TH testing: final results

[Enedin]

As the title suggest, these are my last TH findings. I never completed this monstrous project, but I concluded some things from it. W/e you take from it, just know that this project had me running all over the world killing mobs, trying drops and remembering pops.

It's been fun, if FFXIV has a similiar ability, I might test that as well.

I'd like to say thanks to Shamaya for the mutual interest in each other's results, and also the support I got.

http://enedin.be/RandomPics/FINALTHTEST.png

EDIT: forgot to say that all tests were done solo, and no TH enhancements (TK or AF2 hands) were used.

[Shamaya]

First!

[Rocl]

So what you're saying is TH4 is a 200% increase on drop rate? That makes some drop rates on some items very

[Enedin]

So what you're saying is TH4 is a 200% increase on drop rate? That makes some drop rates on some items very

I only tested TH1 and TH2, TK or AF2 hands didn't come into play ever.

All tests were done solo, as well.

But yeah, on some mobs having TH2 (and thus also TH1) can make a 20% drop rate go up to 50%.

[Cybaster]

Thanks for the testing!

It's very curious how some mob can give you 100% increase on TH1 instead of 50%. From your tests it looked like all mob family smaller than the size of taru will get 100% bonus lol.

[Amok]

I'm not sure I'm understanding the data. Why is it TH1 is sometimes a 100% increase and other times 50%? And this was expected?

[drwaffles]

Props for testing, although I have the say the chart made my brain hurt slightly. What's most interesting is that TH1 and TH2 stack rather than the highest one essentially overwriting, which is kind of what had been assumed thus far... does this mean people are going to start bringing something/THF along to events in addition to a THF main?!

[Amok]

I think you're reading it wrong. The TH2-only is theoretical, from what he said.

[drwaffles]

Oh right, the layout of the chart was confusing me since he seemed to have data gathered for TH2 only even though there's a disclaimer about it at the top... Like I said, the chart hurts my brain.

[Dominionix]

There are so many variables we don't understand about TH, I hope someone can continue from this and perhaps answer some of them. Well done on what you achieved though. The idea of TH stacking raises a whole load of questions: Does it only stack if the levels are different? Would multiple TH of the same level stack? (Alliance of THF's anyone? lol) Does the THF / THF's party / THF's alliance having the most hate / killshot affect TH's proc rate? Does the THF being first person to gain hate on the mob... Bla bla bla.

Thanks for the hard work.

[Dominionix]

Deleted.

[Miraun]

Was just doing some theoretical maths on TH1 to try to figure out how it may function based upon the information provided above... and going back to the initial thoughts of potentially rolling twice if you don't get the drop, it seems to moderately fit for TH1...

Knowing that SE seems to work off of loot tables, which are more easilly defined in BCNM drops, this concept could be applied to normal monsters, but with more information that we're not aware of.

Some monsters that only drop a limit of 1 item, like an orcish axe, could actually have us rolling twice, and if either of them are successful, you get the drop... while a terrible way to code, it could curb the effects of TH if TH adds an additional roll to the item...

Continuing on with the orcish axe drop rate, lets go with an observed drop rate of 36%, it could be comprised of two individual rolls of 20%, and if the result is "Yes" in either case, you obtain the drop. The implications of an additional roll in TH would be an increase from 36% to 59.04% in the case of a single roll, while if it was already comprised of two rolls with a single Yes/no outcome, it would increase the observed 36% drop rate to 48.88%



I haven't crunched the numbers for all of the data... but here's two of them:

Gigantobugard:
Base drop rate: 32%
TH1: 49%
Maths:
(1-.32)=.21, or an increase in drop rate of 68%
Which results in a drop rate of 53.76%, which is within the margin of error based upon the 4% thresholds for both starting drop rate and ending drop rate.

Stirge:
No TH: 18%
TH1: 37%

Since the base % is so low, having that second roll increases the probability that the second roll will be needed...
(1-.18)*.18 = .1476, or an increase in drop rate of 82%
Or a total drop rate of 32.76%
Again, within the margin of error...

While of course, this is just the first potential way that I've looked at it, the higher base % drop rates would be a lower return on TH effects.

[Snprphnx]

TH is a myth, perpetuated by SE to get thieves invited to the show for at least 10 secs.

In all seriousness, thanks for testing. Good to see some people still striving to help us gain even more understandings of the details of the game.

[Amok]

While of course, this is just the first potential way that I've looked at it, the higher base % drop rates would be a lower return on TH effects.

I think this idea has merit. How does it hold up when applying it to TH2's increase? Does it fit a third 'roll'? I wish I paid better attention when I learned how to calculate there things.

For Strige Bat fang it would be?
0.18 +
0.82 * 0.18 +
0.82 * 0.82 * 0.18 = 0.44

Cus that fits it perfectly...

EDIT For the hell of it:
TH3 would then be 0.54
and TH4 0.62

All speculation, but I believe most players will agree the difference between TH2 and 3 and 4 has never been very noticable.

[Miraun]

I think this idea has merit. How does it hold up when applying it to TH2's increase? Does it fit a third 'roll'? I wish I paid better attention when I learned how to calculate there things.

By this, the subsequent additional rolls will always be less substantial than the preceding additional rolls, which is in line with observations...

If we were to abstract this out to TH3 and TH4, assuming that they followed similar mechanics, then that would be in line with the anecdotal statement that TH3 and TH4 are less significant than the increases of TH1 and TH2.

I'll try to make a spreadsheet.

Added Spreadsheet Link
http://code.google.com/p/mirauns-drg...downloads/list



Editted to add more stuff:

also, taking a look at the drop rates, they seem to be following either two patterns: 20/36/45 or 32/51/62, in line with a 20% drop rate, at 1 or 2 base rolls respectively.

