Tradeskill Skill Up Formula

[Bidoof]

You're right, there is a change in % between 10 and say 180 when leveling up, on both our formulas. The change is where I disagree.

I haven't messed around with the formula you posted and use in your spreadsheet, and if its giving you results like that, its significantly different from the linked one. The simple first formula in the link I posted is merely -after a simple reduction- this:

((STAT-x)/(Skill_Diff*y))/100

where x is 0 or 15 (depending on if its a INT/WIS skill or INT/WIS/other skill) and y is 1 or 2 (1 for succes, 2 for failure).

With this formula, successes increase your chance to skill up (obviously). Let's use those arbitrary numbers above. The best trivial to use at 10 is a theoretical 11, since you're most likely to succeed with it. At 200 INT, x=15 and Skill_Diff=2 (baking!), using the success equation for under 68 in the wiki gives us a success rate of 65%. If we do some ratio math and use just the above equation, we get a skill up chance of 76.31% overall.

Now, using the same INT, x and Skill_Diff, let's say our skill is 180 and our trivial is 181. Using the over 68 equation (what about equal to 68?!?), you shoot up to a success rate of 95%, hitting the 5% chance to fail cap. As would be expected from a higher chance to succeed, using *just* the formula above gives us a 90.19% chance to skill up. Obviously you might not be able to find a 181 trivial combine, so let's say 200 trivial. Then, your success rate is still 81.5% (higher than 65%), and your skill up chance is 83.94% (again higher than 76.31%).

If you add the second check- the one that actually directly uses your skill level- the 11 trivial with 10 skill barely drops to a 72.5% chance, while the 181 trivial with 180 skill drops drastically to 9.02%.

What I'm very lengthily trying to say is: either your formula (which I will admit I haven't done much with) is completely different from the first part of the one I'm asking about, or someone's math is off. I'd probably just say mine is wrong, as I like the appeal to authority fallacy, but feel free to take a look at it. This is more of a plunge back into math for me (even if its mostly just algebra), so I probably got something wrong.

Either way, I like this conversation. Its not often I get to talk math with anyone.

[August]

Yeah you are right - and in fact you do see that chance to succeed increases as you go up due to the formula change.

Looking at my code, I have another modifier in there. Don't ask me where i got it- I got it off of EQTraders boards some time ago.

* MIN (.95 , (200-SKILL/200))

Basically an overarching curve handler. I BELIEVE this was the result of the second roll that is mentioned, and perhaps was more complicated and boiled down to this result? I honestly can't remember.

What you can notice is that this also follows the rules of the '190' gap that was mentioned -- at 190 skill MIN becomes always .95 since 190/200 = .95. So, it could have been from different sources, or i could have used this very source and done the conversion, but there you have it.

FWIW This is a tool I used very heavily during leveling on FIPPY (which once again, doesn't answer your question about being valid here). My extended spreadsheet calculated 'expected combines til next gem' and so I used this to prebuy my mats so I could go sit in EC and sell while i enchanted my bars / leveled up. I was very nearly spot on, so I believe it has real validity. I know this server does not completely emulate what live was, however, so in all honesty we need a dev to chime in on the tradeskill formulas. If you look at things like the AC spawn and Pyzjin spawn, we clearly have our own separate implementaiton compared to live.

-tomtee

[August]

Yep figured it out. I took the original code, split it up into ratios, and then divided by 1000 since that is what the roll of N is for - giving the percentage of the chance that we would pass the first roll, given that, when you go the second roll, there is the curve modifier that checks to see if you win the roll based on the 190 artificial modifier.

So, more or less, my formula in the spreadsheet is that exact post bubbled down into some crappy algebra. It isn't pretty, but the curves it produces seem fine to me. (pun intended).

Whether or not that is P1999 implementation, who knows?

[Bidoof]

Figured I'd ask one more time, then I'll give up. Anyone happen to know if the formula post at the EQ Traders boards from 2004 from Live is the formula used here?

Also, shameless plug for something almost completely useless. I made my php script that does a bunch of randoms and calculations to give you a sample set of possibilities while raising your tradeskills. You input what trade, what your stat and skill scores are, and a set of trivials you'll use to level up with. In return it'll spit out a randomized outcome, including how many combines it takes, how many successful combines that includes, and the like.

Its in my sig, but I might as well post it here too: eq.e42randy.com/test.php. Its running off the formula above, so again, if you know of a different formula, please respond!

(Useless program? Yes. But at least its slightly more useful than the last thing I wrote.)