Source data. Public Hyperliquid PERPETUALS trades from Tardis, not builder archives. Lighter books from Tardis full-depth incremental L2 data.
Counterfactual. Public Hyperliquid trade prints are grouped by exact exchange timestamp, symbol, and side, then each group is simulated as one market order walking the observed Lighter book at the same timestamp.
Fees. Public trades do not include wallet fee tiers, so fee savings use the base Hyperliquid taker fee assumption on the portion the observed Lighter book could fill: 4.5 bps on standard perps, 0.9 bps on xyz markets, and 9 bps on xyz:GOLD. Lighter taker fee is modeled as $0 for standard accounts; unfilled notional gets no fee credit.
Order grouping. The grouping rule assumes fills from one taker order share the exact same exchange timestamp, symbol, and side. Public prints do not expose parent order ids.
Coverage. Coverage is the share of each grouped trade filled by the reconstructed full-depth Lighter L2 book at that timestamp; any order larger than the observed book is price-compared and fee-credited only on the filled portion.
HIP-3 / RWA markets. Markets with no Lighter equivalent are excluded and listed
above with their volume.
Slippage is signed distance of execution VWAP from the Lighter mid at fill time —
one consistent reference for both venues.
Cancel priority. Hyperliquid gives cancellations priority over incoming taker
orders: market makers can pull their quotes before your order reaches the book, so
Hyperliquid fills are often worse than the liquidity you saw when you clicked. Lighter
has no cancel priority.