Info: The Effect of In-Game Skills on PvP

[Doer]

This is a subject that hasn't received a lot of public scrutiny. My experiences to date lead me to believe that is for the following reasons:

• PKers are generally willing to believe that any success they have is due to their own IRL awesomeness
• PKers are generally of a breed that is too short on patience for conducting boring tests and reading of lengthy discourse

In in words, hyper adolescents and rough-and-tumble brutes are not that interested in divining game mechanics. Fortunately, i am fairly safe in making these statements on account of the the poor chance any of the same will actually read the thread. To be fair, if i was more into PK, my original assumptions regarding this issue would have been better informed.

Anyway, now that i've vented my frustration at dealing with one too many trigger-happy little punks in the process of conducting tests, let's get down to business. This is a thread about effective hit ability in player-versus-player combat, and the role that Evader and Dodger play. Here's a history of what has been tested before, and the hypotheses and observations. Feel free to skip past the quotes for the executive summary.

I have a bit of info that might be interesting for your research.
With imk2 without attachments I have 100% hit rate in pvp (tested on some very high dodge avatars)
With mm on the same avatars I have quite a bit less hit rate without attachments (my base laser pistoleer HA is 8.0), but even with scopes and lasers that would make my laser pistoleer HA greater than my laser sniper I don't get quite 100%
Against mobs I don't notice any difference between imk2 and mm with lasers. There might be one but difference between 100% and slightly below 100% is much easier to see than 91% or just under 91%
So maybe you should add pvp testing to your research as some thing may be easier to see.

Thanks for sharing this mjukis. This shows that there is clearly a different system used to determine the hit rate on avatars versus on mobs. I didn't realize that 100% hit rate was possible in PvP. I surmise you have >= level 100 (Hit) in sniper.

It suggests that the skill mod either 1) works differently in PvP situations, 2) scales non-linearly near 10.0 HA (one of the two cases i mentioned in a previous post) or 3) doesn't increase effective HA.

Ok, curiosity was killing me. I spent some time in the octagon with a mann and m2100. Here's what i've got so far thanks to some help. These are all vs. a player with no Dodger levels.

Maxed SIB weapon (m2100): 100 shots
Hit rate 1.0
No crit hits

< Level 1 Laser Pistoleer (Hit)) with Mann MPH: 5x100 shots
Hit rate 0.818 +- 0.043
No crit hits

Level ~47 Laser Pistoleer (Hit) with Mann MPH: 2x100 shots
Hit rate 0.97 +- 0
1 total crit hit (out of 200)

Level ~47 Laser Pistoleer (Hit) with Mann MPH + 2x SE600 and 1900RS (28% skill mod -- Effective level 60 (Hit): 2x100 shots
Hit rate 0.98 +- 0.02
1 total crit hit (out of 200)

It apparently is indeed possible to get 100% hits in PvP. What's not so certain is the nature of the range from 0 HA to 10 HA. The base HA gives around 0.8 hit rate -- the same as against mobs -- but even at just halfway to maxed i was getting nearly perfect hits. I think it might be a higher order curve than linear, which would mean that skill mod attachments in PvP would be less effective than against mobs (and also mean that low (Hit) skills in PvP aren't as important); however, it is not certain from my tests if skill mod does anything at all in PvP.

Using the asymptotic fn(x) = 1/sqrt(x) gives a result that has hit rate ~0.97 at 4.7 HA (plotted above). This function predicts a hit rate of ~0.9 at HA 0.8 and a hit rate of ~0.95 at HA 3. Some tests by players in the range 0.5-3.0 HA would be very helpful at verifying this.

It may be that the opponent's Dodger level modifies the curve something like this:

The problem is, at my HA this equation gives a difference of just 1% in hitrate between a level 0 Dodger and a level 30 Dodger. This function predicts a hit rate of ~0.9 at HA 0.8 and a hit rate of ~0.95 at HA 3. We need some tests by players in the range 0.5-3.0 HA laser pistoleer vs. a level 0 Dodger and a high level Dodger. Volunteers?

One thing that has been nagging me is that i know i have successfully dodged grenades multiple times. I guess it's possible that the grenader had < 10/10 on the launcher. Less likely is that the hit ability on those works differently. I'll need to ask next time it happens to be sure.

And finally, thanks to Nebu for challenging my assumptions based on the earlier tests:

I dont have tests, but I know from experience if you fire three rockets into a group of people and hit all of them but miss Joda every time, then his dodger level must play a role.
You say a different effective HA in PVP, I agree with that. But that is something different than saying "dodge has no effect in PVP".

