon spammer guilds, groups, specs, will be on top.

Then you spread them out over 12 servers and..... they still feed off pugs and not each other to keep their ratings.

End of the day doesn't really accomplish much, and not with the WR coming "soon", they can all move as a single guild anyways.

🤷‍♂️

[DeceiverX.8361]

Collecting metrics for player deaths and objective captures and using those to drive links should be a part of the game and is a concept I proposed like eight years ago to help fix nightcapping problem back when playercounts were actually good, rather than the slow and older methods they used, and is also less exploitable than 1U1D and can force more balanced matchups based on playstyles.

Making that data public to the players, however, is terrible for a number of reasons:

- It'll be gamified and generally just encourage only fighting with vastly superior numbers, because of how hard it is to die and lose any amount of rating with drastic player disparities.
- It can and will be used in horribly toxic ways, be it from bullying good players who die outnumbered or get unlucky to being exclusive of newcomers like how raid people required gear links for a long time.
- It discourages taking risks, and rewards players for just running away from combat rather than trying to out-fight a counter matchup or limit testing against multiple other skilled opponents.
- By extension, it massively buffs already-overpowered sustain builds because not dying simply won't reduce their score.
- It can't really be balanced between stuff like 1v1 and 50v50 in the same mode/map.  If a DPS player in a 50v50 is killed by a gank group on their way from respawning, the gank group deserves no credit even if the ZvZ DPS player just emerged from just having a very high damage score and K:D ratio.

Edited September 21, 2023 by DeceiverX.8361

[blp.3489]

One could come up with formulas with factors and offsets so that, for example, if the ratio of attacker ratings to the killed player's rating is over some amount the attackers don't get any rating boost and the dead player doesn't lose any rating.

Fighting with vastly superior numbers means that any gain you might get will be small, or as above reduced to nil.

If you have one factor F1 that is used to calculate gain for attackers and different factor F2 that is used to determine rating loss when killed, you can adjust those two factors any way that you want to favor aggressive or defensive behavior.  Just for illustration, if you set F2 to 0 then you never lose points for dying so you might as well attack.  That would mean that rating are always increasing which would devolve into amount of play being a bigger factor than skill so you probably wouldn't want to go that far.  You could also adjust the formulas so that a group with a high enough rating, due to size or skill, that kills a player with a low enough rating actually loses rating.  It's just a matter of adjusting your formulas.

[Hesione.9412]

There is no point to this. Also, what you propose is a system that will put even more of a load on Anet's servers than there is now. WvW is a group activity, not a single-player activity. If you want to see how good you are against other players, there is PvP.

[RisingDawn.5796]

Even being able to track kill stats, most of the time the top 5 are clouders or pug farmers.

I think I've managed 7-8k kills one week following multiple open tags, so any other 'stat' would be pointless in this game mode anyway.

Edited September 24, 2023 by CrimsonOneThree.5682
Stop following me to spam confused emojis

[blp.3489]

8 hours ago, CrimsonOneThree.5682 said:

Even being able to track kill stats, most of the time the top 5 are clouders.

I think I've managed 7-8k kills one week following multiple open tags, so any other 'stat' would be pointless in this game mode anyway.

That is where my proposal differs from the way most things work in GW2, the rating gain is proportional to the percent of the total damage you did.  If a zerg of 100 people all do equal damage on a player they kill they all get 1% of the rating increase.  In most cases in GW2 if 100 people participate in a kill they all get 100% of the reward and the total reward distributed is 100 times the reward that would be distributed if a single player did the killing.  That's great for community building and prevents people resenting other people joining in a kill, the downside is that people "tag" things being killed, often deliberately doing minimal damage so that lots of players can tag it and lots of people get the full reward, and the most rewarding way to play is to run around with a large zerg.  That is I think a big factor in the deterioration of wvw into mindless zerging.

[joneirikb.7506]

21 minutes ago, blp.3489 said:

That is where my proposal differs from the way most things work in GW2, the rating gain is proportional to the percent of the total damage you did.  If a zerg of 100 people all do equal damage on a player they kill they all get 1% of the rating increase.  In most cases in GW2 if 100 people participate in a kill they all get 100% of the reward and the total reward distributed is 100 times the reward that would be distributed if a single player did the killing.  That's great for community building and prevents people resenting other people joining in a kill, the downside is that people "tag" things being killed, often deliberately doing minimal damage so that lots of players can tag it and lots of people get the full reward, and the most rewarding way to play is to run around with a large zerg.  That is I think a big factor in the deterioration of wvw into mindless zerging.

It might be different for a rating system like this, at least if it is invisible (as DeceiverX said above).

If it's public info, it would probably hit the "ANet Care-Bear Rule". Then one that they always want you to be happy to see another player, and thus never penalise you for having another player nearby. Which is naturally the main cause why we'll never see a change away from the Zerg-style, it's part of why both OW-PVE and WvW will always be casual formats (among other things).

So yeah, if they added a system like this, it would have to be invisible to the player. That said, it would probably have just been an adoption of the PVP ranking system. And I don't see a particular purpose/need for this myself, considering how casual WvW is.

