Accuracy balancing

[Naraka]

9 minutes ago, SwitchFade said:

Obviously. I didn't mention the original poster, original post or any names.

 

Fairly clearly, I quoted someone and directed the comment to their response to feedback and advice.

Not clearly at all.  The person you quoted didn't ask for advice or provided a response that would validate feedback, much less the feedback you provided.

 

11 minutes ago, SwitchFade said:

Further, I asked that specific person to post data we could analyze. Notice, by looking carefully at the conversation progression, that that specific person posted snarky comments FIRST, to which I requested they run a regression analysis and post, since they decided to chime in. THEN, I replied after they posted another snide comment. I concluded with it, as it was evident they were uninterested in feedback. The OP never weighed in on it.

 

I'm sure you realize that though, as you were pointed in making sure you knew that I knew. Which is quite clear I do, if one were to follow the chain of conversation. 😌

 

 

Which convinces me you didn't realize you were responding to a unique individual who was making a jab not at what you responded to the OP with but how you went about it (big words).  Asking such a responder to give you what you want then being perplexed by their lack of cooperation is what convinces me.

[SwitchFade]

1 minute ago, Naraka said:

Not clearly at all.  The person you quoted didn't ask for advice or provided a response that would validate feedback, much less the feedback you provided.

 

Which convinces me you didn't realize you were responding to a unique individual who was making a jab not at what you responded to the OP with but how you went about it (big words).  Asking such a responder to give you what you want then being perplexed by their lack of cooperation is what convinces me.

You can think anything you like. Doesn't change facts or truth. I'm not overly concerned with you miseprception or misinterpretation, that's on you.

 

 

[Naraka]

24 minutes ago, SwitchFade said:

You can think anything you like. Doesn't change facts or truth. I'm not overly concerned with you miseprception or misinterpretation, that's on you.

Sure sure.  The main reason I pointed it out was no one else bothered.  Funnily enough, I suffer the same mistakes at times, replying to a post and stating it as such that it's directed at a particular side of the argument but not clearly affirming it's not at the quoted individual because I didn't realize fully the stance of the quoted individual didn't completely align with the OP or their side of the argument.

 

But so long as no one actually points that out, I don't have to always admit that mistake.

[macskull]

8 hours ago, Greycat said:

Oh, if I didn't mention it... I disagree with removing that 5% - with very few exceptions. Why am I missing fires with extinguishers in Steel Canyon? Why am I missing boxes? I don't *think* we tend to miss objects in Mayhem missions. Those.. should really probably be just hittable.

The issue there is that with the current combat mechanics (other than the PvE/PvP split) it's the power attributes that determine whether a power is autohit, not the target of that power. So the fire extinguisher temp that doesn't do anything except put out those fires? Sure, make it autohit.

[MTeague]

On 5/13/2020 at 8:04 AM, Greycat said:

Oh, if I didn't mention it... I disagree with removing that 5% - with very few exceptions. Why am I missing fires with extinguishers in Steel Canyon? Why am I missing boxes? I don't *think* we tend to miss objects in Mayhem missions. Those.. should really probably be just hittable.

What if that box is actually a ninja hiding inside a cardboard box?  It might be harder to hit than you think.

 

[RikOz]

On 5/12/2020 at 3:14 PM, Ukase said:

Nemesis Staff misses more often than not when in melee compared to being at range. 

I've been feeling for quite some time that ranged attacks in general have some sort of hidden, stacking accuracy debuff when used in melee range. My main is a blaster, and I see all the time that I can be pew-pew-ing enemies on the other side of the room, nailing them nearly every time, but as soon as some of them get all up in my face, I start blasting holes in the ceiling.

 

The 5% miss chance no-matter-what is also kind of comical when there's a massive level difference. I mean, my level 50, T3 incarnate blaster hurls a purple-IO'd fireball at a crowd of level 6 Hellions in Perez Park, they all go down in one shot, except for that one guy who emerges completely unscathed. And half the time it's the specific mob I had targeted. Actually, I'd swear that, half the time, my targeted AoEs hit everything except the specific mob I had targeted.

