When a derivatives screen shows three different prices, they are answering different questions. Last price is the venue’s most recent matched trade. An index or oracle price is a reference calculated from specified spot or external inputs. A mark price is a venue-defined valuation measure. Their values can diverge, and there is no universal ordering that says which one must be highest.
The words “index” and “oracle” are not interchangeable across every venue. Hyperliquid calls its funding reference an oracle price. For a particular contract, read that venue’s definitions and record the instrument as well as the field name.
Hyperliquid’s oracle and mark use different inputs
Hyperliquid’s oracle documentation says validators publish an oracle for each perpetual asset about every three seconds. It describes asset-dependent spot inputs: for assets with primary spot liquidity on Hyperliquid, external sources may be excluded until sufficient liquidity is met; for assets whose primary spot liquidity is elsewhere, Hyperliquid spot is excluded. Each validator submits an oracle, and the clearinghouse combines validator submissions by stake-weighted median. The oracle feeds funding and is one input to the mark.
Hyperliquid’s robust price documentation defines mark as the median of three components:
- Oracle price plus a 150-second EMA of the difference between Hyperliquid’s mid price and the oracle.
- The median of Hyperliquid’s best bid, best ask and last trade.
- A weighted median of external perpetual mids from Binance, OKX, Bybit, Gate.io and MEXC, weighted 3, 2, 2, 1 and 1.
If exactly two components are available, the documented method adds a 30-second EMA of the venue book/last median to the median inputs. Oracle and mark update when validators publish oracle values, approximately every three seconds. Hyperliquid’s robust-price page summarizes oracle as an external CEX weighted median independent of Hyperliquid market data, while its dedicated oracle page describes asset-dependent source scope that can exclude external prices until liquidity conditions are met. Those summaries are not fully aligned. In the worked example, treat the oracle value as a supplied venue field, and verify the asset’s current inputs before rebuilding it. Hyperliquid documents mark as an input to margin, liquidation, TP/SL triggers and unrealized P&L. These rules describe this venue; another contract may use a different index, mark formula, source list or update process.
A fictional snapshot: last can differ from mark
Suppose the oracle is $100.00 and the existing 150-second EMA basis is zero. The venue bid is $99.98, ask $100.02 and last trade $96.00. The median of bid, ask and last is $99.98. Now suppose external perpetual mids are $100.00 with weight 3, $100.10 with weight 2, $100.20 with weight 2, $100.30 with weight 1 and $100.40 with weight 1. Their weighted median is $100.10. The three mark inputs are $100.00, $99.98 and $100.10, so their median is $100.00.
The $96 trade remains a real print in this fictional snapshot, but it does not alone determine the $100.00 mark.
The calculation omits feed timing, validator updates and actual EMA history; it is a teaching example, not a live mark reconstruction. For a fictional one-unit linear long entered at $100, last-price arithmetic is −$4 at $96, while mark-price arithmetic is zero at $100, before funding and costs. These are reference-price comparisons, not an account ledger. Hyperliquid’s TP/SL docs use mark as the trigger reference, but a triggered market or limit order has separate fill conditions.
A trigger price is not an execution price
On Hyperliquid, liquidation uses mark price, and its TP/SL documentation says mark triggers those orders. A triggered order still has to execute. A market TP/SL trades against available book liquidity under its documented slippage tolerance; a limit TP/SL can remain unfilled if the book does not reach the limit. Record the trigger reference separately from the actual fill and fees. A last-price wick does not, by itself, prove that a mark-based TP/SL or liquidation condition occurred.
This is also why chart data must be labeled. Candles built from last trades can show a sharp wick that a mark chart does not; mark and oracle values can update on their own cadence. An hourly OHLC bar cannot tell you every within-bar change or which source crossed a threshold first. Before saying “the stop was hit,” identify the venue, contract, trigger field, timestamp and order type.
Check the reference before making a claim
For a live position or historical test, name which series answers each question: last trade for the venue’s latest match; oracle/index for the contract’s reference feed; mark for the venue’s valuation or risk logic. Then verify the selected source, asset-specific constituents, update cadence, missing-input behavior and trigger settings in the current product documentation. Preserve timestamps and the exact contract. Do not substitute one venue’s mark history for another venue’s candles without labeling that mismatch. If the trigger field or source history is missing, keep the claim unverified rather than inferring it from a different candle series.
The CFTC’s virtual-currency advisory discusses volatility and leverage risk in spot and futures/options markets; it does not define an exchange’s price formula. SuperQuant’s historical engine identifies its valuation assumptions in the interface and methodology. Its bundled hourly histories are synthetic, not exchange mark prices, index values or actual trades, and its simplified model is not a liquidation oracle.
Continue with spot versus perpetual contracts, funding rates and crypto liquidations. This guide is general education, not venue-specific account advice.