[blp.3489]

13 hours ago, DeceiverX.8361 said:

- It'll be gamified and generally just encourage only fighting with vastly superior numbers, because of how hard it is to die and lose any amount of rating with drastic player disparities.

First of all, thank you for the thoughtful and detailed feedback, I appreciate it.

Due to the rating gain being divided among all the participating killers, fighting with vastly superior numbers will result in extremely small gain, or even zero gain depending on the parameters of the formula.

 

13 hours ago, DeceiverX.8361 said:

- It can and will be used in horribly toxic ways, be it from bullying good players who die outnumbered or get unlucky

With the right parameters I think you can eliminate or severely limit losses from uneven encounters, that's the whole point of summing the ratings of all the attackers for the ratio.

I didn't state it but I've been thinking in terms of the rating developing over long periods where it would probably take weeks or months of playing for your rating to settle at a level that reflects your fight history.  So each encounter would only adjust your rating by small increments and an occasional unlucky encounter would have minimal effect on your rating.  If you only participate in zerging and never die then your rating would very very slowly build.  If you built in minimum thresholds based on your current rating then there would be a rating level at which the tiny amounts of rating gained from tagging kills while in a large zerg would fall below the threshold and you would stop progressing your rating until you started taking on less lopsided battles or at least started doing a significant portion of the damage that kills the opponent.  Likewise if you have achieved a high rating then even solo killing of newbies would fall below the threshold and you would stop progressing until you started beating higher rated players.

14 hours ago, DeceiverX.8361 said:

... to being exclusive of newcomers like how raid people required gear links for a long time.

That would be difficult to address.  On the other hand, unlike raids, you don't have to join a party/squad to participate in wvw and if the rating was somewhat accurate then at least the filtering would be based on something better than what gear you have.

14 hours ago, DeceiverX.8361 said:

- It discourages taking risks, and rewards players for just running away from combat rather than trying to out-fight a counter matchup or limit testing against multiple other skilled opponents.

Not necessarily, it depends on the parameters you use, if you use a parameter that results in very small rating losses for losing then there is little to no motivation to run as the only way to raise your rating is to score kills.  And fighting against skilled players, who are likely to have high ratings, will allow you to increase your rating faster and to higher levels than fighting players with less skill and lower ratings.

14 hours ago, DeceiverX.8361 said:

- By extension, it massively buffs already-overpowered sustain builds because not dying simply won't reduce their score.

Using a sustain build might limit your losses from defeats but, in theory at least, a sustain build will not achieve as much damage to other players and hence reduce your ability to raise you rating.  But, no, these ratings wouldn't solve the current problems with sustain builds.  It also doesn't address non-damage forms of play such as healing, CC, cleansing etc.

 

14 hours ago, DeceiverX.8361 said:

- It can't really be balanced between stuff like 1v1 and 50v50 in the same mode/map.  If a DPS player in a 50v50 is killed by a gank group on their way from respawning, the gank group deserves no credit even if the ZvZ DPS player just emerged from just having a very high damage score and K:D ratio.

This goes back to the (previously unstated) long term nature of ratings, the rating changes for a single encounter won't significantly change your rating so the timing of when you get ganked won't have much effect.

[Stand The Wall.6987]

well i think you would have to have a cut off at around 20-30% total hp where no one who does dmg less than that amount of health to an enemy player would get rating or else everyone who does zerg combat would automatically have the highest rating regardless of their skill. that would obviously leave out support players so something set up for them would be necessary. you know what the rating could be guild based. that would solve the problem of support players and there would already be a guild rating anyways so theres no sense in having 2 numbers. solos would be left out but WR isn't for them anyways so who cares :)

[ArchonWing.9480]

That rating would just be a running away rating. And that is the last thing this game mode needs.

Your ability to avoid death in most WvW scenarios is your build's capability to disengage as well as how tanky it is.

And your ability to damage many players also depends on mobility as well.

And of course, victory and defeat in larger scale fights involves personal skill the least. The people that contribute to victory the most are not always going to be the one to survive.

For example, I could be on my LB ranger, quad signet Mesmer, or Willbender and just burst people on the tail then coming in for tags when my side is winning while running away while my side is losing.  Basically never die while scoring kills, while your scourges and support mesmers are sacrificing personal glory and survivability and doing much more to the team but have  a significantly worse kdr/damage.

 

Honestly, I'd probably argue the game already sufficiently incentivizes selfish play.

Edited September 21, 2023 by ArchonWing.9480

[blp.3489]

3 hours ago, ArchonWing.9480 said:

For example, I could be on my LB ranger, quad signet Mesmer, or Willbender and just burst people on the tail then coming in for tags when my side is winning while running away while my side is losing.

Isn't bursting people on the tail considered to be making a contribution?  And if those players are on the tail because they are newbies and don't know any better then they won't have high ratings and you won't get much rating benefit from killing them unless you are also a newbie with a low rating.  If they are a skilled player with a high rating than I think you deserve any rating increase you get from killing them.  As for jumping in for tags when your side is winning, small damage tags won't get you significant rating increases because those are scaled by the damage you did divided by the total damage done by all attackers.

