Ippei Mizuhara: 19,000 bets, a 12% loss rate, and the bankroll that was never his

The numbers in the court record

Most gambling stories arrive as anecdotes. This one arrived as a federal complaint with a ledger attached, because the bookmaker kept one and investigators seized it. Ippei Mizuhara had been Shohei Ohtani’s interpreter since the player’s arrival in Major League Baseball in 2018, sat beside him at every press conference, and had access to his bank accounts. He met an illegal bookmaker at a San Diego poker game in late 2021 and began betting on credit.

Figures from the US Attorney’s Office (Central District of California) complaint and plea agreement as reported by iGaming Business and CBS Sports. He did not bet on baseball; MLB’s investigation treated Ohtani as the victim.

Two of these numbers do the work. The first is the pace: 25 bets a day for over two years is not a hobby, it is a second job with negative pay. The second is the ratio: $40.7 million lost on roughly $325 million wagered is 12.5% of handle, and that figure says something precise about how he was betting.

What 12.5% means

A sports book prices a coin-flip game at −110 on each side: risk $110 to win $100. A bettor who picks winners exactly half the time loses the vig and nothing else — about 4.5% of every dollar wagered. That is the book’s edge, the sports equivalent of a casino’s house edge, and it is the most a 50% bettor can lose in the long run.

Mizuhara lost 12.5%, nearly three times the vig. At standard prices that only happens one way: his picks were winning well under half the time. Solve for the win rate that produces a 12.5% loss on −110 bets and you get about 46%. The break-even rate at those odds is 52.4%. He was not an unlucky sharp; over 19,000 bets, luck has nothing left to say. He was a losing bettor at scale, and the scale is what turned a bad habit into a federal case.

Expected loss = money wagered × edge. The first three rows are what the same volume costs a player who is merely paying the house; the last is what it cost a player who was also losing the picks.

The bankroll that was never his

The loss rate explains the hole. It does not explain why the hole reached $40 million, because a bettor who loses 12.5% of handle runs out of money long before that — unless somebody keeps refilling it. Two things did. The bookmaker extended credit, so losses became debts rather than an empty balance, and the debts were settled from an account that was not Mizuhara’s. The complaint describes him calling the bank, identifying himself as Ohtani, and authorising wire transfers to the bookmaker’s associates: about $17 million over roughly two and a half years, plus winnings routed to his own account.

This is the mechanism every ruin story shares, from the Monte Carlo streak to the martingale: the maths of a losing game is survivable only while the stakes are bounded by what you have. Credit removes the bound. Risk of ruin is a calculation about a fixed bankroll; with an elastic one, ruin is not a probability but a schedule.

• 19,000 bets is 19,000 decisions, each small against the total. No single bet looks like a disaster, which is why the gambler’s fallacy — the next one evens it out — survives contact with a ledger that says otherwise.
• Credit converts a loss into a loan. A prepaid balance stops the session when it reaches zero; a credit line stops it when someone else says so.
• The win total is the trap. $142 million of winning bets feels like a record of skill. It is the numerator of a 46% hit rate.

Betkyo has no credit line and no way to bet money that is not already in the balance; that is a design fact, not a virtue claim, and it cuts both ways — it also means nobody will lend you the next bet. The responsible-play toolkit covers the tools that bound a session on purpose: deposit limits, loss limits, time-outs. They exist because the arithmetic above does not care whose money it is.