skillup formula

[Sylphan]

Here's how we're told skillups are calculated.

Each tradeskill has a difficulty Y: 2, 3, or 4.
Tailoring is 3
Smithing is probably 4
Pottery used to be 2, now it's probably 4
Jewelcraft is probably 2
Fletching
Baking
Brewing is probably 2
Tinkering
Alchemy
Poisonmaking

For tradeskills with an alternate stat set S = max(int,wis,alt)
For tradeskills with no alternate stat set S = max(int,wis)-15

If the combine failed set F=2
If the combine succeeded set F=1

Set N=(S*10)/(Y*F)

If ran(1000) < N then proceed to the second check

Second check: If ran(200) >= min(190, skill) then skillup!

===========================

At skill 0 the second check is basically 100 percent, so the chance of skillup is just the first formula.

Suppose you were doing brewing (no alternate) at skill 0... your percent chance of skillup with various INT values:

int success failure
200 92.5 41.2
250 100 58.7
255 100 60
280 100 66.2
305 100 72.5
355 100 85

Now suppose you're doing smithing (alternate is Strength) at skill 0... percent chance of skillup becomes:

int success failure
200 50.0 25
250 62.5 31.2
255 63.7 31.8
280 70.0 35
305 76.2 38.1
355 88.8 44.4


The same charts at skill 100 would show 50 percent of the listed numbers, due to the second check.
The same charts at skill 190 would show 5 percent of the listed numbers, due to the second check.

I think there's one detail missing from the formula: we haven't yet explained why chance of skillup is better with 200+ skill than it is in the 190s.

[Verdandi]

Is this all information that was provided by the devs in the tradeskill forums at the Vegas FF? For the record....

[Nizanko]

The explanation is right in there regarding the easier skillups post-210. It's related to the "bend" that occurs in the second check from the min(190,skill) term.

An upward trend appears anytime you are working on something far above your trivial level. This is caused by your increased chances of success at higher skill.

Since success has a significant impact on skilling up, being at the point where you just about always get the maximum success rate is where you want to be. It's a lot easier to get that the higher your skill is.

I plugged these numbers into a spreadsheet, and found that the knee in the graph occurs just above 190 skill (which is where the second check stops getting worse). From then on, the skillup rate at best increases very very slowly, depending on how much above your trivial level the item is as well as your stat.

For something like JC, where the items you make are always just a few points over your current skill, the chances never increase - they just bottom out. But that's okay, because they bottom out at the best chance for skillups you can achieve.

For something where you may be working quite a bit above your skill, your chances of skilling up actually go below that threshold and then increase once you pass 190. For example, if you do something that trivs at 252 for a really long time, then your chances will bottom out at about skill 190 and then increase until 233, when the success chance reaches 95% again. In the JC case, if you worked on something way harder than your current skill, your chances would increase past 190, but that's only because your chance to skill up was already way below the best possible.

[Sylphan]

It's from the programmer guy on the tradeskills panel at 2004 Vegas FF. He related it to us from memory, but said he'd spent hours poring over this particular bit of code to be absolutely sure he understood it... and he certainly seemed to.

The actual code apparently used floor() rather than min(), or even calculated it manually (gotta love spaghetti code)... and there were apparently some case structures not just if-then-else. Also the second check probably uses "<" rather than ">=". But only old programmers like me would care about such subtleties; what I've described is technically accurate.

[Hibashira]

Let me try a summary. Assumming the above are accurate (I guess there is no reason to suspect otherwise since it came from the Dev):

- Max your stat (maximize N, easier to pass check 1)
- Its never an overkill to have max stat since the chances of passing check 1 on failed combine is always below 100%.
- For smithing, the max effective stat is 355. 340 for all other skills. (Whether Dex is an alt stat for fletching is still uncertain, I presume...)
- Skill up is more likely on successful combine (due to higher N, thus passing check 1). But success/failure does not affect check 2, which can be viewed as the "capped" skill up rate.
- The lower your skill, the more likely you are to skill up ... up to a point.
- As you go pass 190, your current skill is no longer relevant in terms of skill ups. Chances of passing check 2 is 5%. i.e. 1 in 20 ... if you can get pass check 1.

[Renjah]

Head hurting...... must contol... pain.....


BLAM!

K made my head explode It seems they had a good working equasion then somewhere down the line people "fixed" it without cleaning up the code. I bet this is somewhat standard As a programmer I know this is extremely common practice (heck I do it myself!). I'm glad they did finally explain it. As soon as our great minds pick it apart we can answer so many questions.

[kiztent]

edit: covered by skillup calculators

[papyscott]

Double, triple, no quadruple STICKY!