 

On a general note, I suspect that the accuracy thing in this game just feels out of proportion when fighting enemies that are just a bit above you. That is, there's too big of a power difference between adjacent levels, so that a mere +1 level difference is a relatively large improvement in damage and defense. Specifically, this is when compared to other MMOs (and let's face it, it's likely that CoH is not the one and only MMO that most players have tried). In WoW, for example, you stand a chance when fighting a single, basic mob up to 5 levels higher than you. Sure, you'll do less damage to that +5 mob, and that mob will do more damage to you, and it will require some skill, but it's not an almost hopeless fight. In particular, a WoW +5 mob doesn't have such a massive defense advantage that you simply can't land an attack.

 

And, yeah, it works in the opposite direction as well. If I level up inside a mission and I'm now +1 to the mobs, I suddenly turn into a lawnmower and they're the grass, even after the level-up inspirations wear off. This leads to metagaming the system so that, when I know the next mission is tough (a good example would be the Midnighter mission to rescue Lady Jane, which is often filled with the big, debuffing daemons), I'll deliberately disable XP gain to freeze my character just a hair shy of leveling up while in the current mission, and then turn it back on before entering the next mission, to guarantee that I'll level up on the first or second spawn and give myself a big advantage for the rest of the mission.

[tidge]

19 hours ago, RikOz said:

The 5% miss chance no-matter-what is also kind of comical when there's a massive level difference. I mean, my level 50, T3 incarnate blaster hurls a purple-IO'd fireball at a crowd of level 6 Hellions in Perez Park, they all go down in one shot, except for that one guy who emerges completely unscathed. And half the time it's the specific mob I had targeted. Actually, I'd swear that, half the time, my targeted AoEs hit everything except the specific mob I had targeted.

 

Most annoying to my PBAoE toons: run into a mob of 16(?) enemies and toss a 'final chance to hit'=95% PBAoE. Two minions remain untouched. This means that at least two of the attack rolls was >95% in 16 tries (possibly only 14 tries because of Streakbreaker). Followup with ANOTHER FCtH=95% PBAoE and one of them is still untouched. That means that the ToHit roll was > 95% at least three times in 18 (possibly 15, again because of Streakbreaker) rolls.

[Doc_Scorpion]

On 5/12/2020 at 5:02 AM, Chance Jackson said:

One of the 1st things I do on all my toons is set up the 10 combat attributes I'm gonna monitor

I'm curious as to what you monitor and why...

[Chance Jackson]

5 hours ago, Doc_Scorpion said:

I'm curious as to what you monitor and why...

Depends on the character but in addition to last hit chance, I like to monitor the the def positions on positional toons or types on typed def toons, various unpaired dmg resistances,  hold protection for toons with click protection for mez, dmg bonus & if I have a spot free for it I'll monitor recharge as well 

 

The why is just to monitor various aspects crucial to survival & effectiveness for ex. if my Def &/or Res get floored I'll counter with the appropriate insp, rune of protection or use barrier etc

Edited May 16, 2020 by Chance Jackson

[Sakura Tenshi]

related to this, I feel like an assassin's Strike from Hide with Build Up should be an auto hit in PvE. It's really annoying to do that and end up whiffing and dealing with your strongest power combo (and what's supposed to be your opener with built in mitigation against retaliatory alpha strikes) being on cooldown. You either have to wait for it to recharge or start swinging with the next strongest attack, which in several stalker sets either has no guaranteed Crit (Golden Dragonfly and Headsplitter in Kat and BS' case) or outright does not crit period (energy transfer and concentrated strike)

[agentx5]

On 5/13/2020 at 5:16 AM, Blastit said:

I think the 5% minimum 95% maximum hit chance is good for keeping people humble. Some variance in gameplay is good because you then have to possibly readjust your tactics.

The humility comes from killed, or team-wiped by the enemies from a mistake or being out matches..
Missing shots just because it's a 5% guaranteed miss chance isn't humbling or adding fun variety, it's just kind of annoying.

 

An example of where critical fails can be amusing:

  

On 5/13/2020 at 7:21 AM, AerialAssault said:

You notice the misses on 95% chance more because they're rarer. How many times have you used Assassin's Strike or Total Focus and it has hit? I guarantee it is more times than you've missed!

 

The 5% miss chance is the equivalent of rolling a 1 in D&D, which is always a miss, regardless of whether your attack bonus is +9 or +99. Without that element of failure, the challenge of the game is reduced significantly. Is it annoying? Yes. Is it bad design? No.

