Prediction markets are already liquid, but they remain inefficient until positions become usable collateral.
There’s a thing that happens to every derivatives market, right before it gets serious.
The instrument already exists. People are already trading it. The volume is real, maybe even impressive. But the capital is inefficient — locked up, idle, waiting. And then someone builds a lending layer, and everything changes.
This is one of the most repeatable pattern in financial history.
The Pattern That’s Hard to Ignore
When equities markets matured, it wasn’t better orderbooks that unlocked institutional participation. It was margin lending, the ability to post stock as collateral and borrow against it.
Prime brokerage followed, and with it came rehypothecation, securities lending, and cross-margining. By 2024, prime brokerage borrowing had reached $2.5 trillion globally, more than double what it was in 2020. None of that was possible until someone decided that a stock position was worth something as collateral.
The repo market tells the same story in fixed income. U.S. Treasuries were already the world’s most liquid asset. Repo made them productive. The ability to post bonds overnight and receive cash is what gave the bond market its depth and allowed institutions to run the strategies that define modern rates trading.
Strip out repo and the entire architecture of government debt financing breaks down. The instrument existed first. The collateral layer made it institutional.
Crypto repeated this cycle at speed. Bitcoin spot markets were live for years before perpetuals arrived. Perps were already doing serious volume before Aave and Compound gave traders the ability to borrow against their holdings.
Each time a lending primitive appeared, the market it touched grew by an order of magnitude — not because new participants flooded in, but because existing capital could do more with itself.
The entity that builds collateral infrastructure captures the market. This has been true in every asset class. It will be true in prediction markets.
What’s Actually Happening on Polymarket
Prediction markets have graduated from novelty. The evidence is overwhelming.
Polymarket’s cumulative trading volume surpassed $9 billion in 2024, with monthly volume growing at 66.5% through the year. The 2024 U.S. election became the single most traded event in prediction market history. Open interest hit an all-time high of $510 million during the November election, with active traders reaching a peak of 314,500 in December.
Then the institutional signal came. In October 2025, ICE, the parent company of the New York Stock Exchange, wrote a $2 billion check to Polymarket, valuing it at $9 billion.
The NYSE’s parent company doesn’t make $2 billion strategic bets on gambling products. That’s not what that investment was. It was a bet on financial infrastructure.
Meanwhile, Kalshi, the regulated U.S. competitor, raised over $300 million at a $5 billion valuation, backed by Sequoia, a16z, and Paradigm.
Kalshi achieved $50 billion in annualized volume in 2025, up from just $300 million the prior year. Sequoia and Paradigm don’t back gambling products either.
The market has already decided. Prediction markets are a serious asset class. What they are not yet, is a capital-efficient one.
The Locked Capital Problem
Here’s the friction that nobody has solved:
A trader puts $500,000 into a YES position on a geopolitical event they’re highly convicted on. The contract doesn’t resolve for three months.
That capital is frozen and it can’t be used as margin elsewhere, can’t fund another trade, can’t participate in a time-sensitive opportunity that appears the following week.
The trader isn’t wrong about their position. They’re just illiquid.
This is not a feature of prediction markets. It’s a structural artifact of missing infrastructure. In every mature market, this problem has the same solution: let the position serve as collateral.
A bond trader doesn’t sit idle waiting for a Treasury to mature. They repo it and deploy the proceeds. An equity fund doesn’t liquidate longs to fund a new position, their prime broker lends against them. The capital keeps working.
Polymarket’s average open interest-to-volume ratio sits at 0.38, higher than Kalshi’s 0.29, suggesting Polymarket positions tend to remain static for longer periods.
That stickiness is a signal that people hold strong-conviction, longer-dated positions on Polymarket. Those are exactly the positions that create demand for borrowing.
High conviction, long duration, binary payoff — this is the collateral waiting to be unlocked.
Why This Is a Hard Problem Worth Solving
Borrowing against a Polymarket position isn’t straightforward, and that’s exactly why the opportunity exists.
Binary resolution creates a risk structure that standard margin models weren’t designed for. When a traditional asset declines in value, it does so continuously, giving liquidation engines time to act.
A prediction market position can go from $0.85 to $0.00 in a single oracle call. The liquidation window is zero. Haircuts, collateralization ratios, and risk models all need to be re-thought from the ground up for this specific structure.
This is an engineering problem. A real one. But engineering problems have solutions, and the economic logic on the other side is solid.
Every sophisticated trading workflow that touches prediction markets will eventually need a borrowing layer. The demand is structural, not speculative.
The Moat Builds Itself
Once a protocol becomes the canonical place where prediction market positions are posted as collateral, displacement becomes extremely expensive.
It’s not just switching costs in the narrow sense. It’s that risk models get calibrated around the protocol’s haircuts.
Counterparties build integrations assuming the collateral layer exists. Trading strategies get designed with borrowing capacity as a given. Every new workflow that plugs into the ecosystem makes the next defection less likely.
This is how CME maintains dominance in U.S. futures despite decades of better-funded competitors trying to unseat it — not because their matching engine is unbeatable, but because they sit at the center of collateral flows.
Clearing infrastructure, cross-margining, and collateral netting create lock-in that has nothing to do with the quality of the product and everything to do with the cost of rebuilding the dependencies.
The timing is right. Prediction markets have proven product-market fit. Institutional capital is arriving. The open interest is real and growing. What doesn’t exist yet is the financial infrastructure that lets that capital work at institutional scale.
That gap is a business. And the entity that fills it first will find the moat compounds long after the product ships.
Collateral Is The Moat — And Prediction Markets Are Next was originally published in DataDrivenInvestor on Medium, where people are continuing the conversation by highlighting and responding to this story.
