Superfecta probability tables or a simulator would come in handy
here. My own simulator stops when the Trifecta is decided, but
could be modified to handle the supers. Outcome tables would help
you find cases where consolations are more likely (but without
letting the chance get too small).
Suppose you believe you have the trifecta (let's say x-y-x) nailed.
Now, an x-y-z-ALL Superfecta wheel (5 bets) will win you either
a straight Superfecta (perhaps shared) or 5 consolation winners.
I'd suggest $1 bets.
It's conceivable you could win the entire pool with your
5 consolations, with less tax consequences.
But, of course, you still need x-y-z 
