Not sure, but here's my take...
On one of those Travel Channel shows all about Las Vegas, I recall the narrator said that "insiders" peg the odds of hitting the big one on a the Nevada Megabucks progressive at about 50-million to one (he said there are 400 or so machines linked statewide). At 3 coins per spin, with the denomination being dollars, you'd hit the jackpot once, on average, after spending $150-million.
So, assuming those 50m:1 odds are correct, with a current top prize of $27-million, the jackpot would, on average, be responsible for returning fully 18% of all the funds deposited - and you'd get it all back just as you were completing your 50 millionth spin (on average!). Math: $150 mil/$27 mil = 18%. (Of course, along the way, you'd hit all kinds of other, albeit much smaller, payoffs.)
Remember, when the jackpot is reset it is seeded at $7-million. At this lower level, trusting the 50m:1 odds to be correct, the jackpot would pay back only 4.7% of all the funds blown up to your 50 millionth spin, where, on average, you'd be hitting the jackpot. Math $150 mil/$7 mil = 4.7%.
My take is that if you play the Megabucks game today rather than in its "just-seeded" state, you're playing a game with a 13.4% higher return. Math: 18% - 4.7% = 13.3%. If the "at-seed level" game has a payback rate of at least 87%, the game now has a slight positive expectation.
Have I made a mistake somewhere in this analysis? I would never personally try to beat a game such as Megabucks because of the horrid variance - I don't have the bankroll - but some wild teams might be trying to knock this one down as we speak.