Recently, some folks here and in
ColorlessGreen's
UEI thread were discussing the method that the GDocs spreadsheet uses to calculate FGei. In particular, people noted that the GDocs spreadsheet uses an average card spin rate for all FGs, while the STATMASTA™realtec spreadsheet (hereinafter called "STATMASTA") uses FG-specific card spin rates.
Today, I researched this issue for more than four hours. Yes, four.
EDIT: After posting this, subsequent discussions and research revealed that I misunderstood the timeline of the events I found in my research. As a result, I drew faulty conclusions. I have stricken bad info below, and inserted some good info in yellow. My most abject and sincere apologies to the individuals I named as having committed errors, and to the Community at large. I should have done this better the first time.DISCOVERIES1) STATMASTA calculates an individual FGei(c) for each FG, then calculates FGei(cn) for the deck by adding together all the individual FGei(c)'s.
2) STATMASTA's individual FGei(c)'s are calculated using DarkWeaver's
original calculations, which are hard-coded into the spreadsheet.
3) DarkWeaver's original calculations are wrong. Various people, including DarkWeaver, noted errors in his method.* In addition, an empirical study of FG drop rates provided widely differing results. Many of DarkWeaver's figures were off by more than 10%. At least one was off by more than 20%.
*) DarkWeaver's original post provided highly detailed statistics. After correcting (some? most? all?) of his errors, he never again posted statistics with the same level of detail.
4) The empirical study mentioned in #3 is outdated. The FG decks have changed since that study, and those changes affect the drop rates.
5) Even if DarkWeaver's (and, by extension, STATMASTA's) drop rate figures were completely correct, they are outdated. See #4.6)
XenoSim™ (Xenocidius's
Slot Machine Simulator) produces a "cards per game" statistic. This statistic is exactly equivalent to the drop rate statistic that FGei uses.
7) XenoSim™ was last updated in 2011. Although doubtful, it is possible that Zanz has modified the algorithm since that update.
CONCLUSIONS1) STATMASTA is inaccurate, if not useless. Its core calculations rely on figures that are both wrong and outdated.2) The Community has no information that is both accurate and current on individual FG drop rates
for Akebono or Osiris.
RECOMMENDATIONS1) The Community must obtain accurate, current individual FG drop rates
for Akebono and Osiris, by one or
both more of the following methods.
a) Use XenoSim™ to calculate expected drop rates for each FG. Use of this method assumes that its algorithm still matches the algorithm the game uses.
b) Conduct large-scale drop rate testing similar to the previous study.
c) Duplicate DarkWeaver's original calculations for the two changed FGs. This is the preferred method.2) Modify the GDocs spreadsheet to use the
figures from recommendation #1a known good FG spin rates to calculate more useful FGei's.
3) Use the current GDocs spreadsheet to achieve recommendation #1b. Compile all current and future data to calculate empirical drop rates for each FG.
4) Seek a member of the Community both able and willing to achieve recommendation #1c.I sent a PM to Xenocidius, asking about the algorithm and such. Once I know the tool is good, I will spend
a few days an hour to achieve recommendation #1a. Once I've done that, achieving recommendation #2 will take just an hour or two.
Recommendation #3 will take more time. Like, a bunch more. Don't even ask how long. I don't know. I have several RL priorities that are much higher. Once recommendation #2 is complete, recommendation #3 is little more than a backup plan, or maybe a method for testing future changes to XenoSim™. I suspect it would be a low priority for the Community at that point, anyway, and the data aren't going anywhere, so no big.
The release of v1.4 "sometime" may change things so drastically that none of the above matters, anyway. Stay tuned.
