Parse Results! Materia Conversion Probabilities

BACKGROUND
In lieu of the game developers posting any details on the materia or stat changes I have decided to take matters into my own hands with regard to the probability of what materia you can expect to get by converting gear.

Please see my blog for other information and testing:
Deus Diabolus' Blog

Over the last month I have collected log entries for all materia I have converted. Special thanks goes to my LS, Gin Ichimaru, Khymerra Rhae, Mizzy Lowe, Dark Chops for helping me convert all this gear.

I wrote a program to count and calculate the probability to create each specific materia for the gear I mass produced. Pay close attention to the gears where I converted over 100 for more accurate readings.


GOOGLE SPREADSHEET RESULTS -- UPDATED 2/10


Initial analysis showed a correlation between:
--The type of materia you can insert into gear slot X
--The type of materia you will receive by converting gear that equips to slot X

UPDATE 11/7
I have fully analyzed the results in an optimal format. I created a conversion probability chart organized by materia of each grade in one dimension and the number and percentage for each gear of each slot converted via parsing logs in the other dimension.

Further, cells that coincide with gear slots where that materia can be melded are highlighted to show the correlation or lack thereof indicated by my initial analysis above.

The Raw Data is still retained and is in a separate tab in the spreadsheet. The Ranks of all gears are now indicated. Gears of less than Rank 41 are removed.

MATERIA STAT WITHIN EACH GRADE
The quality of the stat on each materia obtained will not be published here, but I can tell you that materia of say MNDIV or STRIV +16, 17, 18, 19, 20. The probability of obtaining each point respectively within the grade IV stat materia is roughly: 40%, 40%, 15%, 0%, 5%. These are not parsed measurements but to give you a rough idea of the rarity. According to tymora in the reply below, the dat files were observed to confirm that the +19 does not exist.


UPDATE 2/10
--Updated the results to reflect a 2-3x increase in raw data to produce more accurate assimilation probabilities
--Integrated the December 1 Lodestone assimilation table to definitely show that empty cells mean the materia is not able to be assimilated from that slot. Items with numbers other than zero are results I myself saw and integrated into the probability calculation.
--Green cells continue to be accurate and show that that type of materia is able to be melded to that slot.
--Raw data is provided on the second sheet.
--This is my final update.

Thanks but just a few things...
1) Each tier has 4 grades for a total of 16 in all. Tier 4 stats give 16 17 18 20. (no +19) This is from the Dats.
2) It would be nice to include the level of the gear converted in your spreadsheet as that affects the probabilities.

In my own findings, it appears that all gears of the same equipment slot give the same types of materia. So I agree with your organization of the output based on slots.
E.g. Crafting tools, both main and offhand only gives crafting related materia. Even though you can use stat materias on mainhand tools, you don't get them from crafting tools.

Hey... I have been screenshotting the chat log when we have been breaking materia as a linkshell....

Please read the updated original post. Please also email your conversion logs to me to contribute to this project. See OP for details.

UPDATE 11/7:
I have fully analyzed the results in an optimal format. I created a conversion probability chart organized by materia of each grade in one dimension and the number of percentage of each gear slot converted via parsing in the other dimension.

Further, cells that coincide with gear slots where that materia can be melded are highlighted to show the correlation or lack thereof indicated by my initial analysis above.

The Raw Data is still retained and is in a separate tab in the spreadsheet. The Ranks of all gears are now indicated. Gears of less than Rank 41 are removed.

Emailed you all the screenies

Thank you for sharing this thorough information. Very fascinating.

Have you noticed an effect from the moon phase?

Normally getting IV is rare for me, but recently I got 8/10 IV materia (4 in a row) and noticed it was a full moon.

It might very well be a coincidence but thought it would be worth mentioning to someone who has the resources to test this out as I do not.

Thanks again for this info and all your other great posts, I look forward to your future articles.

P.S. Have you derived the changes in strategy of the Darkhold raid since 1.19? I have not been able to find this info anywhere and would love an update.

UPDATE 2/10
--Updated the results to reflect a 2-3x increase in raw data to produce more accurate assimilation probabilities
--Integrated the December 1 Lodestone assimilation table to definitely show that empty cells mean the materia is not able to be assimilated from that slot. Items with numbers other than zero are results I myself saw and integrated into the probability calculation.
--Green cells continue to be accurate and show that that type of materia is able to be melded to that slot.
--Raw data is provided on the second sheet.
--This is my final update.

I have around 1600+ materia conversions logged but they don't have the exact #'s its like:
Dated Electrum Ring, STR Materia IV, PlayerName

If this would help you I can email/PM you the results. I made a mod to my parser which collects materia conversions. Since I don't think the base item affects the outcome IE R41~R50 NQ/HQ has the same exact result % so I was most concerned with just what the %'s were to evaluate the market value associated with a type of materia, IE its a 5% chance at on getting healer hand so the cost justifies the conversion % where PIE is a 40% chance on earrings so the cost does not justify the %.

Yeah I noticed that in the parser output actually. Unfortunately it took me a few hours of manual work after my automated portion to translate the data into the google spreadsheet so the probability results are read-ably useful.

The actual output from my script takes the log line (the same data you captured minus player name) and creates totals and %s like what you see in my Raw Data tab. Then I have to sort the raw data manually by slot type and enter in the total quantities and formulas cell by cell. Excel has some annoying limitations on the formula. So I'm going to hang my cap on this project so I can focus on other ones, but thanks for your contribution.