LotteryIn 2003, a retired Michigan convenience store owner named Jerry Selbee picked up a marketing brochure for a new lottery game called Winfall. He read it, and within three minutes, spotted something the state lottery apparently hadn't fully accounted for: the game had a rule that made certain weeks, and only certain weeks, a genuinely good bet. He and his wife Marge went on to make more than $26 million from it — legally, and without ever hitting the jackpot itself.
The Mechanism: a "Roll-Down" Is the Opposite of a Rollover
Powerball and Mega Millions both use a rollover: if nobody wins the jackpot, it simply grows and carries into the next drawing, which is why headline jackpots can climb into the hundreds of millions. Winfall (Michigan) and its sister game Cash WinFall (Massachusetts) worked differently. Once the jackpot reached a set cap — $5 million in Michigan, a lower $2 million in Massachusetts — and still had no winner, the money didn't roll forward. It rolled down into the lower prize tiers instead, matching 3, 4, or 5 numbers, dramatically inflating what those smaller prizes paid out. See how lottery jackpots grow and roll over for how the normal version of this works.
The Specific Math Selbee Actually Ran
Jerry Selbee described a concrete example of the calculation himself: spending $1,100 on 1,100 tickets during a roll-down week, he could expect roughly one 4-number winner (paying around $1,000) and around 18-19 three-number winners (paying around $900 total) — an expected return of about $1,900 on an $1,100 outlay, a real, positive expected profit of roughly $800. The key detail is that this math only worked out favorably during roll-down weeks specifically — buying tickets in an ordinary week, with no roll-down happening, was just a normal, losing bet like any other lottery ticket.
From a Few Thousand Dollars to Tens of Millions
The Selbees scaled from Jerry's original few-thousand-dollar tests into a formal operation, forming a corporation and selling $500 shares to dozens of friends and family so they could buy tickets by the hundreds of thousands on roll-down weeks. Separately, an MIT student doing an independent study project discovered the exact same mechanism in the Massachusetts version of the game and built a similar operation with fellow students. Both groups made real, substantial profits over years of consistently playing only the mathematically favorable weeks — not through luck, and not by ever hitting the actual jackpot.
Why This Wasn't Cheating
State investigations into both groups, prompted by media coverage once the scale of their winnings became public, found no rule-breaking or fraud. Every ticket was purchased through completely normal, published channels, following the game's own publicly stated rules. The "loophole" wasn't a bug or a hack — it was a real, calculable mathematical consequence of how the roll-down rule was designed, one the lottery's own officials apparently hadn't fully modeled out themselves. The game was eventually discontinued once the scale of coordinated roll-down betting became widely known and competitive, since it was cutting into the lottery's own margins.
The Takeaway
This wasn't a system for picking winning numbers, and it doesn't apply to how Powerball or Mega Millions work today — those games use a standard rollover, not a roll-down, so this specific mechanism simply doesn't exist in them. What makes the story genuinely instructive is what it actually was: a real, verifiable case where a specific game's rules briefly created a genuine positive expected value on certain purchases, and a small number of people did the math carefully enough to notice before anyone else did. See the expected return rate calculator to see what a ticket's real expected value looks like in today's games, where no comparable roll-down mechanism exists. For a very different real case built on brute-force math instead of a rule quirk, see the man who bought every lottery combination.