I can try to give you some input, but I cannot explain JM's calculation.
_RefDropItemGroup : SeleteRatio
This should be a % value inside a give dropgroup (=D15_B). If you add only one item into the group, it will always choose the same item. Normally you need to add multiple items to your dropgroup and from the given set of items it will roll out one item using the SeleteRatio.
Example:
Item1: SeleteRatio = 0.25
Item2: SeleteRatio = 0.5
Item3: SeleteRatio = 0.125
Item4: SeleteRatio = 0.125
If roll for %drop of droupgroup on given object is true, then roll2 for SeleteRatio within the given dropgroup. With 25% prohability Item1 will be selected and dropped (DropAmount and DropAmount Max = 1).
If roll for %drop of droupgroup on given object is true, then roll2 for SeleteRatio within the given dropgroup. With 25% prohability Item1 will be selected and dropped. Additionally with 50% prohability Item2 will be choosen and dropped too (DropAmount and DropAmount Max = 2).
And so on.
Regarding DropRatio question:
This one is more complicated, because JM choose several settings to modify %drop on monsters (server setting, database setting, buffs/items, amount of drop stacks per mob type...). To get the real values, you can only run a long term test multiple times (/zoe2).
If you set a prohability of 1/15000, there is no guarantee, that you drop exactly one item within 15000 kills. The drop will vary according to a normal distribution, unless you choose to modify something. Everytime you kill a monster, the GS will roll out a dice. With 1/15000, you can roll 1 to 14999 and drop nothing. Only if you roll 15000, you will receive a drop.
I created a simulation within Excel, to test your setup.
With 1/15000 and 30k kills, you can get 4 drops with ~4% prohability.
With 1/15000 and 30k kills, you can get 2 drops with 25% prohability.
With 1/15000 and 30k kills, you can get 0 drops with 15% prohability.
With 1/1500 and 30k kills, you can get 30 drops with ~2% prohability.
With 1/1500 and 30k kills, you can get 18 drops with ~29% prohability.
With 1/1500 and 30k kills, you can get 14 drops with ~20% prohability.
With 1/1000 and 30k kills, you can get 45 drops with ~2% prohability.
With 1/1000 and 30k kills, you can get 31 drops with ~28% prohability.
With 1/1000 and 30k kills, you can get 22 drops with ~16% prohability.
With 1/1000 and 60k kills, you can get 74 drops with ~4% prohability.
With 1/1000 and 60k kills, you can get 59 drops with 25% prohability.
With 1/1000 and 60k kills, you can get 49 drops with ~12% prohability.
Maybe you should keep your current setting of DropRatio and do some more testing:
(Please turn off random drops for your monster and use a lv char = mob lv without any bonus.)
1: 10x use /zoe2 with at least 30k mobs and collect all item drops of your dropgroup
Afterwards you can calculate the mean amount of drops for your given DropRatio. You can compare the result to the DropRatio and try to match it. E.g. if the drop of items is 10 times higher, set the DropRatio to 1/10 of the value and test again.
Like that you should be able to calculate a factor f of: Item dropped = DropRatio x f
2: Continue with testing by using the same factor and change other parameters (server.cfg x10, using different char lv, type of mob and so on).
I've attached the simulator, if you want to test it. Adjust the yellow cells to whatever you need. I've put a limit of 1k sims and 10m sims * kills, but you change it in the code.