In short,

• PvP appears to have a different effective HA than hunting
• It is possible to get 100% hit rate in PvP
• The minimum hit rate for a noob vs a noob is 77-85% (possibly the same as in player vs. mob)
• The hit rate goes up much more quickly than in hunting
• It's not clear that skill mod attachments play a role
• The target's skill does affect the hit rate!

The last point is the big one, here. I didn't test that at all previously. This is an invitation to start exploring the effect of the target's skill level on hit rate in PvP. I welcome results from others to add to the plot.

With the focus here being on skills, rather than skill mod attachments, the tests were done with maxed SIB weapons. Specifically, the M2100 and Powerfist from the TT. I also tested a Mann MPH in similar conditions and found that my miss rate was higher (as would be expected), but beyond that, did not investigate the effect of HA on the hit rate. To repeat: all of the results plotted here are with a maxed SIB weapon. Now, with no further ado, here's what i collected yesterday, with many thanks to Hijacker hijacker27 inferno for his patience and helpfulness.

What you are seeing is that it appears to be a relationship--between the attacker's (Hit) professional standing and the defender's Dodger (or Evader in the case of Powerfist) standing--that determines the hit rate. More tests are needed, but it seems that when the attacker has significantly higher (where "significantly" is possibly 20 levels) Hit than the defender's Dodger, the hit rate will be almost flawless. As the two professions draw closer together, the hit rate declines to at least 85%, which may be the minimum. This makes it possible to reconcile mjukis' experience (getting 100% hits on avatars with high dodge, and then less than 100% with a lower Hit profession) and the previous tests.

It would be very helpful to have players with high Dodger volunteer to stand still for 10 minutes in the ring.

PvPer's version: Skills matter.

Thanks to the following for their helpful participation:
Hijacker hijacker27 inferno
Black Hawk Hawk
Noober to Uber
Lance Blackthorn
jwuan marcus
Micah Hoos McDowell
Runner Blade Runner
Cloud
Thaddeuss
Beatrix

 

[Coop_]

Wow

So in pvp higher hit pro standing helps even when using maxed SIB weapons!

This shatters the idea that skills beyond a certain level are worthless. At least in pvp they can be significant.

+Rep, and I'll see if I can do a bit of testing.

 

[mandu]

As always, good work Doer

And for the record, I consider myself A white Hat PKer Honor amongst thieves !!

I'm glad you posted this, because I had heard once, long ago, that dodger/evader didn't affect PVP.

Back to drone Skilling!!

 

[Doer]

Some further thoughts:

• Still no tests on someone more than 20 levels higher Dodger. Does it level off, or keep going down?
• It's easier to find high Evader, but harder to test melee (no ammo to keep count for you)
• This changes greatly the criteria for choosing a weapon for PvP:
  • Although many good PvP armors are deficient in Melee protection,
  • most avatars have higher ranged (Hit) than melee (Hit)
  • and most avatars have higher Evader than Dodger
  • Plasma is powerful and long range, but who has good plasma skills?
  • Explosives are " " " ", " " " "
• It may be that using a SIB isn't as important in PvP as thought
• But damage is still important

 

[Hurrikane]

I'd say a fair amount of RL factors are important too; speed of connection, speed of PC, responsiveness of mouse, and RL playing skills; I'd say anyone who can charge through Halo or FarCry on hard-as-hell setting would be at a distinct advantage to someone like myself, who still has fun but dies quite a lot

Hurrikane

 

[Kahr]

Interesting. I do hope that skills do have an effect of some kind. I have done a few tests my self that turned out that dodge might not matter in pvp, but as you graceful described it, most pkers dont have much patience, so my tests where very short
If you need a test person with dodge over lvl 20, i would gladly help. I have just chipped out a bit, needed a bit of ammo money, but i am well over lvl 20

 

[Torresta]

what i want to see is "rl skills vs. ingame skills" thread.

 

[Thantos]

I'm sorry, this discourse was too lengthy for me to read.

Keep up the good work.

 

[Coop_]

Tested 100 shots with maxed emik s30 by shooter with prof level 6.6

Dodger Lvl            Hitrate
   0                     92%
  15                     85%
  28.8                   81%
  31.5                   76% (not on graph yet)

There are 25 levels between the shooter and highest target tested, and it seems the hitrate continues to go down.

 

[Doer]

Thanks Coop; interesting to see that the baseline for low (Hit) and 0 Dodger is the same as maxed player vs mob. Maybe there's a unified formula.

