In the early hours of March 9, the situation in Iran escalated. CME closed, ICE closed, and major global futures exchanges all shut their doors. The next official price for crude oil will not be available until the early Monday morning opening, hours away.

However, the crude oil contract CL-USDC on Hyperliquid did not wait. On that day, the trading volume of this on-chain perpetual contract surged from a daily $21 million to over $1.2 billion. Traders used an on-chain protocol to instantly price geopolitical risk during the time window when traditional markets were closed.

This event was widely shared in the crypto community as another triumph of DeFi. But few have asked a more fundamental question: where does the price on this on-chain exchange come from when external markets are closed?

When there is no external price, where does the price come from?

Trade.xyz is the largest provider of traditional asset perpetual contracts on Hyperliquid, operating on the HIP-3 protocol, accounting for 90% of the total HIP-3 open interest. The S&P 500, Nasdaq 100, WTI crude oil, gold, silver, South Korean stocks—all can be traded 24/7 on it. However, the pricing logic of perpetual contracts is completely different from spot. The price on spot exchanges is generated by direct matching of buyers and sellers, while perpetual contracts require an "anchor" to peg the contract price to the real price of the underlying asset. This anchor is the oracle.

In the traditional futures market, the pricing anchor is the exchange itself. The price of CME's crude oil futures is the price of oil, without needing additional reference. But Trade.xyz's contracts run on the Hyperliquid chain, with no direct connection to Chicago's matching engine. When CME is open, Trade.xyz's oracle directly references CME's quote, which is straightforward. The real challenge arises after CME closes.

Trade.xyz's solution is to have the oracle extract information from its own order book. The system calculates an "impact price difference," simply put: if someone wants to buy a large amount now, how much higher will the average price be than the current price? If someone wants to sell a large amount, how much lower will it be? This deviation reflects the imbalance of buying and selling pressure in the order book. The oracle adds this deviation to the current price to get a "target price," then slowly converges the current price to the target price using a decay function.

The keyword is "slowly." The oracle updates every 3 seconds, but each time it only moves a small portion of the gap between the current price and the target price. This movement speed is controlled by a time constant. The larger the time constant, the slower the oracle, the harder to manipulate, but also the less able to reflect true market sentiment.

During Trade.xyz's early days, this time constant was set to 8 hours. In November 2025, this parameter was reduced to 1 hour. The reason for the reduction was related to traders' real money: Trade.xyz settles funding rates every hour. The oracle tracking the real price was too slow, causing profitable traders to be continuously drained by the funding rate.

As shown by the red line in the figure below, after the same 8 hours, if you are long on crude oil and the direction is correct, but the oracle took 8 hours to catch up to the real price, and during those 8 hours, the price never reached your target (the real price), your gains were heavily eroded by the funding rate.

After reducing the parameter to 1 hour, the price only took 5 hours to reach your expected position (blue line). The price can confirm your judgment more quickly, reducing the number of funding rate payments compared to before.

However, a faster oracle also brings new risks. If the oracle fails for 6 hours and then suddenly recovers, according to the formula, it will jump directly to 99.7% of the target price. This instantaneous price jump can trigger large-scale liquidation. Trade.xyz's solution is to add a safety valve: no matter how long has actually passed, the maximum time difference for each update is calculated at 6 minutes. Even if the oracle crashes and then recovers, the price can only catch up in small steps.

Cage, Heavy Anchor, and Monday Morning Gap

Oracle pricing solved the problem of "how to price over the weekend." But another problem arises: to what extent can the price freely move?

Trade.xyz drew a "cage" for each contract. The mark price is restricted within a certain percentage above and below the last external closing price. This percentage is equal to the inverse of the maximum leverage. With a maximum leverage of 20x for the crude oil contract, the cage is 5% above and below the closing price. If crude oil closed at $100 on Friday, the mark price can only fluctuate between $95 and $105 over the weekend. Touching the boundary leads to immediate trading halts.

Weekend trading halt of early March crude oil contract

During a normal weekend, this mechanism works well. The 5% range is sufficient to absorb most overnight fluctuations. However, a geopolitical event of the magnitude seen on March 9 will push the price directly to the edge of the cage. With all market information backlogged, by the time CME opens on Monday, if the real price has jumped by 8%, a massive gap will form. Those who were short get liquidated instantly, and market makers suffer losses due to the inability to hedge gradually.

In March 2026, Trade.xyz deployed "Price Discovery Boundary v2" on the oil contract. Key change: the cage's size remains the same, but the cage can move. When the oracle price reaches the 90% position of the current boundary, the system reanchors the cage's center to the boundary value, drawing a new cage of the same size around the new anchor point. A maximum of two reanchors is executed in each direction.

In specific numbers: initial cage is $95 to $105. When the oracle rises to $104.50, a reanchor is triggered, and the new cage becomes $99.75 to $110.25. After another trigger, it becomes $104.74 to $115.76, which is the endpoint. Starting from $100, the maximum discoverable range expands to about $115.76.

This design ensures that the instantaneous volatility range is always kept at 5%, and market makers' risk models do not need to change. At the same time, reanchoring means the system "acknowledges" the price movement that has already occurred, narrowing the gap at Monday's opening. But the trade-off is clear: a long position with a liquidation price at -8% is absolutely safe under v1 (because the price cannot reach -8%), but under v2, it may enter the liquidation range after one downward reanchor. Trade.xyz chose to deploy v2 on two oil contracts first and will decide whether to expand after observing the effects.

