Can Math Actually Beat the Lottery? The Two Who Did
“Almost no one beats the lottery with math — but two people genuinely did, legally, by spotting structural flaws. Stefan Mandel and the Selbees.”
'You can't beat the lottery with math' is almost always true. Almost. There are two extensively documented cases — covered by major newspapers, a Harvard study, 60 Minutes, and a 2022 feature film — where people legally extracted millions from lotteries. Crucially, neither used a number-picking 'system.' Both exploited a structural flaw in how a specific game was run.
Stefan Mandel, a Romanian-born economist who later operated from Australia, reasoned that any lottery small enough to let a syndicate buy every possible combination, when the jackpot grew larger than the total cost of all those tickets, became a positive-expected-value bet. He won smaller Australian lotteries this way before his most famous operation: the 1992 Virginia state lottery, which had only about 7.06 million combinations. With the jackpot around $27 million and tickets $1 each, his investor syndicate attempted to buy every combination. They printed and delivered millions of tickets, secured roughly five million of them in time, and won the jackpot plus a large number of lower-tier prizes — a payout reported around $27.9 million.
Mandel broke no laws, but lottery operators noticed. Jurisdictions changed rules to prevent bulk printing and buying of full combination sets — limiting ticket runs, banning third-party mass purchase, or barring computer-printed tickets bought in volume. The flaw he exploited was closed by regulation, not by probability. His edge existed only because a specific game allowed total coverage cheaper than the prize.
The second case is gentler and arguably more remarkable. Jerry Selbee, a retired Michigan convenience-store owner with a math background, noticed that the Massachusetts game WinFall had a 'roll-down': when the jackpot hit a cap without a top winner, the money rolled down to lower tiers, briefly making the expected value of a ticket positive. He and his wife Marge formed 'GS Investment Strategies LLC,' recruited friends and family, and bought tickets in enormous volume during roll-down weeks for roughly nine years.
Over that period the Selbees' group bought an estimated $26 million in tickets and netted on the order of $7.75 million in profit before the game was shut down. They were not the only group doing this — a Harvard/MIT-linked syndicate ran the same exploit — and a Massachusetts inspector general's report concluded the players had broken no rules; the lottery itself had designed a game with a temporarily positive expected value. Their story became the 2022 film 'Jerry & Marge Go Large.'
Both cases share a pattern that destroys most 'beat the lottery' marketing. Neither involved predicting numbers, hot/cold analysis, dreams, or systems. Both required a game with a specific exploitable structure (cheap full coverage; a roll-down that flipped expected value positive), enormous capital, industrial-scale logistics, and the math literacy to recognize a fleeting edge — and both edges were eliminated once operators noticed.
The lesson is not 'math can beat the lottery.' It is that the only real edges in lottery history came from operational flaws in particular games, exploited briefly by well-capitalized groups, and then closed. For an ordinary player buying a few lines of a modern, properly designed game, no math changes the negative expected value. The Mandel and Selbee stories are fascinating precisely because they are the rare, documented exceptions that prove the rule. For the rest of us, the most useful thing to do is check whether your Powerball combination has actually been drawn before believing any system.
Sources