Kneeling/offensive while skinning [Archive] - The Gemstone IV Players' Corner

Tgo01

01-28-2014, 02:19 PM

Has it ever been proven that kneeling or being in offensive stance while skinning helped in the success rate of skinning? I always thought it was one of those old GS wives' tales from the early GS3 days and has carried on since.

After a few days of skinning aivrens I've thus far collected thus data:

aivren gizzard success rates:
Offensive|Kneeling: 0 (0%)
Offensive|Not Kneeling: 60 (47.619%)
Not Offensive|Kneeling: 120 (64.516%)
Not Offensive|Not Kneeling: 117 (55.981%)
Total Attempts: 521

So far it would appear kneeling provides a bonus of around 10% to success rate and it's a bit too early to tell about offensive stance.

Someone gather 10,000 of each stat and report back so we can decide once and for all.

Latrinsorm

01-28-2014, 02:35 PM

1000 each should be plenty.

Tgo01

01-28-2014, 02:36 PM

10,000!

BigWorm

01-28-2014, 02:39 PM

Would be interested to see a value comparison of the resulting skins as well.

Jhynnifer

01-28-2014, 03:02 PM

eleventy billion!

Tgo01

01-28-2014, 03:49 PM

Hmm..still ongoing but...

aivren gizzard success rates:
Offensive|Kneeling: 0 (0%)
Offensive|Not Kneeling: 81 (47.647%)
Not Offensive|Kneeling: 120 (64.516%)
Not Offensive|Not Kneeling: 117 (55.981%)
Total Attempts: 565

21 successful skinning attempts later and the success rate is pretty much unchanged for offensive stance.

Is it possible offensive stance actually hinders your ability to skin?

Is this possible, Latrin?

Latrinsorm

01-28-2014, 04:35 PM

Everything is possible in LeBron.

But you have not demonstrated a statistically significant difference. If your percentages stay exactly the same you will do so with 572 samples a piece.

Tgo01

01-28-2014, 04:39 PM

Why is 572 the magic number?

Latrinsorm

01-28-2014, 04:51 PM

You just unleashed the mother fucking fury, pal.

2 * sqrt((81/170) * (1 - 81/170) * X) / X + 2 * sqrt((117/209) * (1 - 117/209) * X) / X < 117/209 - 81/170
Solve for X.

Let Y = 1 / sqrt(X)
.998892 * Y + .99282 * Y < .083338
Y < .041842
1 / sqrt(X) < .041852
X > (1 / .041852) ^2
X > 571
X = 572

/drops mic

Narthsin

01-28-2014, 04:53 PM

Note to self: Never ask Latrin about magic numbers. They are magic.

Tgo01

01-28-2014, 04:55 PM

You just unleashed the mother fucking fury, pal.

2 * sqrt((81/170) * (1 - 81/170) * X) / X + 2 * sqrt((117/209) * (1 - 117/209) * X) / X < 117/209 - 81/170
Solve for X.

Let Y = 1 / sqrt(X)
.998892 * Y + .99282 * Y < .083338
Y < .041842
1 / sqrt(X) < .041852
X > (1 / .041852) ^2
X > 571
X = 572

/drops mic

I can spout a lot of gibberish too but I don't need to.

All I have to say is:

573 > 572.

I win.

Merala

01-28-2014, 05:23 PM

I wonder if the Statistics class I started yesterday will help me figure out what the fuck is going on in this thread. Remind me to check back in May and find out.

Latrinsorm

01-28-2014, 05:43 PM

You'll probably learn all I did in two weeks, I'm really not very well versed in statistics, it's just a little standard deviation business and Gaussian noise.

But tell me more about this camel.

Merala

01-28-2014, 06:08 PM

You'll probably learn all I did in two weeks, I'm really not very well versed in statistics, it's just a little standard deviation business and Gaussian noise.

But tell me more about this camel.

It's tan, has a hump, and can go a hundred miles on a single gallon of water.

