The House Always Wins — Until Someone Pulls the Three Jobs Apart

When DraftKings limited my colleague's account last year, they didn't call it banning a winning customer. They called it "risk management." He'd made the mistake of being consistently right about player props in the NBA. His reward? A polite email and a $50 max bet limit. The house had done its job: taken the other side of his action, decided he'd won too often, and exercised its right to protect the business model.

I thought about him when a16z published their explainer on prediction markets this week. Clean writeup — explained how these markets turn future beliefs into live prices, how a contract trading at 50 cents implies 50/50 odds, how your money moves the number closer to truth. But they glossed over the part that matters most to anyone who's ever placed a bet or reconciled a brokerage statement: why you can trust the payout when you can't trust the institution holding your money.

Because here's what we don't say out loud about sportsbooks, brokerages, and most financial intermediaries: they're your counterparty, your referee, and your bank — all at once. And those three jobs create a conflict of interest so fundamental we've just learned to live with it.

The Three Jobs Problem

Walk through what happens when you place a $100 bet on the Packers.

Job One: Counterparty. The sportsbook takes the other side. They win when you lose. Their risk desk adjusts the line to manage exposure. You're not betting against other fans — you're betting against the house.

Job Two: Referee. The sportsbook decides whether you won. Yes, the Packers either covered or they didn't, and the result seems objective. But edge cases happen. What if the game's suspended? What if there's a scoring error later corrected? The same entity that profits from you losing also grades your bet.

Job Three: Bank. The sportsbook holds your $100. And their $100. And everyone else's money. If they go under, or if they decide you've won too consistently, or if they just change their terms of service — you're holding a claim against an institution, not cash in your pocket.

Three jobs, one entity, every structural incentive pointed against you. And we accept this because, well, what's the alternative? Regulation helps. Licensing helps. Reputation helps. But the underlying architecture remains: trust us.

I've watched this movie before. Different industry, same script.

The Last Time We Pulled Jobs Apart

In 2002, I was working with a regional exchange that was still doing some trades via phone. A client wanted to move a block of shares. The broker on the other end of the line took the order, confirmed the price, executed the trade, and settled it through his firm's clearing operation. Three jobs: counterparty, referee, bank. When someone questioned the execution price a week later, guess who arbitrated the dispute? The same firm that had profited from the spread.

Then electronic trading and central clearinghouses pulled those jobs apart. The exchange became the neutral venue. The clearinghouse became the neutral settler. The broker's job shrank to just: find the other side. It didn't eliminate conflicts overnight — we still argue about payment for order flow and maker-taker fees — but it broke the fundamental three-jobs problem. You could audit the tape. You could see the clearing process. The referee was no longer also the counterparty.

Nobody talks about that transition anymore because it worked. The plumbing got boring. Which is exactly the point.

What Blockchain Actually Fixes Here

This is where prediction markets start to look less like gambling and more like infrastructure.

Take Polymarket. You want to bet that a specific bill passes Congress. The platform doesn't take the other side of your bet — other traders do. The market sets the odds through supply and demand, not a risk desk trying to balance its book. Counterparty: separated.

When the bill either passes or fails, the resolution isn't Polymarket's call. There's a public resolution process using oracles that anyone can audit and challenge. The mechanism is transparent. Polymarket doesn't grade its own homework. Referee: separated.

Your money sits in a smart contract — code that holds the funds and pays out automatically when the condition resolves. Polymarket can't run off with the float. They can't selectively limit winners. The contract executes. Bank: separated.

The odds are credible because the entity showing you the price isn't the same entity that profits from you being wrong.

This isn't some crypto-native magic trick. It's the same architectural fix we applied to securities trading twenty years ago: when you can't trust the institution, you build a neutral mechanism and make it auditable. Blockchain just makes the mechanism code instead of a regulated clearinghouse. The "trust" moves from trusting an entity to verifying a process.

And before you ask: yes, there are still trust points. Oracle manipulation is real. Smart contract bugs are real. Regulatory uncertainty is real. I'm not arguing prediction markets are perfect — I'm arguing they've unbundled a conflict of interest that's so old we forgot it was a choice.

The Uncomfortable Question for Finance Professionals

Here's where I keep landing when I walk clients through this: if you can run a credible market without a trusted intermediary holding all three jobs, where else does this pattern apply?

Your escrow company: counterparty, referee, bank.

Your payment processor: same.

Your clearing firm: we fixed part of this already, but wallet custody still bundles bank and counterparty risk.

I've spent twenty years in rooms where someone says "We need a trusted third party" and everyone nods because, obviously, how else would you do it? The assumption was so embedded we stopped seeing it as an assumption. Blockchain is the first technology that lets you pull those jobs apart without requiring a new regulated institution in the middle.

Which makes this the question to sit with: if your client's money is sitting with an institution that's simultaneously their counterparty, their referee, and their bank — and a decentralized alternative exists that separates those roles — how long before "we've always done it this way" stops being an acceptable answer?

What to Do Monday Morning

If you're auditing a financial services client, ask this: "Walk me through what happens if your top ten most profitable customers all try to withdraw their funds this week. Who decides whether they can?" If the answer involves discretion, limits, or "risk management" — you're looking at the three jobs problem.

If you're advising a client exploring blockchain use cases, skip the "efficiency" and "transparency" pitches. Those are nice-to-haves. Start here: "Show me where you're acting as counterparty, referee, and bank simultaneously. That's your unbundling opportunity."

And if you're trying to explain prediction markets to a skeptical CFO, don't lead with "decentralization" or "trustlessness." Lead with my colleague who got banned for winning. Then ask: "What if the house couldn't ban the smart money — and the odds actually meant something as a result?"

The railroad's already here. The question isn't whether intermediaries will get unbundled. The question is whether you'll recognize the pattern before your clients do.

---

I'm working with clients navigating blockchain's intersection with traditional finance — particularly where "trusted intermediary" assumptions are getting stress-tested. If you're seeing the three jobs problem in your world and want to pressure-test the alternatives, I'm at [email protected].