Prediction markets have indeed been popular over the past two years. The trading volume looks impressive, but after taking a closer look, I realized—these platforms are actually trapped in a dangerous local optimum.

Liquidity issues are the core problem. In seemingly lively data, most trading volume is concentrated in hot markets like the US elections and Federal Reserve, while other markets have absurdly wide bid-ask spreads, and no one wants to trade. Market makers are losing money, incentives and subsidies are burning cash, fake trades are rampant—these are red flags indicating insufficient organic growth.

Deeper issues lie in the product design itself. Pure binary structures can't compete with perpetual contracts, with capital efficiency being shockingly low, retail user experience poor, and market makers unable to profit. Jump risk, lack of hedging mechanisms, no leverage support—these are not minor bugs but fundamental architectural flaws.

Kalshi and Polymarket are like BlackBerry back in the day—optimized to the extreme in the wrong track. They use liquidity rewards and trading rebates, seemingly "fixing" problems, but in reality just burning money to maintain a false appearance. Real solutions require a radical redesign of the protocol, introducing mechanisms similar to perpetual contracts.

Ironically, the users who should be most attracted—perpetual contract traders—are precisely the ones who see through these issues first. But their voices are often drowned out by the trading volume figures.

The endgame of prediction markets shouldn't stop here. To truly realize the vision of "everything can be traded," the current model must be completely overhauled. Otherwise, it will only sink deeper.