Yeah but in D&D you can also roll a natural 20.

A 5% guaranteed critical hit to go with a 5% guaranteed miss... that actually sounds good!

But how would that work with Stalkers & Scrappers though? Additive? Multiplicative?

 

 

39 minutes ago, Sakura Tenshi said:

related to this, I feel like an assassin's Strike from Hide with Build Up should be an auto hit in PvE. It's really annoying to do that and end up whiffing and dealing with your strongest power combo (and what's supposed to be your opener with built in mitigation against retaliatory alpha strikes) being on cooldown. You either have to wait for it to recharge or start swinging with the next strongest attack, which in several stalker sets either has no guaranteed Crit (Golden Dragonfly and Headsplitter in Kat and BS' case) or outright does not crit period (energy transfer and concentrated strike)

 I agree. I used dual-blades and a miss can actually mess up the combo chain.  Thankfully placate can't miss, but yeah Assassin's Strike missing and then the target getting knocked by a teammate or moving away to pursue a teammate... it can be a bit frustrating.  Sometimes I really need that knock up to connect, and I'm already fighting level 54's so I don't need ANOTHER extra 5% miss chance lol

Edited May 16, 2020 by agentx5

[Noyjitat]

Its an issue 24 thing or homecoming. Accuracy isn't an issue on the pre shut down servers. I think the still not finished issue 24 that was in beta before the game shutdown didn't catch this because they were busy testing so many other changes.

[Solarverse]

I will admit, it can be frustrating to see the system message say, "You had a 95% chance to hit, you rolled a 96" three times out of 5 attacks. I scratch my head and go, "How?"

It does feel off, almost like the chance to roll above 95 is a higher chance than a 5% chance. But since I don't have any numbers to back this up other than how it "feels" I have kept my mouth shut and dealt with it.

If anything, I would say the percentiles need to be looked at just to make sure they do not need to be recalibrated. It may be working just fine and this may be all in my head, but holy cow, these NPC's are lucky as hell when it comes to my roles of 96% or higher.

[macskull]

2 hours ago, Solarverse said:

I will admit, it can be frustrating to see the system message say, "You had a 95% chance to hit, you rolled a 96" three times out of 5 attacks. I scratch my head and go, "How?"

It does feel off, almost like the chance to roll above 95 is a higher chance than a 5% chance. But since I don't have any numbers to back this up other than how it "feels" I have kept my mouth shut and dealt with it.

If anything, I would say the percentiles need to be looked at just to make sure they do not need to be recalibrated. It may be working just fine and this may be all in my head, but holy cow, these NPC's are lucky as hell when it comes to my roles of 96% or higher.

There's a thread somewhere around here where the OP's premise is about how the game's RNG seems to be broken and weighs rolls above 95 more heavily, based on parsing combat logs... but two things came from other posters there:

1. The odd results only happened for players and not NPCs (even though they use the same functions)
2. Some powers do not show a hit roll in the combat logs unless they miss, leading to rolls above a 95 being incorrectly over-represented

Based on those observations I'd say it's working fine and it's all in your head. For another data point, a few months back I did some testing in response to a Discord user who thought a certain proc was firing less often than it should. I loaded myself into a test AE mission and let it run overnight. Not only did my testing show the proc was firing at the expected rate, but it also showed that 95.02% of my attacks hit.

[Doc_Scorpion]

31 minutes ago, macskull said:

Based on those observations I'd say it's working fine and it's all in your head.

*nods*  People only remember the ends of the bell curves - the frustrating misses and amazing Hail Mary hits.  The middle, the vast majority, is forgotten as fast as it happens.

[Solarverse]

42 minutes ago, macskull said:

There's a thread somewhere around here where the OP's premise is about how the game's RNG seems to be broken and weighs rolls above 95 more heavily, based on parsing combat logs... but two things came from other posters there:

1. The odd results only happened for players and not NPCs (even though they use the same functions)
2. Some powers do not show a hit roll in the combat logs unless they miss, leading to rolls above a 95 being incorrectly over-represented

Based on those observations I'd say it's working fine and it's all in your head. For another data point, a few months back I did some testing in response to a Discord user who thought a certain proc was firing less often than it should. I loaded myself into a test AE mission and let it run overnight. Not only did my testing show the proc was firing at the expected rate, but it also showed that 95.02% of my attacks hit.