[Thorny]

Granted I'm me and I'm an asshole to begin with;

Your calculations are off, you have a 6000 mob sample size(scattered around so much that most of the samples are too small to work with), and while your chart is pretty it's ultimately set up very poorly. There's no relevant data here, if trash mob droprates interest people that much I can start parsing my colibri logs and compare TH2 to TH0 over a good 10,000 kills with and without.

[Byrthnoth]

That'd actually probably be useful Thorny. Though we don't know for sure, we can assume that (at least) non-rare things follow this pattern. One prediction would be that things with higher drop rates to start with are more affected by TH than those with lower drop rates.

Think of Bahamut v2, with vs. without THF. My experience is that with a THF, you generally get a full or near-full loot pool and without a THF, you get ~5 items.

Everything in Bahamut v2 has a 30-50% drop rate, eh?
.3 + .7*.3 + .7^2*.3 + .7^3*.3 + .7^4*.3 = ~83%
.5 + .5^2 + .5^3 + .5^4 + .5^5 = ~97%

Now, this might not hold true for "Rare" drops, like Gaiters, Haidates, etc.

[Shamaya]

Good work Enedin!
Not that I need to tell you. The thief community has been behind you for quite awhile anyway, cuz you began this long ago. Speaking of that, I could swear you posted about this on BG before? Maybe I'm just thinking of something else.

Just to clear up a few concerns some had in this thread,
-TH doesn't stack from multiple players. The "highest TH taking precedence" was established when SE stated it at fanfest. The translation came out saying "for the party", but I have tested this for alliances to remove the last shreds of doubt that some had. Result was as expected; does not stack.
-The thief can be in any party in the alliance. It doesn't matter which party has the most hate or which party kills the mob. The SE commentaries indicate this, but the results of my testing (inside LJ for reference) also confirm this.
-Thf doesn't have to be the first person to get hate on the mob. Per SE commentary, they simply have to fulfill two conditions: (1)The thief must produce enmity (of any amount) on the target mob, and (2)The thief must perform an action on the mob. Whether these conditions have to occur at the same time or can be met separately is unknown, but my bet is that they have to be concurrent.
-It's been long believed that TH+1 items have, in common with other +1 items, a +1% effect. That is, in the case of treasure hunter and drops, each +1 piece adds a solid +1% onto the drop rate. This is added after the treasure hunter trait(s) is calculated. This is based of a test done years back by Divisortheory aka Ladyofhonor; link should be on wiki.

Many questions that people might have about TH are answered right on the 'other' wiki page. There are also links to testing and SE's commentary on the article and the discussion page.

@Enedin

Spoiler: show

About Beadeaux/Carpenter'sLanding. Chocobo & Mazurka work in landing but neither in beadeaux, so landing is outside and beadeaux has 'dungeon' status.

@Thorny

Spoiler: show

Your assessment is unduly harsh as you say. Given that you are right that essentially no TH test to date has met the rigorous Margin of Error standards that, for example, science communities require.. that doesn't mean data samples below that margin are useless. And it would be very wasteful and silly to ignore them completely. Remember, you don't always have to choose between "I believe in X" and "I don't believe in X"; it is perfectly intelligent and probably advisable to neutrally accept that "X is more probable than not", based on the evidence. If you want to do colibri parses though, by all means go ahead. A sample of that size could do nothing but help.

Miraun, after looking at those numbers I agree that what you're talking about is definitely worth looking into.

[Gryffes]

Something more rigorous such as Thorny stated would be nice.

If someone happened to save all their parses while meriting and analysed their drop rates on TH jobs and TH in party vs not you'd actually get something statistically worthwhile.

( It's not even unduly harsh people could randomly contribute parses and you'd get some sick sample size going on. )

[Enedin]

Very interesting replies.

@thorny

I know my sample sizes are laughable (I tried to find mobs with a drop rate as close as possible to 50%, which would make me require less mobs for accurate results), but gathering all this already took up so much time, it's impossible for one person to test this. I know someone around here started an online drops chart once, but I think that's discontinued.

@shamaya

I posted this before, but with less drops tested. It was the same lay-out, just less results. This is what I leave here, as my final chart. In retrospect, it's amazing I killed so many Bugards (got me loads of cash way back when though), Stirges and all kinds of other stupid mobs already XD

Also, Ranguemont Pass is actually a pretty cool zone, if you put up the right background music :D Those 1250 Stirges netted me Taisaijin ONCE, got the title! 1250 Stirges, that's like... 100 rounds of farming. 16 min repop = 1600 minutes = ~26 hours. Omg, I spent an entire day of my life killing Bats for virtual science.

@miraun

I love your idea, and it does seem pretty close. Most of us here (including me) suck at advanced statistical functions (although I understand them, just takes me a few minutes to get my brain to start thinking like a math wizz), so it would be nice to see all of this typed out, like you did in the Excel file (excellent work), for every mob I posted results for.

I input some of my numbers and it was all, indeed, very close. Assuming you're on the right track here, this could mean you've cracked the evil TH code SE put in their game. Please work your magic a bit more

TL;DR

Thorny, you're right, but it's an enormous task to do alone, and I'm doing whatever I can do to help. I hate error margins XD Even 1250 Stirges isn't enough.

Shamaya, thanks for the support. The THF community has had tons of shit over itself, and we're here to prove we're useful XD You're more useful though, Mandau and all that... >.>

Miraun, awesome stuff dude. It all seems to fit within the error margin. Try some more numbers and keep the goodies coming!

@everyone: thanks for the kind replies