A Second Identity: Prediction Markets as Financial Derivatives

Key Takeaways

• Prediction markets have crossed into institutional territory, yet the capital sitting inside them remains almost entirely idle - unable to be borrowed against, leveraged, or deployed elsewhere.
• Jump risk remains a structural problem that has made leverage nearly impossible to introduce, as capital providers can be wiped out instantly with no window for liquidations.
• Difficulty surrounding pricing and valuation, alongside historical regulatory uncertainty have caused the slowed adoption of prediction market positions within existing lending protocols.
• New approaches to fee models, developments at the product layer, on-chain representation of prediction market positions, and Kalshi's CFTC legal victory have significantly lowered barriers to DeFi composability, and opened up room for innovation and further development

Having generated over $6b in weekly notional volume across 8 of the past 15 weeks, prediction markets have cemented themselves as a legitimate financial sector, a far cry from their origins as a crypto-native experiment into crowd-sourced forecasting. Kalshi and Polymarket - the sector’s two largest players, now stand at valuations of $22b and $12 billion respectively, having closed successful fundraising rounds from blue-chip investors. Talks of a fresh funding round have also been circulating, potentially valuing the latter at $15 billion. 

Yet for all this momentum, a fundamental inefficiency persists. A sophisticated trader sitting on $500k worth of prediction market positions faces the stark reality that his capital is almost completely inert and sitting idle. It cannot be borrowed against, leveraged, or deployed elsewhere while positions remain open. While users are able to borrow up to 90% on their standard crypto assets and take up to 100x leverage on high conviction bets, this is not the reality for prediction market positions. This efficiency gap is jarring, and it stems from a few long-standing structural limitations.

Structural Limitations

The first is design-related. Prediction market positions have been largely binary by nature, acting as contracts that pay out either $1 or $0 at resolution, with no in between. This makes them difficult to value as collateral using the same frameworks and guidelines applied to assets like ETH or BTC, where liquidation engines can gradually unwind positions as prices move over time. The immense and unpredictable volatility of their value prior to resolution further complicates this, making it understandably difficult for lenders to extend sizable credit against them if any at all. 

The second is technical. Until Kalshi’s launch of tokenised prediction market positions on Solana in December 2025, these positions have remained locked within centralised platforms and proxy wallets, inaccessible to users and external smart contracts. As such, DeFi protocols lacked any means of integrating the positions, regardless of their interest in doing so. 

Together, these two major constraints have kept prediction market capital effectively walled off from the broader DeFi ecosystem, limiting their potential as financial primitives. This however, is beginning to change. New layers of infrastructure and applications are being built directly on top of these markets - leveraged prediction markets, lending markets accepting positions as collateral, and more. 

Raising the Ceiling

Prediction markets have long been subjected to a built-in leverage ceiling, essentially behaving like a spot trading market. On Polymarket for example, each position/share is capped at a $1 payout at resolution, and while traders may wish to increase their exposure, their only means to do so would be to purchase an additional share at its current market value. This creates a frustrating scenario for traders where profitability is extremely limited once a market already trades at a high probability (e.g. 90% on a binary outcome), and simply purchasing more shares offers diminishing returns for massive capital commitment. For a sector that is increasingly attracting more institutional players like hedge funds, quant desks, etc, this spot-like design is extremely limiting. 

Jump Risk

A reason leverage has been proven so difficult to introduce into prediction markets is the presence of jump risk, an extension of the design limitation described in the above segment. In equity or even other crypto asset markets, brokers extending margin are protected by the continuity of price movement - if a position moves against a trader on margin, there is sufficient time to issue a margin call or or gradually liquidate before losses become unrecoverable. On the other hand, prediction markets do not offer such a window. A single news event or play during a sports match could send the value of a position close to 0 in an instant, leaving capital providers and financiers with no time to act, and prevent losses. 

New Fee Models

To introduce leverage, capital providers must be sufficiently compensated for a risk they cannot hedge in real time. Messari analyst Kaleb Rasmussen suggests an epoch-based fee model as the most viable path forward, where capital providers price risk over short rolling windows rather than across the entire life of a position, similar to how the funding rate mechanism in perps works. Within each epoch, the fee will account for two sources of loss: gradual price drift towards the liquidation barrier, and sudden jumps that cross it instantly. By repricing at each rollover, capital providers avoid the near-impossible task of forecasting jump risk and volatility months in advance, instead committing only to the next short window. Rasmussen further proposes pairing this with an auction-based rebate system at the platform level, designed to capture the arbitrage value generated by sudden price shocks and redistribute it back to capital providers, further offsetting potential losses.