SHAFT

01-28-2014, 06:12 PM

I once asked latrin what would happen if you googled google. Once.

Tgo01

01-28-2014, 07:08 PM

Irregardless of Latrin's lack of grammar and his negativity I have proven beyond a shadow of a doubt that offensive stance lowers your skinning success rate by about 7% and kneeling increases your skinning success rate by about 10%.

Behold!

aivren gizzard success rates:
Successes|Fails|Total|Percent chance of success
Offensive|Kneeling: 0 (0%)
Offensive|Not Kneeling: 106|114|220|(48.182%)
Not Offensive|Kneeling: 143|77|220|(65.000%)
Not Offensive|Not Kneeling: 124|96|220|(56.364%)
Total Attempts: 660

But just to be on the safe side I'm going to do about 352 more tests on the ones I've already done and about oh...572 tests on the offensive/kneeling one.

ETA:

I have now proven that offensive stance and kneeling increases your success rate to 100%.

aivren gizzard success rates:
Successes|Fails|Total|Percent chance of success
Offensive|Kneeling: 2||2|(100.000%)
Offensive|Not Kneeling: 107|114|221|(48.416%)
Not Offensive|Kneeling: 143|78|221|(64.706%)
Not Offensive|Not Kneeling: 124|96|220|(56.364%)
Total Attempts: 664

You can't explain that!

Narthsin

01-28-2014, 07:31 PM

Very cool. Now test the other stances too! Guarded stance would be especially good to know, since that's what most casters sit in.

Tgo01

01-28-2014, 07:34 PM

In all tests if I wasn't in offensive stance I was in guarded or perhaps less likely defensive stance.

KneeCapper

01-28-2014, 08:26 PM

Encumberance plays a role as well. Keep your off hand open too, you'll see the results

m444w

01-28-2014, 08:37 PM

Hold your smooth stone in your left hand and lay down for the best skinning results.

Tgo01

01-28-2014, 08:41 PM

Encumberance plays a role as well. Keep your off hand open too, you'll see the results

Hold your smooth stone in your left hand and lay down for the best skinning results.

I can't keep up with all of these rules! :O

Although I could see encumbrance playing a role, my character is never encumbered so that's no problem.

cwolff

01-28-2014, 08:49 PM

I can spout a lot of gibberish too but I don't need to.

All I have to say is:

573 > 572.

I win.

TG for the win. I have no idea what Latrin wrote, but I do understand 573>572!

Dendum

01-28-2014, 09:07 PM

Hands open is a big one in my experience. I see no difference in stance

Tgo01

01-28-2014, 09:39 PM

I figured if stance affects success rate then it makes sense to track success rate based on stance as well instead of just offensive/not offensive because the latter doesn't tell us much other than...y'know...offensive/not offensive.

So I have started my tracking all over again with better and improved tracking! YAY!

aivren gizzard success rates:
Successes|Fails|Total|Percent chance of success
Offensive|Kneeling: 21|10|31|(67.742%)
Offensive|Not Kneeling: 0 (0%)
Advance|Kneeling: 0 (0%)
Advance|Not Kneeling: 0 (0%)
Forward|Kneeling: 0 (0%)
Forward|Not Kneeling: 0 (0%)
Neutral|Kneeling: 0 (0%)
Neutral|Not Kneeling: 0 (0%)
Guarded|Kneeling: 0 (0%)
Guarded|Not Kneeling: 0 (0%)
Defensive|Kneeling: 0 (0%)
Defensive|Not Kneeling: 0 (0%)
Total Attempts: 31

I haven't heard of the open hand thing helping before, that's a new one on me so I didn't account for it in my script but this character always has a shield out when skinning so at least it will be consistent results.

Everyone should download the script skinme and post their results up! Just in case make sure to note whether or not you have an open hand when skinning when posting results. Also probably a good idea to remain wound free, at least with the hand/arm you skin with.

