Ask anybody why gamblers keep gambling and you'll get the same theory: they remember the wins and forget the losses. It sounds right.
Somebody actually tested that in 1983, and it came back backwards.
The gamblers remembered their losses better. Three weeks later, in a recall test, the losses were the ones that stuck. They'd also spent more time chewing on those losses than on any win, and what they were doing with all that attention was arguing. The kicker took a bad bounce, the starter was hurt and nobody mentioned it, the refs blew that one.
The wins needed no explanation. A win was proof you were good at this.
So the ledger in a gambler's head runs two sets of books. Wins count as evidence, losses get explained away one at a time, and they'll tell you with total sincerity that they're up.
The researcher was Thomas Gilovich, and memory wasn't what he was after. He wanted to know why people keep going.
Prediction markets handed a few million people the same problem with a tax bill attached. Kalshi's annualized volume tripled in six months, from $52 billion to $178 billion. The World Cup alone pushed about $27 billion through roughly 3 million users, double what the company projected. Sports is 80% of the whole damn thing. Every one of those accounts shows a balance, and a balance tells you nothing about what you owe in April.
A trader with hundreds of fills needs cost basis per contract, fee attribution, and an open-position inventory at year end. Kalshi's own docs say its figures aren't tax advice and your return is your problem. Tax pros are arguing four different treatments and the IRS hasn't said a word.
So build the boring thing: a tax and portfolio ledger for prediction market traders. Read-only API sync, penny-perfect reconciliation against the exchange statement, an export a CPA will actually sign. A thousand traders at $90 a year is $90K ARR, five thousand is $450K. Half a dozen teams are already racing toward free dashboards, and free is the easy half. Nobody's built the one an accountant trusts.
Read the full playbook here:
Kalshi's $178 billion in volume minted millions of new prediction market traders. The exchange shows a balance; nobody turns fills, fees, and cost basis into records a CPA will sign.
From the Vault:
Okara runs AI marketing for 120,000 businesses at $99 a month. Small law firms can't use it. Bar filing rules, disclosures, and attorney approval turn that gap into $499 software.
Housecall Pro just verticalized into HVAC, plumbing, and electrical. Septic pumping got skipped, leaving 7,717 operators with three-to-five-year service cycles and no software built for the paperwork.
