... are you hoping to get this variance number for playboy specifically for some reason or is it part of a wider effort to get this same data for as many slots as possible?
My ultimate objective is to assign words to numbers. In the thread to which I linked in my previous post was the following:
OK, so we've got the hard data on Gems of Isis:
1. It has a Variance of 51 +/- 1.5.
2. It has a Max Line Hit Frequency of about 35%.
3. It has an "Average Number of Spins Between Free Game Wins" of 140.
Based on "game play feel", ChopleyIOM thinks that this is a "Medium" variance slot.
zap987 feels that this is a "Low" variance slot, based on both "game play feel" as well as on some "max possible wins" calculations.
I think that, just as with Theoretical RTP, Variance is another slot performance value which can be used by the Player to answer the question "OK, what can I expect from this slot?"
before they start playing. I've heard it stated many times that Players like bonuses because of the extended playtime. If they're (unknowingly) dropping $1 a round on a "High Variance" slot, they may find that their extended playtime is cut pretty short, pretty quickly.
I strongly feel that a word (High, Medium, Low) would better serve the Player than a number. However, just as with buying a pair of pants, it would be nice if one slot manufacturer's words were associated with the same Variance numbers as another manufacturer's words. Otherwise, High variance for one manufacturer might be Medium variance for another manufacturer. (To continue the analogy - the pants don't fit.)
Also with regards the free spins data, my understanding is kktmd has his data structured Such that the value of the free spin row is already equivalent to the total of the trigger spin and the bonus round. Assuming I am right, should it not be possible to simply square the existing value to use in the calculation as opposed to any extra code or subselect on the individual free spins? (I am unaware if they were recorded anyway too actually)
If this is true, then I think you're right - it would make the calculation of S1 and S2 very simple. In our database each round, free spin or standard spin, is recorded in a separate row, which is why our S2 calculation was a bit trickier.
If this is true, then it would also make the calculation of WRC a simple "SELECT COUNT(*) FROM table_name". In our table, there is a field gametypeidno, which has a value of 1 for a standard spin and a value of 2 for a free spin, adding a WHERE clause to the count. (I mention that because sometimes I've forgotten that WHERE clause, completely hosing the results.)
I don't think you mentioned which slot(s) have already been analysed in this one you have done the 40 mil spins on, but would love to know?
All of our slots have now been analyzed using a 40-million round sample set. It would be nice to use a 250-million round sample size, but we dump the data into a database, which takes time. However, having the data in a database allows us to do additional analysis, like a CHITEST evaluation of the reel stop value distribution.
If you, or kktmd, can calculate the reported Variance values using the formula provided by Mike, then I, for one, would find it very interesting to see Player's "subjective feel" posts on the slot's variance, and compare a "consensus variance word" for that slot to the calculated Variance number.
Chris