Basis Trading and HIP-3 Spread Capture

Grossman and Stiglitz showed that a perfectly efficient market is impossible. If prices fully reflected all information, no participant would spend resources discovering more of it. Without those informed participants, prices stop being fully efficient.

That leaves an equilibrium level of inefficiency which compensates the operators willing to commit capital, infrastructure, and risk. The spread and funding differentials between HyperCore and HIP-3 markets persist for exactly that reason: harvesting them requires fragmented capital, execution infrastructure, and tolerance for smart contract risk.

Andrew Lo's Adaptive Markets framework completes the picture. HIP-3 introduced a structural change: independent builders can deploy their own perpetual markets with their own liquidity, oracles, and funding parameters. Fragmentation is therefore not noise around one venue, but the source of a new venue topology.

The result is observable across assets like SILVER, HYPE, and OIL. Funding rates diverge, sometimes in opposite directions, because each market is sampling and pricing the same underlying through a different microstructure.

Pallas captures pricing differentials between perpetual markets running on separate liquidity pools and pricing engines. By holding offsetting positions on the same underlying across venues, net directional exposure is flattened while the spread remains monetisable.

The strategy rotates across seven venues: Hyperliquid Core plus six HIP-3 builder markets. Pairs include spot-perp, perp-perp, Core-HIP-3, and HIP-3-HIP-3 expressions, shifted toward the assets showing the widest and most durable dislocations.

The funding mechanism anchors perpetual contracts to a spot reference by transferring payments between longs and shorts. On Hyperliquid, each venue defines this independently with its own oracle, sampling cadence, cap structure, and settlement rhythm.

Each venue therefore computes a different fair price for the same asset at different times using different methodologies. That is not a bug in the system. It is a direct outcome of fragmented market design.

Across the observed 90-day period, Pallas collected positive funding on two builder venues while paying on Core and two others. The differential is structural, not random noise around one market.

HIP-3 lets independent operators deploy their own perpetual markets on Hyperliquid. Each builder maintains its own liquidity pool, oracle configuration, and funding parameters, which creates persistent pricing divergence that a neutral strategy can continuously scan and harvest.

Asset selection remains the discretionary overlay. OIL generated the second-highest PnL contribution in only seven active days, validating rotation into newly listed, high-volatility tokenized markets. COPPER was the only material loss, reinforcing that structural edge still depends on venue and asset choice.

The figures below reflect 90 days of live trading from December 11, 2025 to March 10, 2026. Results are realised PnL, with negligible open exposure at period end because the strategy systematically exits on convergence rather than carrying directional residuals.

The core characteristic is regime resilience. In calmer markets, returns remain positive. In high-volatility environments, spreads widen and funding mismatches amplify, lifting performance materially without changing the underlying structural logic.

PnL composition also evolved as execution improved. December was dominated by funding capture, while March skewed heavily toward spread capture after the system added higher-volatility tokenized assets and refined pair selection.