If you already have skinme be sure to download the latest version, run it and then type "skinme reset" to reset your stats so you're properly tracking the new stats.

KneeCapper

01-28-2014, 10:56 PM

Sorry but don't lich. 2x FA and 2x Survival Halfling Ranger and have skinned thousands of war griffs. Off hand open, unencumbered, stance advance and don't try to get decent skins in the dark. I have tried and tried but my quality level always is off in the night for some reason

Goat

01-29-2014, 12:23 PM

I did some research and stance and open hand last year. I posted it on the officials:

http://forums.play.net/forums/GemStone%20IV/Game%20Design%20Discussions/Research%20and%20Investigation/thread/1530785?get_newest=true
(also http://forums.play.net/forums/GemStone%20IV/Game%20Design%20Discussions/Research%20and%20Investigation/thread/1529400?get_newest=true for 702 vs 711)
(also http://forums.play.net/forums/GemStone%20IV/Game%20Design%20Discussions/Research%20and%20Investigation/thread/1509420?get_newest=true, difference between 1x/.5x and 1x/1x)

tl;dr: I saw a small increase in appraised skin quality from kneeling/stancing, but not real impact on skin value. It was all statistically insignificant. I never got enough trials on any of those posts to lock down small effects.

Tgo01

01-29-2014, 04:51 PM

It was all statistically insignificant. I never got enough trials on any of those posts to lock down small effects.

You should download skinme and get more stats! ;)

Someone in Shattered could have this settled for us in a few hours.

Goat

01-29-2014, 05:27 PM

You should download skinme and get more stats! ;)

Someone in Shattered could have this settled for us in a few hours.

I don't skin any more, sadly. All those TPs got dumped into pole arms/CM.

Tgo01

01-29-2014, 05:28 PM

I don't skin any more, sadly. All those TPs got dumped into pole arms/CM.

Bad Goat!

Goat

01-29-2014, 07:31 PM

Happy to send you my skinning scripts if you want. They're in that hacked-up-from-something-else-barely-makes-sense-and-only-works-for-me state, but they include doing a tally of skin quality, which I gather you haven't covered yet.

Tgo01

01-29-2014, 07:34 PM

I'm just focusing on success rate for now. I figure if something helps with success rate then it only makes sense to do that anyways because you'll be skinning more often anyhow. Unless you're already so awesome at skinning that you have a 100% success rate, then you might skip that stuff.

Tgo01

01-30-2014, 01:07 AM

Alright got some updated stats here.

aivren gizzard success rates:
Successes|Fails|Total|Percent chance of success
Offensive|Kneeling: 145|65|210|(69.048%)
Offensive|Not Kneeling: 120|90|210|(57.143%)
Advance|Kneeling: 0 (0%)
Advance|Not Kneeling: 0 (0%)
Forward|Kneeling: 0 (0%)
Forward|Not Kneeling: 0 (0%)
Neutral|Kneeling: 0 (0%)
Neutral|Not Kneeling: 0 (0%)
Guarded|Kneeling: 0 (0%)
Guarded|Not Kneeling: 0 (0%)
Defensive|Kneeling: 0 (0%)
Defensive|Not Kneeling: 0 (0%)
Total Attempts: 420

I think it's safe to say kneeling does indeed affect success rate. I can't say for sure what the rate is but I'm willing to bet it's 10%, possibly 15%.

Next up I'll be doing 210 tests of skinning while kneeling and in guarded stance to test if stance has any effect.

Also if anyone is using the skinme script I updated it so it now keeps separate stats for whether you are wounded or not when skinning. The wounded stats simply checks to see if you have ANY wounds or scars when skinning, seems to be too much trouble to get too detailed in this department. The other set of stats keeps track of skinning attempts with no wounds or scars at all.

You don't need to reset your stats when downloading this version if you don't want to, all of your current stats will be listed under the no wounds stats.

So typing skinme stats will list no wounds stats and skinme wounded stats will list wounded stats.

