Sucess Rates on high trivials

[Yroc Morf]

My understanding has always been that sucess rate could be calculated as:

(Skill x 1.15) - (Trivial - 70)

This was based on information found in the FAQs from JC.

The highest trivial that I have seen since the new UI was released is 335. This being the case, I calculate the following.

(252 x 1.15) - (335 - 70)
290 - 265
25%

Does anyone have any further detail on the mechanics of this formula. I know there is a suggestion that 5% success rate is the lowest, and that items which are trivial by 200 or more are no fail, but I have seem nothing concrete in this regard.

I'm most curious in the Grimel Robe combine. I assume the difference between 220 and 242 for this combine is a 20% increase in the sucess rate. But for obvious reasons I am unsure.

[Lothay]

See What sort of success rate should I expect? for the more detailed formula.

[Yroc Morf]

I can't get it to work out.

Can someone insert a skill of 252 and a trivial of 335 into this formula.

[trelos]

Originally posted by Yroc Morf
I can't get it to work out.

Can someone insert a skill of 252 and a trivial of 335 into this formula.

52% your example above

[Arghargh]

Keep in mind that those formula are inferred from experimental data. We don't really know what the formula is.

I can say from my experience that with skill 252, items that trivial at 335 have approximately 50% success, not 25% as the best guess formula suggests.

Arghargh Grumble, Darkblood
65 Ogre Shaman of Rallos Zek

200 Alchemy
250 Baking
250 Blacksmithing
250 Brewing
200 Fishing
200 Fletching
200 Jewelry
200 Pottery
179 Tailoring

[Nanndas]

I think the RNG has a lot to do also with qty and skillups.

On picnics:
Skilled up from 224 - 232 on 200 combines with 280 picnics. From 233-238 I made 340. Combined another 200 attempts yesterday and I got a whopping 480 or 60% and made it to 243.

As my skill raised so did the successful attempts. But I never thought I'd get 480 !

Nan

[Yroc Morf]

I understand the formula is based on experimetal data and that is actually part of the reason I am questioning it.

Based on the formula given above, someone at 220 tailor should make the Grimel robe 31.5% of the time. This assumes a trivial of 335 and my having the ability to apply the formula correctly

It also indicates someone with 242-250 skill would suceed 52.5% of the time.

I can say with certainty I was not that lucky (and my skill was much higher), however I do realize I only did 5 attempts so I don't have enough data to say one way or the other.

In addition to this, I tried to make some fineseafarer's items with the introduction of the berzerker class. They all seem to trivial at 335 as well (this is known in this case). I was getting about one sucess per 10 combines at a modified 242 skill.

Based on the formula, I calculated a 42.5% expected success rate.


Now that we know the high lvl trivials, we should be able to determine if this formula is correct in application since the fail rate is much higher on items with these incredibly high trivials.

I tend to think its not accurate based on my own personal experience, but I realize my data is extremely limited.

[KyrosKrane]

A couple of comments on the formula used in the FAQ.

1) It needs to be edited slightly; in the conversion from the old boards, the < symbol got changed into the HTML code &lt; which makes it hard to read.

2) The lower limit of 5% fails was removed some time back. This has been verified experimentally, if memory serves. It is no longer expected that you will fail 5% of all long-trivial combines.

3) There was also talk that items trivial 200 points below your skill level would become no-fail, but I don't know if that was ever rigorously established or not. I do know that in hundreds upon hundreds of low-level combines (blocks to bricks or sheets, bricks to smaller or bigger bricks and pieces, pieces to bits to studs, bricks to boning, and so on), I have never once failed a recipe under 50 trivial since I got to 250 skill. (I reached 250 skill after the trivial changes went in the game.)

[Lothay]

Thanks for the heads up Kyros, I updated the FAQ accordingly.

[KyrosKrane]

You missed a spot. < evil grin >

Success* = Skill - Trivial + 66 (for Trivial &lt; 68)

Also, while the 5% removal was confirmed on these boards (I believe it was either a developer who confirmed it, or numerous players posting experimental results -- I'll try to find the old threads), I'm not positive that the 200-point trivial thing was ever confirmed as having made it off of Test server. Again, let me hunt up the threads to see if I can find the info.

[KyrosKrane]

OK, here's what I found.

Link 1: Splunge working with Frizznik on Test server, June 12, 2003. Frizznik confirms that you cannot fail on combines that are 200 or more points trivial. Less trivial combines have lower fail rates too.

[Thu Jun 12 18:47:34 2003] Frizznik tells the group, 'if you make something that you are 200 over the trivial on'
[Thu Jun 12 18:47:38 2003] Frizznik tells the group, 'you will never fail'

Link 2: From the patch message on July 10, 2003:

- Tradeskill chance of failures changed - The higher your skill is over the trivial of the combine, the less likely you are to fail with a combine. This is a reduction on the chance to always fail.

Link 3: Multiple confirmations that items trivial 200 or more points below you no longer fail. Items with low trivials (but not 200 or more points) still fail, but at lower rate. Many other threads confirm. c. July 2003

[elvirra]

Brewing skill 250 + trophy. KC is trivial at 335 according to the new UI. My success rate is always 45%. No more, no less.

My husband's druid has a skill of about 230 + geerlock, and his success rate hovers around 40%.

Wondering how this fits in with what we know of formulas and the experience of other players.

[KyrosKrane]

My first question would be, over how many combines did you get that 45% (and 40%)? If it's less than, say, 1000 non-trivial combines at the same skill level (including the use of a geerlok), then the results are prone to the effect of the RNG. I'd hazard a guess that 1000 or more combines provides us with a sufficiently large sample size to use in refining the formulae.

[kiztent]

Does anyone have Laurefin and Ishwar's work on the original formula archived? I'm especially interested in the datasets. It would be nice to reconfirm the formula.

Suffice to say, you'd need some large datasets to make any conclusions about success rates. I'll leave it to a statistican to comment further.

[KyrosKrane]

The work is still up on EZBoards, actually. Here are some threads to look at:

The Truth about INT and WIS and item trival on Skill Inc -- the forumla was first posted in the middle of this thread, I believe.

Laurefin discovers success rate formula

*derailed thread. Was re:velious tailoring, now stats&VI

There are a few others too, I think.

edit: Also the first pass at writing the FAQ.