I realize that variance is supposed to be higher -- actually MUCH higher -- on single-play VP games than multiplay at the same relative denomination. By "same relative", I mean that you would be playing a $1 5-play machine instead of a $5 1-play.
According to this article on WizardOfOdds, the variance for 5-play $1 9/6 Jacks or Better is about four times LESS than 1-play $5. (Divide the number by 5 for the 5-play variance in the article to compare them.)
However, I have played a lot of 5-play 9-6 JoB lately and have gotten absolutely killed. Some of it has been due to poor luck with royals -- basically I haven't hit one.
But I'm also running far below expectation even taking royals out of the equation. That is, I'm still doing a lot worse than expected compared to the "didn't hit a royal" return on the machine.
I've found that the big problem with multiplay VP is that it is very deal-heavy regarding how you are going to do.
That is, if you are dealt poor hands, then you are going to lose, and lose badly, barring hitting the one big draw payout (the royal).
Why? Because aside from royals, the only big hits in multiplay JoB come from made hands dealt. For example, a dealt full house at 5-play is worth 225 credits, while a drawn straight flush is only worth 250! So even drawing to the second-best hand (straight flush) is barely worth more than being dealt a full house.
Furthermore, the value of backing into semi-unlikely full house and quad draws (such as when you just hold a pair) again becomes fairly trivial, given that you will only be hitting it on one hand.
On its face, multiplay VP seems like a great variance reducer, as it's basically letting you draw at every hand several times, thus smoothing out the luck factor of drawing. However, if you look at it a different way (as explained above), you basically become destined to lose big if you aren't dealt enough of a share of high-value made hands right off the bat.