Mohan Srivastava and the OLG Tic-Tac-Toe Ticket
In 2003, a geological statistician living in Toronto named Mohan Srivastava scratched a tic-tac-toe lottery ticket he'd received as a gift and won a few dollars. That win was unremarkable. What came next wasn't.
While walking to a local lottery retailer to cash in his ticket, Srivastava found himself thinking about how the game must work. Scratch ticket randomness isn't truly random — the results are pre-determined at the point of printing, and the whole print run must contain a statistically predictable ratio of winners to losers. The numbers visible on the ticket before scratching aren't decorative; they're part of the ticket's underlying structure.
Then a thought hit him: if those visible numbers are generated by the same process that determines whether the ticket is a winner, they might carry a detectable signal.
They did. Srivastava noticed that on the tic-tac-toe boards printed on the front of the ticket, certain numbers appeared multiple times across the eight grids, while others appeared only once. He called these singletons. The tickets that tended to win, he found, had their tic-tac-toe rows filled with singletons — numbers that appeared just once on the entire visible face of the card.
ℹ️ The singleton method: Without scratching anything, Srivastava could scan the eight tic-tac-toe grids on the front of the ticket and mark the numbers that appeared only once. If three singletons appeared in a row on any board, the ticket was very likely a winner. He tested his hypothesis on a batch of unscratched tickets — and predicted winners with roughly 90% accuracy.
The crucial detail: he didn't exploit it. Srivastava did the math and worked out that acting on this edge wasn't worth his time. The tickets paid out small amounts, the process of sorting through piles of cards would be slow, and he had a well-paying job as a consultant. "I'm not the type to do something illegal," he later told Wired, "but I was stunned that nobody had spotted this before."
Instead, he sent a package to the Ontario Lottery and Gaming Corporation — 20 tickets he'd sorted into winners and losers without scratching — along with a letter explaining how he'd done it. OLG confirmed the flaw, pulled the game, and revamped their ticket generation process. Srivastava went back to work.
The uncomfortable implication, which Srivastava raised himself: if he found this vulnerability in a few hours of idle thinking, someone less scrupulous could have been exploiting it for years without the lottery corporation ever knowing.
Joan Ginther: Four Jackpots, $20.4 Million, One Small Town in Texas
Joan Ginther holds a PhD in mathematics from Stanford University. She also won the Texas lottery four times over 17 years, collecting a total of $20.4 million in prizes. The odds of that happening by chance have been calculated at roughly one in 18 septillion — a number so large it's essentially impossible.
The four wins:
- 1993: $5.4 million in Lotto Texas (a draw game)
- 2006: $2 million on a Holiday Millionaire scratch ticket
- 2008: $3 million on a Millions and Millions scratch ticket
- 2010: $10 million on a $140 Million Payout scratch ticket
Two of the winning scratch tickets were purchased from the same convenience store in Bishop, Texas — a small town of about 3,000 people where Ginther grew up.
The Texas Lottery Commission investigated Ginther after her fourth win and concluded there was no evidence of fraud. She had purchased the tickets legally, the games were legitimate, and the wins were declared valid. She has never publicly explained how she did it, and she's given almost no interviews.
The leading theory, advanced by investigative journalist Nathaniel Rich after extensive research, is that Ginther had reverse-engineered the distribution algorithm used to seed scratch ticket print runs. Scratch tickets within a single print run are not distributed randomly to retailers — they're shipped in sequential order from the printing facility. A sufficiently large sample of purchase data, combined with knowledge of how the ticket algorithms work, could theoretically let someone predict which stores were likely to have winners in their current shipment.
What this would mean in practice: Ginther may have bought large quantities of tickets from specific rural stores — stores whose shipment sequences put them in a zone of the print run dense with winners — rather than buying individual tickets randomly. The rural location matters: a small-town store processes fewer tickets than an urban retailer, so the sequence moves more slowly and is easier to model.
