So all I'm doing is crafting today. 3x an 8 piece set for guildies and some weaponcrafting on the side. First set i crafted 4mp gems within 40 tries. Then no more. Switched to weaponcrafter. 30+ tries, 1 99 qual weapon and 3p lost. Then the second 8 piece set armour/weapons for a guildy spellcrafting. 6 mps within 60 tries, 3 mps on one piece alone (no chance to swap), after this no more (or perhaps even 1 more). Hade a couple of tries on the weaponcrafter again and after 10tries, 1 99 qual (I'm not an idiot). Switched back to spellcrafter for the 3rd set. 2mps within 30 tries. Call it conformation bias all you want, but statistically it's just plain wrong.

I'd also like to add that the chance for 99 qual is lower then anything between 94-98. Anyone can confirm?

Yes, ur correct, its random.

Satiah wrote:Call it conformation bias all you want, but statistically it's just plain wrong.

You have some chat logs / analysis of expected vs actual results with a decent sample size to show this?


Sent from my iPhone using Tapatalk

Lurker wrote:

Satiah wrote:Call it conformation bias all you want, but statistically it's just plain wrong.

You have some chat logs / analysis of expected vs actual results with a decent sample size to show this?


Sent from my iPhone using Tapatalk

I have no clue how to do this. Please inform.

You can enable the chat log using /chatlog

This will output the text in any chat window to a chat.log file in (default in windows) C:\Users\<username>\Documents\Electronic Arts\Dark Age of Camelot\

You will want to craft for a while to get enough data on particular gem levels etc for that to be statistically viable dataset. (The larger the better I suppose)

You can find a chat log parser here... http://capnbry.net/waldo/ that will provide some insight into the crafting results you have seen that will be nice and reliable (not subject to confirmation bias).

Im not sure what the expected results would be. You would need to research that or do the same tests in live to demonstrate a difference I suppose.




Sent from my iPhone using Tapatalk

Actually it's besides the point. I get that you need a large data set to get accurate numbers but that's not what I'm trying to say. It's the streaks. As i mentioned with the mp's they all came shortly after eachother. Let's say on average you need 10 tries to craft a 99 qual gem, so that 8x4x10= 320 tries. 3 sets = 960 tries add the couple of tries on weaponcrafting and we got a sample size of ~1000.
I made around 12/14 mps. So yeah, that might even be on the low side (but perhaps the sample size was lower), but statistically it's correct.

However like I mentioned 4 of those mps were made in 40 tries and another 6 of those within 60 tries.

Edit: I tried the parser, but the graph shows 0% for all qual. I take it it requires a large sample size before it gives an output?

http://www.craftsage.com/probability.html

You are welcome.

Very nice write up from somebody that actually understands statistics.
Good Stuff!

Lurker wrote:Very nice write up from somebody that actually understands statistics.
Good Stuff!

Indeed. It kinda proves my point.

Nothing can prove your point except demonstrating with data unfortunately. I'm not arguing that you are wrong or trying to be dismissive at all, but without chat logs / analysis then there's nothing that can be proven and it's all anecdotal.

Lurker wrote:Nothing can prove your point except demonstrating with data unfortunately. I'm not arguing that you are wrong or trying to be dismissive at all, but without chat logs / analysis then there's nothing that can be proven and it's all anecdotal.

How are you gonne prove the point of having streaks if people demand larger data sizes?

By providing the data?

You can analyze the streak distribution against what might be expected given a data set of any given size and the knowledge around what the underlying probabilities are.

Its random of course so there will be some variance but if there was a significant issue this type of analysis would potentially highlight it.

I'm happy to put some work in on this for you if someone can provide the chat logs. Given that, I believe, the chance for an MP is 2% regardless of item level (assuming it's grey). You could do this with low level items so as to save time / gold.

I will admit that my gut feeling at the moment is that there is no significant issue here. But I'm more than happy to look at the numbers in an impartial way and try to see what can be determined from them. I'm also very happy to have my gut feeling proven wrong!

Around 200 tries now. 12 99 quals, no mps.

Edit: thats over 1200 diamond metal bars and 658 arcanite bars and 200+ duskwood (with salvage everytime on the arcanite).

Satiah wrote:Around 200 tries now. 12 99 quals, no mps.

Edit: thats over 1200 diamond metal bars and 658 arcanite bars and 200+ duskwood (with salvage everytime on the arcanite).

the amount of materials used adds nothing to this discussion...
and "around 200" tries is highly unspecific

on top of that, anyone can mkae up numbers, but if we assume 200tries without MP, the chance for that happening is ~1,76%.
ironically almost the same chance as getting an MP with one specific craft

Cant say how many times i wrote this now.. but its not the ONE item you can look over to make the RNG luck..

i needed more then 80 trys for Cloth MP Vest last days..

then i needed 4 more tries for my MP Cloth Legs, and switched to AC to do 20 trys on gloves and got 2 mp too.

so when i now only see the Vest, i would say : omg what a bad mp rate

but when i see the other 3 mp too .. i can say : halleluja what a nice RNG luck