I still don't agree :-)
Let me fine tune the problem to reduce ambiguity.
First, let's assume that 8-deck provides more liberal
rules (DAS and late surrender) so that the off-the-top
house edge is the same for 2-deck and 8-deck. In fact,
this is the case in real life.
Second, the minimum bet is the same for 2-deck and 8-deck,
i.e., 1 unit. Thus the 8-deck has higher maximum bet.
Third, the bankroll is the same for both cases. In fact,
to calculate variance, this condition is not needed.
(Be careful, please don't use ROR in your reasoning.
If you think that ROR is higher for 8-deck, you have
admitted that its variance is higher, because they
both have the same expectation.)
From my original post, we have agreed that the overall
expectation is the same for 2-deck and 8-deck.
One more thing is that the average bet is also approximately
the same for 2-deck and 8-deck, i.e., about 2 units.
Now, could you rethink the problem?