Let's consider a formula where the factor for scaling rating decrease when killed is set to zero so you don't lose rating for being killed.  If the formula for calculating rating increase awarded when you kill another player has a threshold that scales with your rating then as your rating gets higher the kills that you participate in don't increase your rating when your contribution was a small part of the total damage or when the victim has a low rating so your own rating will only increase when you kill players of a similar or greater rating.  Basically the higher rating you have the harder it is to increase your rating and you are incentivized to fight players of similar high ratings.

The problem with not having a loss on being killed is that it makes it easier to kill trade.  You would probably need to not award rating for killing the same player repeatedly, maybe temporarily set a player's rating to zero until they waypoint so that they would have to at least repeatedly cross the map for the next trade.

[ArchonWing.9480]

1 hour ago, blp.3489 said:

Isn't bursting people on the tail considered to be making a contribution?  And if those players are on the tail because they are newbies and don't know any better then they won't have high ratings and you won't get much rating benefit from killing them unless you are also a newbie with a low rating. 

It is. 

But I'm just pointing out it favors a more low risk playstyle. And regardless, the act of being in a zerg is inherently safer and those players that get picked off are probably suffering less deaths than they would outside of a zerg, thus they will always hold some value.

If you possess the damage to take people out, it will probably be pretty easy land damage on a  zerg too. But this is more of an issue of initiative, of being able to choose fights, hence my comment about safety.

The act of running away isn't restricted to individuals too. Entire groups may be more reluctant to take engagements if they think they'd be at a disadvantage.

These are just examples. Some roles are inherently more prone to death than other because they sacrifice more selfish traits to benefit the team.

I know you suggested no penalty for dying but there basically isn't already, yet many groups are already extremely afraid of it. And for whatever reasons I think it'll cause people to gravitate towards these safer options because the game already heavily does.

Kill trading isn't really an issue because people will always find loopholes. A cooldown on any single playier inting too much will mitigate most issues though.

Edited September 21, 2023 by ArchonWing.9480

[Venport.3925]

Honestly I just think theres too many factors to make rank meaningful on an individual level, I could maybe see doing it on the guild/alliance level though for matchmaking purposes but even then it'd be incredibly rough and not actually tell you that much

[pukish.5784]

I have played Lord of the Rings online, that game 's pvp have such rating system, and It encourages coward/dirty/zergy game play. I hate it. 

Rating won't indicate true player skills same like WvW Rank. And I'm sure most rating lovers want easy killing situation for their precious.😶‍🌫️

Edited September 22, 2023 by pukish.5784

[kamikharzeeh.8016]

after 10 years a system like this is just too late to make any further sense, and i doubt u can seperate wvw deaths from pve ones, so the whole statistics would be bonkers if they'd use old data anyways

hence yeah no, kind of pointless

[TheIceman.1039]

On 9/21/2023 at 4:26 AM, Hesione.9412 said:

We have this system already. It's called "PvP".

I agree. If u seek top pvp players look pvp. No need bring elitism on wvw. 

Edited September 24, 2023 by TheIceman.1039

[Zephyrus.9680]

Bad idea in practice that would be inaccurate, achieve nothing and maybe make people even more scared to fight. 

Just as an example of how meaningless it would be, last week I had an 18 minute fight with ~5,138,000 damage near the EBG north keep entrance. They just kept coming back. I didn't go down at all and maybe took 100-150k damage in return or around 30-40 times less.

How do you rate that? You don't. 

I was just doing normal weaver things from the backline, protected by the real heroes, competent support, frontline and midline. And facing a less skilled/numbered server that still likes to fight. If you want stats then use arcdps, keeping in mind it doesn't really mean much in an asymmetrical context like WvW. Otherwise there's WvW rank which is all there should be.

[KrHome.1920]

I like the idea. I would tank my rating to zero then bait some cheese build roamers, destroy them and whisp them how bad they are losing to a noob like me. If they then team up to gank me they don't get any benefit since my rating is low.

[Gotejjeken.1267]

WvW is too volatile in terms of gear, food, utility, etc. to do this.    

It's basically slightly competitive PvE--there is no prestige here.  

On 9/20/2023 at 9:09 PM, blp.3489 said:

If it turned out to be reasonably accurate it could be factored into the WR system so that a guild full of skilled players wouldn't be considered the same as a guild of equal size full of not so great players, or even for distribution of individual players without guilds.

Even devil's advocate, a high-ranking guild of like 5-10 is going to do nothing against a zerg with a rating of 0, because numbers always win in WvW.  You'll also get that same high ranking guild of 5-10 being high ranking because they just comp down PUGs all day--there is no skill there only coordination vs. no coordination.

You'd be better off brainstorming ways to make the sPvP rating system better; for starters, not so reliant on random teammates / lucking into a single or double duo per team / actually matching with people of your rank / etc.

Edited September 24, 2023 by Gotejjeken.1267

[Lincolnbeard.1735]

This would just confirm that zerging while pressing 1 is the epitome of combat ability.

[Junkpile.7439]

I need to play 10 PvP matches to  get some cape and i can say that there isn't any point make any ranking system in sPvP. It's even bigger kitten show than WvW.

Edited September 24, 2023 by Junkpile.7439