Goat

01-30-2014, 01:57 AM

The wounded stats simply checks to see if you have ANY wounds or scars when skinning
On a vaguely related note, I never saw any difference in skinning myklians with rank 1 hand scars vs no injuries. Others have echoed that they believe rank 1 hand scars don't matter.

droit

01-30-2014, 01:58 AM

I hope this line of inquiry extends to value comparisons for those of us who never fail a skin anymore. It's very interesting.

Methais

01-30-2014, 09:46 AM

You just unleashed the mother fucking fury, pal.

2 * sqrt((81/170) * (1 - 81/170) * X) / X + 2 * sqrt((117/209) * (1 - 117/209) * X) / X < 117/209 - 81/170
Solve for X.

Let Y = 1 / sqrt(X)
.998892 * Y + .99282 * Y < .083338
Y < .041842
1 / sqrt(X) < .041852
X > (1 / .041852) ^2
X > 571
X = 572

/drops mic

http://25.media.tumblr.com/tumblr_m3rtyerfHZ1qir45xo1_500.gif

Tgo01

01-30-2014, 09:54 PM

Tgo01

01-31-2014, 10:53 PM

Latrinsorm

02-01-2014, 11:09 AM

I'm just saying, you were obviously willing to run 840 tests, why not run 420 of O/kneel and O/notkneel?

Tgo01

02-01-2014, 11:21 AM

I'm just saying, you were obviously willing to run 840 tests, why not run 420 of O/kneel and O/notkneel?

I thought you said 587 was the magic number?

Latrinsorm

02-01-2014, 11:25 AM

You go to war with the army you have.

Tgo01

02-01-2014, 11:27 AM

Ah ha! 572 was the magic number and you didn't correct my 587 figure just now! You're just making this up as you go along. You make up all your math and statistics as you go along :(

Latrinsorm

02-01-2014, 12:37 PM

Do you know many Secretaries of Defense who spend their time correcting privates third class?

In bed?

Tgo01

02-01-2014, 12:42 PM

Latrin, you're no Jack Kennedy.

Tgo01

02-01-2014, 09:16 PM

Tgo01

02-07-2014, 12:02 AM

Astru

02-07-2014, 09:16 AM

It seems odd that not kneeling in offensive would be worse than not kneeling in other stances while kneeling in offensive turns out better, but that seems to be what the numbers are showing (so far anyway). It's definitely interesting if further tests continue to show this pattern.

masterdtwin

02-07-2014, 10:34 AM

You are unfortunately not taking into consideration one incredibly major factor for these tests. Or at the very least you have not clarified your methodology.

Things I am assuming you took into consideration...
Skills / stats remain constant throughout these tests
Weapon used remains the same

Things I don't believe you took into consideration
Potential damage to the skin based on the method used to kill said creature.

Unless you are using an empath using exclusively bone-shatter, or a cleric using fervent reproach, Your individual success rate changes per actual skin - skewing your results into randomness.

JackWhisper

02-07-2014, 10:40 AM

Dude, he shoots everything in the fucking eyeball. Absolutely no chance of messing up there!

masterdtwin

02-07-2014, 10:54 AM

Dude, he shoots everything in the fucking eyeball. Absolutely no chance of messing up there!

Each and every time without fail, for the entirety of this testing?

JackWhisper

02-07-2014, 11:12 AM

You are damn skippy! If he misses, he doesn't count it! He's that badass!

Tgo01

02-07-2014, 11:56 AM

Things I don't believe you took into consideration
Potential damage to the skin based on the method used to kill said creature.

I thought damage to the skin just affected it's value, not the success chance of skinning it. Even if that were the case since I'm using the same spell for every attack I would imagine the damage to the skin would average out over enough tests. So if I'm totally ruining the skin via damage 10% of the time it will be 10% whether I'm in offensive stance kneeling or in guarded stance not kneeling.