Here's the updated plot with some test results from tonight:

The low Hit (=11) data from from my first plot was based on just one test at each point, so it's probably not very accurate. Our noober had to run...

I added the data from an avatar with Hit=6 to yours on the plot.

 

[Doer]

What, no sine waves? Yawn.

Need more melee vs. Evader...

 

[Xen]

This is a subject that hasn't received a lot of public scrutiny. My experiences to date lead me to believe that is for the following reasons:

• PKers are generally willing to believe that any success they have is due to their own IRL awesomeness
• PKers are generally of a breed that is too short on patience for conducting boring tests and reading of lengthy discourse

In in words, hyper adolescents and rough-and-tumble brutes are not that interested in divining game mechanics.

It would be very helpful to have players with high Dodger volunteer to stand still for 10 minutes in the ring.

Hopefully the people with higher dodger levels who might be of help are not also pk'ers.

 

[wobbledag]

You havn't factored in the proven fact that your HA on other person is increased by x% if you shout "Your mom is .......insert own insult" before you fire

 

[levithanikos]

Great Thread Doer ,

was very enlightening to read !

Demonikos

 

[Doer]

Hopefully the people with higher dodger levels who might be of help are not also pk'ers.

If any of my in-game experiments had ever had significant help from the dedicated PKers i might be concerned about it. As it turns out, the contribution of PKers has overall been negative, with direct interference on a few occasions.

Fortunately, there are usually a few highly-skilled hunters willing to help out. Black Hawk and Hoos were sufficient for this one, as you can see that i already have an expression for the hit rate in PvP with firearms.

It looks like the melee vs. Evader could be different. If anyone wants to submit tests for that, the easiest way is to download Witte's damage parser and then buy a tt powerfist or shortblade and hack at somebody several hundred times. Then, you can get the hit count from the parser and the number of attacks from the decay on the weapon. The parser requires Microsoft's .Net version 2 or 3 installed (from Windows Update) and Entropia.exe to be in W2k Compatibility Mode. You also should be running EU full screen to make it easier (alternatively, line up the window with the bottom left corner of your display).

It may be that the decay for the tt powerfist is wrong on Entropedia, which could be shifting the melee results. Someone tried changing it a couple years ago to make their merchandise look more favorable; maybe it has happened again. Feeling too lazy to check that atm.

You havn't factored in the proven fact that your HA on other person is increased by x% if you shout "Your mom is .......insert own insult" before you fire

Hey, you're right . . . everything fits now! Eureka!

 

[AJack10600]

That's interesting stuff to read... and I am inclined to believe it

Good work !

 

[Coop_]

What, no sine waves? Yawn.

Need more melee vs. Evader...

Your wish is my command.

Whipper lvl 6.4
Evader             Hitrate
 33.7                72%
New tester with whipper lvl 6.6
 20.5                81%
  0.2                88%

 

[Coop_]

Hmmm, just had a thought.

Wasn't most of the old HA testing done in the ring? If there was much difference in prof levels of the testers, it could make a big difference in the data.

 

[Doer]

Hmmm, just had a thought.

Wasn't most of the old HA testing done in the ring? If there was much difference in prof levels of the testers, it could make a big difference in the data.

My PvP HA tests as summarized in the OP were on an avatar with 0 Dodger. Someone may have done some hunting (player vs. mob) HA tests in the ring, but it would have been very inconvenient.

In those tests with a non-SIB weapon (Mann MPH), i got 98% hits at level 48 (Hit) (4 misses out of 200 shots), and 82% +- 4% at level ~0 Hit. And, of course, 100% hits with maxed SIB and level 48 (Hit). I didn't do many repetitions, and a more complete range of tests with non-SIB weapons vs. 0 Dodger would be good. The miss rate is so low that it requires a lot of shots to approach the real value due to "dynamics".

Here's a possible way that HA works for PvP: Multiply the expected hit rate due to Dodger by 0.9+HA*0.01. It gives 81% for 0 HA vs 0 Dodger and 95.2% for me with a Mann MPH vs. 13 Dodger, which coincides nicely with my actual measured value of 95.3 the other day. It doesn't work well for my test from months ago, but the error is very high as i'm sure you are aware.

Strange things are afoot in the ring. What's the error like on those numbers?

 

[Xen]

If any of my in-game experiments had ever had significant help from the dedicated PKers i might be concerned about it. As it turns out, the contribution of PKers has overall been negative, with direct interference on a few occasions.

Fortunately, there are usually a few highly-skilled hunters willing to help out. Black Hawk and Hoos were sufficient for this one, as you can see that i already have an expression for the hit rate in PvP with firearms.

