@Eli5DeFi: Kalshi, Polymarket, and Hyperliquid's HIP-4 prediction markets

$23.5B in prediction market notional traded in March 2026. In a single month. @Kalshi drove $13B. @Polymarket followed with $10.5B. And now @HyperliquidX’s HIP-4 could eclipse both, living up to its name: The House of All Finance. Here’s our comparative analysis. ⤵ — ➠ Kalshi - The Regulated Exchange Kalshi is what you get when prediction markets go through the U.S. regulatory process and come out the other side. ▸ You sign up, verify your identity (KYC required), deposit real USD, and trade yes/no contracts on events like elections, Fed rate decisions, or sports outcomes. ▸ Each contract pays $1 if the event happens, and $0 if it doesn’t. You buy at the current implied probability, then sell before expiry or hold to settlement. ▸ Kalshi resolves outcomes using predefined reference sources. Fast. Centralized. Reliable. ▸ Fees are straightforward: roughly $0.0175–$0.02 per contract near 50¢ probability, dropping toward zero at the extremes. ▸ The constraint is real: Kalshi can’t list markets on war, death, or anything the CFTC considers off-limits. Its market catalog is narrower by design. ▸ What it does well: $13B notional in March 2026. The largest regulated volume. Institutional trust. Fiat on-ramps. If compliance matters to you, there’s no alternative. — ➠ Polymarket - The Crypto-Native Market Polymarket runs on Polygon but plans to migrate to its own layer soon. Anyone with USDC (soon Polymarket USD) and a crypto wallet can trade, as long as they’re not in blocked countries. ▸ Connect a wallet and buy YES/NO shares on any listed market. Share price = implied probability (60¢ = 60% chance). ▸ No KYC for global users. No geographic restrictions outside limited U.S. access. ▸ Anyone can create a market. That’s the power, and also the risk. ▸ Settlement goes through UMA’s Optimistic Oracle, the Markets Team, and Chainlink Oracle: a proposer submits the resolution, there’s a dispute window, and if contested it goes to a community vote. It works most of the time, but can take hours to days when outcomes are genuinely ambiguous. ▸ The fee structure is more complex. Fees are parabolic: near-zero at extreme prices ($0.01/$0.99), but steep in the mid-range. ▸ At 50% probability, the round-trip cost hits 1.80% on Crypto markets and 1.50% on Sports. Geopolitics is the only zero-fee category. Finance markets have the highest maker rebate at 50%. ▸ What it does well: $10.5B in March 2026 volume. Broadest market variety on earth. Permissionless. Transparent on-chain trades. The go-to platform for global, crypto-native event trading. — ➠ HIP-4 - The Composable Primitive HIP-4 is Hyperliquid’s native outcome contract system, built directly into the same trading engine that handles $178B+ in monthly perpetual volume. As of April 2026, it’s still on testnet, but the architecture is worth understanding now. ▸ Contracts trade between 0 and 1. Price = implied probability. At expiry, they settle to 1 (event happens) or 0 (it doesn’t). ▸ You post 100% of your maximum possible loss upfront. No leverage. No margin calls. No liquidation risk, ever. ▸ Settlement is on-chain and deterministic, using curated reference data, resolving in ~0.2 seconds on HyperL1. No dispute window. Here is the part that changes the math on everything else: ▸ Every position you hold on Hyperliquid: spot, perpetuals, and now outcome contracts shares one unified margin account. ▸ The system automatically recognizes offsetting risk. If you hold a short ETH perp and a YES outcome contract on “ETH rallies above $X,” those positions reduce each other’s margin requirements automatically. ▸ Your outcome position isn’t dead capital. It functions as collateral across your broader portfolio. ▸ Every other prediction market silos your capital. Your $500 in Polymarket is invisible to the rest of your trades. On HIP-4, that same position stays alive and productive inside your full account. ▸ Fees follow Hyperliquid’s volume-tiered structure: very low at base (0.045% taker / 0.015% maker), compressing further for active traders. ▸ Fees apply only on close or settlement, not on open. Builders who deploy permissionless markets in Phase 2 earn up to 50% of the fee revenue generated. ▸ Phase 2 permissionless deployment requires staking 1,000,000 $HYPE tokens as a bond. That creates three simultaneous effects: a supply sink for HYPE, aligned builder incentives, and a filter against low-quality market creation, without central gatekeeping. — ➠ Bottom Line Kalshi wins for regulated, fiat, U.S.-friendly trading with strong liquidity in politics/sports. It is the most “institutional-grade” but constrained by rules. Polymarket dominates crypto-native volume and variety globally. It offers the most open market creation today but faces oracle delays and fee friction in high-uncertainty events. HIP-4 is the most innovative structurally. It turns prediction markets into a composable primitive inside the world’s largest perp DEX. The unified margin + no-liquidation design solves the biggest pain point of traditional prediction platforms (dead capital). Once on mainnet with permissionless markets, it could pull high-volume DeFi traders who already trade billions in perps on Hyperliquid.