Plus what JackWhisper said, I am pretty bad ass.

masterdtwin

02-07-2014, 12:10 PM

I thought damage to the skin just affected it's value, not the success chance of skinning it. Even if that were the case since I'm using the same spell for every attack I would imagine the damage to the skin would average out over enough tests. So if I'm totally ruining the skin via damage 10% of the time it will be 10% whether I'm in offensive stance kneeling or in guarded stance not kneeling.

Plus what JackWhisper said, I am pretty bad ass.

That may be the case, but if you have an unknown variable affecting your sample - Then the sample size must be increase to be statistically significant. That does mean in fact... that 572 is no longer the magic number, instead it would be 715. GO GO GO!

Tgo01

02-07-2014, 12:14 PM

You and Latrin are just trying to kill me :O

masterdtwin

02-07-2014, 12:19 PM

You and Latrin are just trying to kill me :O

Actually, if I was trying to kill you - I would give you the actual magic numbers...

In order to have 99% confidence that your results are within a 0.5% range (1% total margin), you would actually need a sample size of 66564.
In order to have 95% confidence that your results are within a 0.5% range (1% total margin), you would actually need a sample size of 38416.

That my friend, is true statistics.

Tgo01

02-07-2014, 12:26 PM

Actually, if I was trying to kill you - I would give you the actual magic numbers...

In order to have 99% confidence that your results are within a 0.5% range (1% total margin), you would actually need a sample size of 66564.
In order to have 95% confidence that your results are within a 0.5% range (1% total margin), you would actually need a sample size of 38416.

That my friend, is true statistics.

Oh snap! You going to take that, Latrin?

masterdtwin

02-07-2014, 12:31 PM

Oh... Based on your sample size of 210... there is a:

95% chance your results have a 6.76% margin of error
99% chance your results have an 8.9% margin of error

Latrinsorm

02-07-2014, 04:28 PM

Oh snap! You going to take that, Latrin?Am I going to take someone agreeing with me? I'm Republican, but I'm not that Republican.

Here is what I see when I look at your results:

kneeling
O: 73.3% - 64.1%
F: 69.1% - 55.8%
G: 71.3% - 61.8%
no statistically significant differences

not kneeling
O: 60.4% - 50.4%
F: 66.6% - 53.2%
G: 65.6% - 54.3%
no statistically significant differences

We can be reasonably confident that the offensive values for kneeling and not kneeling are different. It could be that the kneeling bonus is stance dependent, but if it isn't then we can instead collect all values for kneeling and see:

677 of 1017: 66.6% ± 3.0%
528 of 911: 58.0% ± 3.3%

It is a fact that God designed the universe to include round numbers, therefore the bonus for kneeling is 10%.

masterdtwin

02-07-2014, 10:13 PM

Tgo...

I'm sorry to disappoint you, but according to my statistics there is in fact no bonus to success from kneeling.

Skinme non wounded stats
moor eagle talon success rates:
Successes|Fails|Total|Percent chance of success
Offensive|Kneeling: 31||31|(100.000%)
Offensive|Not Kneeling: 137||137|(100.000%)
Advance|Kneeling: 0 (0%)
Advance|Not Kneeling: 0 (0%)
Forward|Kneeling: 0 (0%)
Forward|Not Kneeling: 0 (0%)
Neutral|Kneeling: 0 (0%)
Neutral|Not Kneeling: 0 (0%)
Guarded|Kneeling: 0 (0%)
Guarded|Not Kneeling: 0 (0%)
Defensive|Kneeling: 0 (0%)
Defensive|Not Kneeling: 0 (0%)
Total Attempts: 168

moor hound paw success rates:
Successes|Fails|Total|Percent chance of success
Offensive|Kneeling: 37||37|(100.000%)
Offensive|Not Kneeling: 154||154|(100.000%)
Advance|Kneeling: 0 (0%)
Advance|Not Kneeling: 0 (0%)
Forward|Kneeling: 0 (0%)
Forward|Not Kneeling: 0 (0%)
Neutral|Kneeling: 0 (0%)
Neutral|Not Kneeling: 0 (0%)
Guarded|Kneeling: 0 (0%)
Guarded|Not Kneeling: 0 (0%)
Defensive|Kneeling: 0 (0%)
Defensive|Not Kneeling: 0 (0%)
Total Attempts: 191

I'm not sure what crazy stats you're seeing, but it's pretty clear to me, no advantage to kneeling.