OMG this information rocks...


Papy

[Furyrpied]

Now if you could break it down so that not only Gnomes can understand it....

Okay, max out your (wis/int/str depending) for better skill up chance
averages start to factor in at 190?
I'm soooo lost

[scarbrowed]

Heh, it's pretty darn complicated. The answer is...... it depends. In another thread I walked through the whole calculation for a baker with WIS of 200, but only 190 with geerlok, with a current skill of 191, making HMP's. In that case, use the geerlok.

On the other hand, if you are tailoring at 191, with a WIS of 255 (245 with geerlok) and making arctic wyvern masks, the extra WIS helps more. 4% more likely to skillup in fact.

I have an Excel spreadsheet that does the calculation, in case you want to try it yourself.

[elty]

I rememebr in a long long long time ago, a patch that said something like
"tradeskill at above 200 will now have a double chance of skill up" or something to that effect. Anyone remember that or I am wrong?

also, they mention in a patch to make brewing, baking, tailoring, smithing easier. So how can tailoring and smithing be like 4? there is no improvement at all

[Yakk]

End game math (skill over 190).

Lets say you have 355 int/wis, and you are working on a "best of 3" skill, and your skill is over 190.

Skillup chance on success/failure:[2]
Hard(4) skill: 4.4% / 2.2%
Medium(3) skill: 5% / 3.0%
Easy(2) skill: 5% / 4.4%

On a "best of int/wis" skilll:
Hard(4) skill: 4.3% / 2.1%
Medium(3) skill: 5% / 2.8%
Easy(2) skill: 5% / 4.3%

With a 305 stat cap, "best of 3" skill:
Hard(4) skill: 3.8% / 1.9%
Medium(3) skill: 5% / 2.5%
Easy(2) skill: 5% / 3.8%

With a 305 stat cap, "best of 2" skill:
Hard(4) skill: 3.6% / 1.8%
Medium(3) skill: 4.8% / 2.4%
Easy(2) skill: 5% / 3.6%

In effect:
More int/wis/secondary stat will help you get skillups, to some degree, no matter what.

However, if you always succeed on your combines, and the skill is 'easy', there is a point where more stats don't matter.

Max useful stat if you always succeed on your combines:[1]
Easy skill, "best of 3": 201
Easy skill, "best of 2": 216
Mediium skill, "best of 3": 301
Medium skill, "best of 2": 316
Hard skill: no cap

On failures, extra stat points will always help your skillup chance. Some skills get lots of failures when skilling up, some get lots of successes.

So, there is no real need to have your int/wis over 215 if you are doing barely-not-trivial recipies in brewing or JCing with a geerlock, apparently.

Footnote(s):
[1] I solved for S in 1000>(S*10)/(Y). You might need 1 less point.
[2] There are two "tests" that have to pass in order for a skillup to happen.

The first is the "int/wis/other" check. The second is the "low skill" check.

Because I'm dealing only with 190+ skills, the second check succeeds 5% of the time.

So, we get the int/wis chance * 5% as our success rate.

Int/wis/etc chance is
(S*10)/(Y*F) / 10 %
with the note you can't have a higher than 100% chance of passing it.

For success:
S/Y %
For failure:
S/(2*Y) %

Multiply by 5% and we get:
Success:
S/(20*Y) %
For failure:
S/(40*Y) %
with a max skillup rate of 100% * 5%, or 5%.

Edit: Noir, looks like you where right btw. =) Skillups on those PotC recipies is crap!

[Tarelna Prexus]

People have always talked of hell levels for tradeskills from about 180 to 220 but with this published formula, its looks like its diminishing returns till it bottoms out at 190 and never gets any better.

So if I get this right, all skill ups from 190 skill onwards are 'hell levels'. The only variable that will effect that is success rate and its obviously more advantageous to go for near trivial combines and combines with a naturally high success rate.

From 190 onwards with maxed stats, depending on which tradeskill you will average from 5% (max) to 1.6% (min) skill ups per combine.

Or am I missing something ?

[Lushissx]

OK, I kow casters can get stats to 355, but how can you get strength past 305. I have the planar power and none of my other aa's appear to allow me to exceed the 305 cap? Are you just saying your strength would be 355 if there was not a cap at 305?

[kiztent]

Originally posted by Lushissx
OK, I kow casters can get stats to 355, but how can you get strength past 305. I have the planar power and none of my other aa's appear to allow me to exceed the 305 cap? Are you just saying your strength would be 355 if there was not a cap at 305?

Only casters can get stats over 305, and only by taking innate enlightenment AAs.