Nobody has proven this. Ginther has never confirmed it. The Texas Lottery Commission maintains her wins were legitimate. What's undeniable is that a Stanford-trained mathematician won four lottery jackpots, and the statistical explanation "extraordinary luck" strains credibility to the breaking point.
The MIT Students and the Massachusetts Cash WinFall
This one is different: it wasn't about scratch tickets, and it wasn't theorized after the fact — it was a deliberate, documented, and entirely legal exploitation of a flaw in a draw game's structure. But it belongs in this list because it shows what happens when people with quantitative training actually look at how lotteries work, rather than accepting the odds at face value.
Massachusetts ran a lottery game called Cash WinFall. Like most lottery games, it had a jackpot that rolled over when nobody won. But Cash WinFall had a quirk: when the jackpot exceeded $2 million and nobody claimed it, the prize money didn't keep rolling — it "rolled down" into the smaller prize tiers instead. On those drawings, the 3-, 4-, and 5-number prizes paid out dramatically more than usual.
A group of MIT students — forming a syndicate called Random Strategies — discovered that on roll-down drawings, the expected value of a $2 ticket could exceed $2. Buying $100,000 in tickets on the right drawing didn't guarantee a profit on any single bet, but in aggregate, across the full statistical distribution of 3-, 4-, and 5-number prizes, the math pointed firmly positive.
They started buying tickets in bulk — eventually hundreds of thousands of dollars worth — during every roll-down week. By the time the scheme was discovered and reported on, the group had accumulated nearly $8 million in winnings.
ℹ️ Was it legal? Yes. They bought tickets through normal retail channels, paid for them legitimately, and collected prizes through the standard process. Massachusetts lottery officials were aware of the group's activity and did nothing to stop it — in part because the bulk buying increased lottery revenues. The game was eventually shut down in 2012 after investigative reporting exposed the scheme, but no criminal charges were ever filed.
The MIT group wasn't alone. A retired Michigan couple, Jerry and Marge Selbee, independently discovered a near-identical roll-down mechanism in a Michigan game called Winfall and extracted over $26 million in winnings over several years. Jerry Selbee, a former cereal company statistician, figured it out on his lunch break after reading the game's rules on a gas station brochure.
What These Stories Actually Tell Us
The common thread isn't genius. It's the willingness to actually look at how the systems work, rather than taking the published odds as the whole story.
Srivastava looked at the visible numbers on a scratch ticket and asked whether they contained information about the hidden result. They did. Ginther (allegedly) looked at how print runs were distributed to retailers and asked whether that distribution was predictable. It appears it was. The MIT students looked at Cash WinFall's payout rules and asked whether there were conditions under which expected value turned positive. There were.
None of these were secret. The information Srivastava used was printed on the face of every ticket. The distribution mechanics Ginther may have exploited were documented in how scratch tickets are shipped. The Cash WinFall rules were public. The edge existed because almost nobody bothered to look carefully at data that was sitting in plain sight.
That's the same principle behind Scratchers Edge. We're not predicting winning tickets — that's not possible for games designed properly. But the prize remaining data that WCLC, OLG, BCLC, and Loto-Quebec publish is real information about the current state of each game's prize pool. Most buyers never see it. We make it easy to check before you buy.
⚠️ Responsible gambling reminder: Scratch tickets are a form of gambling, and no amount of analysis changes the fundamental math — the house wins in aggregate. Information about prize availability helps you avoid games that are already picked clean, but it doesn't create positive expected value. Only spend what you can afford to lose. If gambling is affecting you or someone you care about, visit responsiblegambling.org or call 1-800-522-4700.
Sources
- Cracking the Scratch Lottery Code — Wired, Jonah Lehrer's profile of Mohan Srivastava
- How a Statistician Beat Scratch Lottery Tickets — Gizmodo
- The Stanford PhD Who Allegedly Gamed the Texas Lottery — Inc.
- How MIT Students Won $8 Million in the Massachusetts Lottery — TIME
- How a Couple Won $26 Million Using a Lottery Loophole — Yahoo Finance