> 95%

[Hibashira]

A while back, SoE implemented a change that makes success rate can exceed 95% (the higher your skill is at over trivial, the higher the success rate, etc.)

Anyone done any work on calculating the numbers? I see bits and pieces in different skill forums but nothing substantial.

I was thinking about doing something like that for, say, brewing:

Skill up to 46 (Heady Kiola trivial)
Do 200 combines of HK. Tally results.
Skill up to 47.
Do 200 combines of HK. Tally results.
Skill up to 48.
Do 200 combines of HK. Tally results.
...
etc.
...
Compare.

Would that make sense? Is 200 enough for each skill level? I only say HK because its low trivial, reasonable cheap to make, sells reasonably well in baz, and no big lost if I have to destroy them.

Would that be enough to construct a prediction model to test other tradeskills?

Please comment.

[Cubwynn]

Now, we're all masochists here. But dang... I doubt there's anyone that masochistic.

[Dunthor Warsmith]

At 200 attempts, you would expect 10 failures under the old 95% system. That's a pretty small number, well within the range for reasonable variance (I'm not statistician, don't know the math behind calculating the statistical significance of a few more or less either way).

Still, if you tested it at EVERY skill point like that, you could probably get a half decent curve. But then there's that whole "sanity" bit ... the old system formulas have a 46triv item going 95% at what, ~60? I'd personally start there, and maybe do a run of ~400 every 10 points. That way it would a) be mildly more sane, that's only about 8000 kiolas, and b) a bit bigger pool (20 out of 400) for expected, although fewer data points to less to extrapolate (or here, interpolate as well) from.

But to get a solid point, I'm guessing you'd need in the neighborhood of 8-10k kiolas at that skill, although that number is more off of mathematical intuition than actual formula - go find a statistician, EQTC is full of them

[Nolrog]

At 200 attempts, you would expect 10 failures under the old 95% system.

You would expect 10 failures, on average. You could have 20 or 30 failures one time, and 1 failure the next. Depends on how much the RNG likes you that day. But if you did 200 attempts 200 times, you will see the failure rate get close and closer to 10.

My understanding of the change is that you have a 5% chance to fail on any combine until your skill is 100 points higher than the trivial level. The failure rate then declines at a steady rate until you hit 200 above the trivial rate, at which time, it becomes a nofail combine.

[Myrron]

The no-fail idea existed on test only, it did not make it live. There have been a few reports posted on these boards of 250 tradeskillers failing on something with a trivial below 50, but according to them it seems to be extremely rare.

--Myrron

[Abazagaroth]

I was gone for awhile, so missed the changes they supposedly did. However, last year when I was mass producing blue diamond cultural (I posted the actual numbers on these boards) I *think* over like 2000 trivial combines on the blue diamond combines (bd -> dust, dust -> temper, temper -> ring) I was at like 93% or something success.

However, I can say that something has changed, as over thousands of 3 small bricks -> one large and hundreds of chiselling ore into smaller parts, I have had ZERO failures since I got back last month. So either something changed with trivials (250 smithing skill), or they made those ore conversions no fail.