Product Layer Alternative

An alternative approach to enabling leverage would be to sidestep the fee model problem entirely, and address the issue at the product layer. Perpetual futures on prediction outcomes wrap a binary event in a derivatives structure that traders are already familiar with. Rather than buying a YES or NO share and waiting for resolution, a trader takes a leveraged position on the probability itself, with funding rates keeping the perpetual price anchored to the underlying market. The result preserves the informational core of prediction markets while unlocking the capital efficiency and flexibility that institutional participants expect, without requiring a capital provider to solve for jump risk pricing on the underlying spot market. This is similar to what dYdX implemented with their TRUMPWIN market for the 2024 US Elections.

The appeal for sophisticated traders is straightforward. A trader with strong conviction on a market already pricing at 80 cents has limited upside on a spot position. The same conviction expressed through a leveraged perpetual opens up meaningfully asymmetric returns, with liquidation risk that is at least priceable within a known derivatives framework. It is worth noting that this approach will require deep liquidity and tight spreads just as regular perps do.

From Bet to Balance Sheet

Beyond leverage, a second and equally significant opportunity lies in unlocking prediction market positions as collateral. Currently, traders who hold positions in prediction markets have effectively pre-committed capital to an outcome they believe will happen. While those positions have real, and quantifiable value, they cannot be used to access liquidity in a manner that almost every other financial asset is able to. It simply remains locked and non-yield generating until the market is resolved, or the trader decides to sell his positions prematurely.

Liquidations & Regulatory Risk

The reluctance of existing lending protocols to accommodate prediction market positions stems from two main concerns. The first is design-related, as mentioned in the earlier segment. Unlike equities or crypto assets that trade continuously, prediction markets have a defined end date. The lack of a continuous price value complicates the application of existing loan-to-value frameworks, making it difficult to determine a safe collateralisation ratio without either severely undercollateralising the lender or making the loan economically unattractive to the borrower. A position's entire value potentially collapsing to zero instantly at resolution, also leaves lenders with no ability to gradually unwind exposure if a loan goes underwater, and the platform potentially with bad debt if liquidation cannot be executed in time. On top of this is the presence of regulatory risk. For most of their existence, prediction markets have operated within a legal grey area, with regulators questioning whether event contracts constitute legitimate financial instruments or are simply gambling repackaged. For risk-averse lending protocols, onboarding prediction market positions as collateral assets carrying that classification pose unnecessary legal risks.

Weakening of Barriers

Both of these barriers have weakened considerably in the past few months. Kalshi's legal victory over the CFTC settled the classification question at a federal level, establishing that event contracts are indeed regulated derivatives, a distinction that could matter enormously for lending protocols. Furthermore, Kalshi's launch of tokenised positions on Solana in December 2025 gave prediction market positions an on-chain representation for the first time, making them assets that smart contracts can actually hold, verify, and act against as collateral. While frameworks on valuing these have yet to be completely figured out, this change opens up experimentation with ideas such as probability-weighted valuations, time-decay, and other implementations of oracle infrastructure.

Implementations & Developments

What this enables in practice is still being actively designed. The most straightforward implementation mirrors conventional DeFi lending where a user deposits tokenised prediction market positions as collateral and borrows stablecoins against them, with the loan-to-value ratio calibrated to the position's current probability, time to resolution, and historical volatility. More sophisticated approaches include collateralising a basket of uncorrelated predictions rather than a single one, using portfolio-level health computations to sidestep the binary valuation problem and decreasing the risk of liquidation. This logic is similar to what prime brokerages use, where the entire book is assessed, and not individual assets. In the process of infrastructure being built, the NFT lending market serves as a cautionary tale. What initially started as rapid growth was quickly followed by collapse when underlying valuation models proved insufficient in handling the price changes of the underlying assets. Getting the collateral framework right before scaling should be the appropriate approach.

Conclusion

Prediction markets have long been justified on a single premise: that crowds, given the right financial incentives, price uncertain outcomes better than experts. That premise has proven sufficiently compelling to sustain years of experimentation, and sufficiently validated by recent volume and valuation figures to attract serious institutional capital. This has triggered a monumental shift from forecasting and its informational utility, towards becoming a financial derivative in itself.

The developments outlined in this piece signal that prediction markets are already in the early stages of earning that second identity. The tokenisation of positions, the foundations being laid for leverage, the emerging frameworks for collateral lending all represent key building blocks of a derivatives market, one where traders can express conviction with appropriate capital efficiency, access liquidity without prematurely closing positions, and construct sophisticated strategies around binary outcomes in much the same way they do around any other underlying asset. Parallels to earlier DeFi maturation cycles are clear - spot markets came first, then perpetuals, then structured lending, then composability across all three, alongside regulatory shifts and acceptance.

The transition for prediction markets is already well underway.