Well, that's good to know. I would rather it be all in my head than it be true, heh. Thanks for touching up on that with me, MacSkull.

[aethereal]

1 hour ago, macskull said:

There's a thread somewhere around here where the OP's premise is about how the game's RNG seems to be broken and weighs rolls above 95 more heavily, based on parsing combat logs... but two things came from other posters there:

1. The odd results only happened for players and not NPCs (even though they use the same functions)
2. Some powers do not show a hit roll in the combat logs unless they miss, leading to rolls above a 95 being incorrectly over-represented

Based on those observations I'd say it's working fine and it's all in your head. For another data point, a few months back I did some testing in response to a Discord user who thought a certain proc was firing less often than it should. I loaded myself into a test AE mission and let it run overnight. Not only did my testing show the proc was firing at the expected rate, but it also showed that 95.02% of my attacks hit.

Just out of curiosity, does that 95.02% of attacks hitting not count streakbreaker attacks?  Because the true rate of hitting should be higher than 95% due to streakbreaker.

 

Basically, to get 100 attack rolls, you should get 105 attacks, since about 5 attacks should miss, triggering streakbreaker next.  So you should have about 100 hits from 105 attacks (95 normal hits, 5 misses, 5 streakbreaker hits), 95.24% of all attacks hit, including streakbreaker.  So 95.02% would be a little low.

[Vanden]

2 hours ago, aethereal said:

Just out of curiosity, does that 95.02% of attacks hitting not count streakbreaker attacks?  Because the true rate of hitting should be higher than 95% due to streakbreaker.

 

Basically, to get 100 attack rolls, you should get 105 attacks, since about 5 attacks should miss, triggering streakbreaker next.  So you should have about 100 hits from 105 attacks (95 normal hits, 5 misses, 5 streakbreaker hits), 95.24% of all attacks hit, including streakbreaker.  So 95.02% would be a little low.

I've never done such testing myself, but people do those tests with scripts that parse the text in the combat log; it'd be easy to write the scripts so they ignore any attacks that are forced to hit by the streakbreaker when tallying the results.

[macskull]

7 hours ago, Vanden said:

I've never done such testing myself, but people do those tests with scripts that parse the text in the combat log; it'd be easy to write the scripts so they ignore any attacks that are forced to hit by the streakbreaker when tallying the results.

The results people were posting in the thread I mentioned were all done using a script that ignored streakbreaker (easy to do since the way it appears in combat logs isn't the same as a normal hit). I ran my results through that same script but the sample size was small so the results weren't super accurate - some of the posters in that thread had months of logs they could dig through but all I had were about 3000 attacks over a six-hour period from the aforementioned proc testing I was doing.

 

9 hours ago, aethereal said:

Just out of curiosity, does that 95.02% of attacks hitting not count streakbreaker attacks?  Because the true rate of hitting should be higher than 95% due to streakbreaker.

 

Basically, to get 100 attack rolls, you should get 105 attacks, since about 5 attacks should miss, triggering streakbreaker next.  So you should have about 100 hits from 105 attacks (95 normal hits, 5 misses, 5 streakbreaker hits), 95.24% of all attacks hit, including streakbreaker.  So 95.02% would be a little low.

The 95.02% was including streakbreaker, though my sample size was only 2732 attacks. I'd need more data to nail down a final number, but the issue with running through months of combat logs is that things start to get skewed. It's easy enough when you're running through logs and only using one power that you know shows hit rolls for hits and misses, but there are a large number of powers that don't show hit rolls in the combat logs unless they miss so it's extremely difficult to get actual numbers unless you go through and ignore every hit roll from those powers.

[Replacement]

The 5%/95% thresholds are old school and outdated.  They were tuned for PnP games where combat is typically resolved within about 15 dice rolls tops.

 

A standard CoH combat sees at least 25 dice rolls per second.  It will just always feel worse by volume, and at the same time less impactful than it does in a tabletop game. 

 

But I have a different beef with Accuracy: it's boring.  It's a tax to itemize against (Enhancements).  If the sky were the limit, I'd buff everyone's base ToHit chance.  Maybe combine accuracy and end mod enhancements, or something.