Tgo01

02-07-2014, 10:15 PM

I'll get you for this, Captain Planet!

masterdtwin

02-07-2014, 10:30 PM

I'll get you for this, Captain Planet!

To be fair, I was actually trying to help... I didn't actually realize my skin rate was 100%, but that's actually comforting to know :)

Edit: I just realized I'm the idiot who used 'actually' three times in a row...

Tgo01

02-08-2014, 11:33 PM

Latrinsorm

02-09-2014, 11:55 AM

How many more tests do I need to run to be certain? 100,000?There are no certainties in statistics. If we assume your true rates of success are kneeling = 67% | not kneeling = 57%, and we want each value to be ± 1%, then to get to 2 standard deviations we need to do...

2 * sqrt (67% * 33% * x) / x = 1%
1 / x = (1% / 2) ^ 2 / (67% * 33%)
x = 8844

2 * sqrt (57% * 43% * y) / y = 1%
y = 9804

This assumes that stance has no overall effect and that kneeling's effect is stance-independent. With all that said, we do not yet have statistically significant differences for guarded [kneeling or not] or forward [kneeling or not]. The only ones we have are...
o k > o ~k
o k > f ~k
d k > o ~k
d k > f ~k
1000 each would give ±s of about 3% for each stance/kneel set and, if it turns out that stance is still pointless, about 1.8% for kneel vs. not kneel.

1000 each should be plenty.

Tgo01

02-09-2014, 11:59 AM

1000 each should be plenty.

It's almost like someone said this exact same thing in the very beginning of this thread.

Tgo01

02-11-2014, 12:07 AM

masterdtwin

02-11-2014, 08:12 AM

Booya!

So, what you're saying is... You still don't have enough data points, Latrin was right, the kneeling bonus is probably 10%, and for the most part it looks like stance doesn't have any effect on success.

Tgo01

02-12-2014, 01:53 AM

Latrin was right

For the love of God man, speak English! I don't understand gibberish.

Tgo01

02-13-2014, 07:06 PM

Latrinsorm

02-13-2014, 07:55 PM

This is what I've been trying to tell Latrin all along. You need a larger sample size to really get a feel for averages and numbers and whatnot. The larger your sample size the more accurate your numbers are.

Now it's looking more like stance has no effect on skinning at all because kneeling in guarded stance and kneeling in offensive stance are pretty even.When you're right, you're right.

Tgo01

02-14-2014, 07:39 PM

Tgo01

02-16-2014, 08:53 PM

Buckwheet

02-17-2014, 09:51 AM

Does the roll you make for the attempt impact the final quality of the skin?

Tgo01

02-17-2014, 03:00 PM

Does the roll you make for the attempt impact the final quality of the skin?

That's a good question. I'm assuming yes but of course that's just a guess on my part.

One of these days I'll run tests that look at the quality of the skin.

Latrinsorm

02-17-2014, 05:51 PM

Nice job, Terry. Can I call you Terry? Terry, that was a nice job.

Tgo01

02-17-2014, 05:53 PM

Nice job, Terry. Can I call you Terry? Terry, that was a nice job.

What? You're not going to insist I do 1000 tests? :O

Latrinsorm

02-17-2014, 06:04 PM

I said nice, it's your choice if you don't want to do a great job. :)

Tgo01

02-17-2014, 06:06 PM

I said nice, it's your choice if you don't want to do a great job. :)

:(