The rise of prediction markets, and the account plumbing underneath

A prediction market is a place to take a chance on what will happen. Will a central bank cut rates this month, who wins the World Cup final, what the price of oil is in May. You buy a contract that pays out a fixed amount if you're right and nothing if you're wrong. The price of the contract moves with the crowd's confidence, so a contract trading at 70 cents means the market thinks the outcome is about 70% likely. When the event happens, the contract settles: winners get paid, losers get zero.

That's it. The clever part, the reason economists and journalists pay attention, is that these prices turn out to be decent forecasts. A market pricing an event at 70% is often more accurate than a pundit saying it'll probably happen.

Prediction markets have grown fast over the past two years. Combined monthly trading volume rose from under USD 5 billion in September 2025 to about USD 24 billion by April 2026. By June, a major sporting event pushed that figure to USD 44.8 billion across the main platforms, a 75% jump in a single month. The number of people using these markets grew just as quickly: unique accounts nearly tripled in the six months to February 2026, reaching 840,000.

Almost all the attention goes to the forecasting. Can a market really predict an election better than a poll? The more practical question gets skipped. Where does the money go while the market is open?

A prediction market is a deposit business

Think about what has to happen for a single transaction to work. You put money into an account. That balance sits there while the event you've bet on plays out, which might be minutes or might be months. When the result comes in, your balance goes up if you won and down if you lost, and you want to withdraw the money quickly. Multiply that by hundreds of thousands of people, all funding accounts, placing bets, and cashing out at different times, and you have the core problem these platforms actually solve.

It's a deposit problem. Money comes in, it has to be held accurately and safely, and it has to move out fast the moment an event resolves. The forecasting is what users notice. The account holding their money is what has to work every single time.

This is the layer that decides whether a platform can grow. When a big event lands, volume can jump 80% in a month. Every one of those transactions is a balance that has to be tracked in real time, matched against the outcome, and paid out with no overnight delay. Get that wrong and the platform loses trust immediately, because the one thing a user will not forgive is money that's slow to come back or wrong when it does. Reviewers who rank these platforms put speed and reliability of cash-out ahead of how many markets are on offer or what the fees are.

What the account layer actually does

A prediction market needs the same building blocks as a savings bank. It has to open an account in under a minute, because a user who can't fund and start quickly gives up. It has to track every balance the instant it changes, because a resolved trasnaction pays out immediately and there's no room to wait for an end-of-day tally. It has to handle more than one currency at once, since money funded in one form and settled in another are just two views of the same balance. And it has to keep a complete, tamper-proof record of every movement, because anyone holding customer money has to prove where it is.

Holding balances this way is what deposit management software is for: running the deposit product lifecycle end to end, from account creation and real-time balance tracking through interest accrual and the compliance controls that come with customer funds. The same infrastructure underpins savings accounts, term deposits, and demand deposits, with balances updated the moment money moves rather than in nightly batches. A prediction market platform doesn't sell a savings account, but it needs almost everything that sits under one: fast onboarding, instant posting, multi-currency balances, and an auditable trail of every dollar.

Most platforms don't get this right the first time, and the early ones didn't survive it. The first wave of prediction markets a decade ago forced users to run their own technical setup, and a single transaction could cost more than USD 50 in fees during busy periods. The forecasting idea was fine. Funding an account and getting paid was painful enough that the users never came. The platforms that broke through did it by hiding the plumbing: sign-up in about a minute, and transaction fees made invisible to the user.

Why this matters beyond predictions

The lesson generalises to anything that holds customer money. A wallet, a savings app, a brokerage, a loyalty scheme with a cash balance. The product people talk about sits on top, but the thing that has to work is the account underneath: open it fast, track it in real time, hold it safely, pay it out on demand, and keep a clean record the whole way through.

Serious money has already noticed. In the first half of 2026, prediction markets were the single most-funded category in crypto, drawing USD 1.85 billion of the USD 7.1 billion raised across the top ten categories. That capital isn't buying better forecasting math. It's buying the ability to hold hundreds of thousands of accounts and settle billions in volume, fast enough and cleanly enough to keep both users and regulators satisfied.

The forecast is the part everyone talks about. The account is the part that has to hold.