Another key component of the pricing system is the funding rate. The funding rate is like a rubber band that ties the perpetual contract price to the oracle price: when the mark price is higher than the oracle, longs pay shorts; when it's lower, shorts pay longs. Trade.xyz's funding rate formula is similar to most crypto exchange structures, but with a 0.5 scaling factor applied at the beginning.

This 0.5 is calibrated for traditional assets. The basic annualized funding rate for crypto perpetual contracts is about 11%, reflecting the pure leveraged holding cost, which is reasonable for assets like Bitcoin that do not have dividends. However, for stocks and commodities, the actual holding cost is close to SOFR plus 1 to 2 percentage points, around 5% to 6%. Multiplying by 0.5 reduces the basic annualized rate from 11% to about 5.5%, aligning with traditional assets. This is especially crucial on weekends: the scaling factor directly halves the weekend funding rate, and in conjunction with a 1-hour time constant oracle, allows traders in the right direction to retain most of their profits.

Different Assets, Different Processing Pipelines

Precious metals have an active global spot market. The external prices of gold, silver, platinum, and palladium are directly taken from spot quotations, eliminating futures roll issues. However, crude oil and industrial metals do not have a unified spot quotation; Trade.xyz can only use CME futures contracts as the pricing basis. Futures have expiration dates, and the system needs to switch from the current month contract to the next month contract every month. The issue is that the prices of the two contracts are usually different. Storage costs and supply-demand expectations can make the far-month contract price higher than the near-month. If there is a price jump during the switch, the P&L of the holders may show unrealistic fluctuations, potentially triggering unwarranted liquidations.

The handling by Trade.xyz is to transition gradually over 5 trading days: from the 5th working day of each month to the 10th working day, using the weighted average of the near and far month contract oracle prices, with the weight linearly changing daily.

Pricing of the stock index contract is a bit more complex. The XYZ100 tracks the Nasdaq 100, but CME's Nasdaq futures trade almost 24/5, providing a longer price reference than the spot market. Trade.xyz initially reverse-engineered the spot price from the futures price, fixing a 4% discount rate to strip out holding costs. However, this fixed value deviates when the Fed raises rates. The v2 solution launched in February 2026 changed to a dynamic calculation: using the spot index value directly at the U.S. stock market open, while reverse-engineering an implied discount rate from the futures and spot basis; this discount rate is then used after-hours to reverse-calculate the spot price.

There is also a special case: South Korean individual stocks. Trade.xyz has listed Samsung Electronics, SK Hynix, and Hyundai Motor, which are priced in Korean Won on the Korean exchange. The oracle needs to overlay a USD/KRW exchange rate conversion on the original price. Holders' gains and losses reflect both stock price and exchange rate fluctuations.

Who is responsible for the consequences of parameter choices?

All these pricing mechanisms are built on one premise: there are enough market makers willing to provide liquidity continuously. Hyperliquid's HLP liquidity mining treasury provides liquidity for native BTC, ETH perpetual contracts but does not cover third-party contracts deployed on HIP-3. Trade.xyz's liquidity relies entirely on external market makers voluntarily participating. In extreme market conditions, if a liquidated position cannot find a counterparty to take over, the system will not have a treasury like Hyperliquid's main site to backstop but will directly trigger ADL (Auto-Deleveraging), forcibly liquidating the most profitable counterparty positions based on profit ranking.

The subtlety of this pricing system lies in using a set of interrelated parameters—the oracle's tracking speed, the boundary of price discovery, the scaling factor of funding rates—to build a pricing environment that can be self-sufficient in the absence of external quotes. On March 18, S&P's decision to authorize Trade.xyz may be based on this infrastructure that has passed the test in a real geopolitical crisis.

However, this system also has its costs. The oracle extracts information from the order book, meaning that in periods of thin liquidity (such as late-night Korean stock contract trading), a small number of orders can lead to significant oracle movements. The price discovery boundary v2 expands the liquidation range over weekends, requiring leveraged traders to reassess their safety margins. ADL means that even if you are correct in your judgment, you may still be forcibly liquidated in extreme market conditions.

Trade.xyz has chosen a path completely different from traditional exchanges: transferring the pricing power from a centralized matching engine to an on-chain parameter system. Traditional exchanges close because clearing, risk control, and market-making all require a window of human intervention. Trade.xyz never closes its doors because an on-chain smart contract has no concept of "closing time." It must be able to provide a price at any moment. The oil event on March 9 demonstrated that this system can operate under pressure. However, it also exposed a deep-rooted issue: when a blockchain protocol undertakes the pricing function of traditional financial infrastructure, who is responsible for the consequences of parameter choices?

The team at Trade.xyz made a parameter decision by reducing the time constant from 8 hours to 1 hour. The price discovery boundary upgraded from v1 to v2 as well. These decisions affect the liquidation threshold and funding rate of every position holder. In a traditional exchange, such rule changes require regulatory approval and a public notice period. On-chain, a parameter update can be done in one go.

In a system without HLP backstops, regulatory arbitration, entirely relying on parameter design to maintain order, understanding how these parameters affect your position is understanding the risk you are truly taking on.

Welcome to join the official BlockBeats community:

Telegram Subscription Group: https://t.me/theblockbeats

Telegram Discussion Group: https://t.me/BlockBeats_App

Official Twitter Account: https://twitter.com/BlockBeatsAsia