Ok, what dodger levels would be helpful for more data points, or do you not need more dodger helpers?

 

[Doer]

Stop the presses. Er, i mean, Coop.

Coop, i need Dodger for those avatars the whip was tested on.

Behold:

Now that is interesting. Selling Dodge@boxes, 40k%

Question: do you get Dodge gains when taking melee PvP? I don't think so, but i don't remember. This may be the only example in game where the contributing profession is not the same as the one where gains happen.

Ok, what dodger levels would be helpful for more data points, or do you not need more dodger helpers?

Dodger over 50 would be good, but we just need a couple tests on someone really high (70+) to make sure the trend is really linear and that would be enough.

 

[Coop_]

The lowest evader to be whipped would have very low dodger, probably 0.1 at most.

The other two test subjects were Alice and I. Both of us have dodger that is several levels at least lower than evader. I'll get the precise levels later (or maybe Alice will see this and post hers).

 

[Xen]

Dodger over 50 would be good, but we just need a couple tests on someone really high (70+) to make sure the trend is really linear and that would be enough.

Ok, I'm afraid I'm only 39 dodger. If you ever need that, let me know and I'll be glad to be shot repeatedly.

 

[Coop_]

Strange things are afoot in the ring. What's the error like on those numbers?

All my tests were 100 attacks. 90% confidence interval is ~ +/- 5, 95% c.i. is ~ +/- 6

 

[Doer]

The lowest evader to be whipped would have very low dodger, probably 0.1 at most.

The other two test subjects were Alice and I. Both of us have dodger that is several levels at least lower than evader. I'll get the precise levels later (or maybe Alice will see this and post hers).

Hmm, lower is not good. The values are already lower than expected. The other set of melee data shifted right over to the line when i changed the Evader to Dodger. Still ambiguous...

Edit: 100 is probably too few. Could you repeat one or more, please? I sometimes get one out of three with quite a different result. I will be happy to pay for your whip decay. heheh

Ok, I'm afraid I'm only 39 dodger. If you ever need that, let me know and I'll be glad to be shot repeatedly.

It may turn out to be useful. Thanks for the offer and i'll let you know if so.

 

[Coop_]

Hmm, lower is not good. The values are already lower than expected. The other set of melee data shifted right over to the line when i changed the Evader to Dodger. Still ambiguous...

Edit: 100 is probably too few. Could you repeat one or more, please? I sometimes get one out of three with quite a different result. I will be happy to pay for your whip decay. heheh

How many attacks were your pf tests? I think I just took 100 from some of the previous tests mentioned in the OP, but that is a bit low for this.

 

[Doer]

How many attacks were your pf tests? I think I just took 100 from some of the previous tests mentioned in the OP, but that is a bit low for this.

300 or 400. Just started the damage parser and started wailing on the target for several minutes.

With the firearm tests, if the helper will allow, i do three series of 100 shots and then do more if the deviance is high.

 

[mandu]

Did you get any perc gains during the deviant data sets

I'll just go back to watching now

 

[Doer]

Did you get any perc gains during the deviant data sets

I'll just go back to watching now

Well, i am conducting the tests in the ring, sometimes even in Twin Peaks. There're lots of deviants. *rimshot*

 

[Sidney]

I'd say a fair amount of RL factors are important too; speed of connection, speed of PC, responsiveness of mouse, and RL playing skills; I'd say anyone who can charge through Halo or FarCry on hard-as-hell setting would be at a distinct advantage to someone like myself, who still has fun but dies quite a lot

Hurrikane

i can't agree with this.

when u get green to green dot action in pvp then it will start from far away.

(as soon dot's pop up in radar)


when this far away the players only have to select the other person.
and thats not hard to do,
when a person is far away and he runs from left to right and jumping up and down his avatar only moves like a inch for me.
so its really easy to select and the just auto fire.

so skills v skills and no real life skills.

as i play Quake from time to time (and in the early day's a lot) i manage well in there.
aswell in battlefield, and crysis.

here in entropia, is different story.
i get killed almost allways , as when i start i take it up against higher lvl players.
why ,well i think its to easy to only shoot low lvl people.


and yes my connection is fine.
and yes i have a state off the art PC.

this meaning i can download a movie , listen to music , watch a movie , play crysis, play EU window mode , and chat on msn with webcam on.
without having any lag in EU being in twin peaks.
and all this running at the same time offcourse........


so my thoughts are to keep its skill v skill and nothing more.

edit: and the longer range gun offcourse