Data Research: How Big Is the Liquidity Gap Between Hyperliquid and CME Crude Oil?

null

Original author / Castle Labs

Compiled / Odaily Planet Daily Golem（@web 3_golem）

Editor’s note: This article systematically studies the differences in crude oil futures trading data between Hyperliquid and CME during weekday and weekend sessions, and draws several important conclusions. At present, Hyperliquid indeed cannot match CME on absolute metrics such as liquidity depth or slippage; overall liquidity is still below 1%, which is consistent with the fact that the primary users of RWA trading platforms are still crypto-native retail traders.

What’s different about Hyperliquid is that during weekend sessions, trading volume in crude oil futures on Hyperliquid increases noticeably. This suggests that besides retail traders with speculative demand, traders who want to obtain crude oil trading exposure before Monday and carry out hedging operations are also trading on Hyperliquid during weekends. Moreover, this trend is becoming increasingly evident, indicating that Hyperliquid has already developed price discovery capabilities for large commodities.

However, for institutional investors, Hyperliquid’s high trading costs—compared with CME—remain the main obstacle to its expansion in the large-commodities trading space. If Hyperliquid does not improve its ability to handle orders at the institutional level soon enough, it can only serve as a temporary trading venue for traditional traders on weekends, and ultimately become a small supplement in the broader traditional finance landscape.

Research methodology and data sources

This analysis evaluates the microstructure of the crude oil market through two studies, covering the market during weekdays and weekends, respectively, and using trade-by-trade execution data from two trading venues: Hyperliquid’s xyz:CL perpetual contract and the Chicago Mercantile Exchange (CME)’s CLJ6 (April 2026 NYMEX WTI crude oil futures).

CME data comes from Databento’s trading data feed. This feed captures trade-by-trade execution data rather than order book snapshots. Therefore, all of CME’s depth and slippage estimates are based on actual traded volume rather than quoted depth. Hyperliquid data comes from Hyperliquid’s publicly available S3 database, which contains complete on-chain execution records.

Accordingly, the analysis of both trading venues is based on actual traded volume. All depth data represent visible liquidity—i.e., the traded volume within a specific-basis-point range in a 5-minute window around the VWAP mid-price—not the full dormant depth from the order book.

Study period and market context

The research period runs from February 27, 2026 to March 16, 2026—a window that coincides with geopolitical turmoil following an attack launched by Iran on February 28, 2026.

Market close before the attack: the last CME trading day before the attack event.

Monday open: there is huge reopening pressure; CME’s stock price gaps up sharply at the open, while the Hyperliquid xyz:CL market is constrained by price discovery boundaries.

The following few weekends: because oil prices remain high, market volatility keeps crude oil trading volume on the Hyperliquid platform elevated.

xyz:CL launched in early 2026, meaning that the observation windows for these three weekends cover the early maturation stage of the Hyperliquid market. The observed trends—including improving liquidity depth, increasing trading volume, and growth in user numbers—partly reflect market maturity. But we believe that, in absolute metrics such as liquidity depth or slippage, on-chain exchanges like Hyperliquid still cannot match traditional exchanges.

Our goal is to track directional trends: whether the price spread between the two is shrinking, how fast it is shrinking, and under what conditions it shrinks.

Data analysis

The data analysis is divided into two parts by time period:

Weekday sessions: covering the full three-week period, comparing Hyperliquid and CME weekday depth, slippage, and the premium/discount at which the two exchanges trade. For Hyperliquid, we also analyze its funding rates across the entire period.

Weekend sessions: within a given time period, containing three weekends; we analyze price discovery and the deviation of the price gap on Hyperliquid relative to the CME opening price.

Weekday session data analysis

This analysis covers the full three-week period and focuses on the overlapping times when both exchanges are active.

Liquidity depth is measured by dollar traded volume within ±2, ±3, and ±5 basis-point ranges around the VWAP mid-price for each 5-minute interval, and aggregated into the median across all weekday intervals. As stated above, this reflects traded volume within the range, not full resting quoted depth. This approach may underestimate CME and Hyperliquid liquidity depth.

Execution slippage is estimated using a synthetic order book constructed by sorting executions by execution price. Within each 5-minute time interval, the observed aggressive execution records are sorted in ascending order by price (simulating sweeping sell orders), and the orders are swept sequentially until reaching the target order size. The arrival price is set as the lowest execution price within that time period (representing the best sell price at the time the order arrives). Slippage is calculated as the difference between the VWAP (volume-weighted average execution price) and the arrival price, expressed in basis points. This method is applied to incremental order sizes ranging from $10k to $1M.

Weekday-session Hypeliquid-CME basis: track the signed price difference between Hyperliquid’s mid-price and CME’s latest price within all weekday 5-minute windows. This reflects any structural premium or discount at which Hyperliquid trades relative to CME reference prices during active windows. Hyperliquid mid-price is sourced from the VWAP (volume-weighted average price) of executions within each 5-minute trading interval, rather than from real-time order book quotes.

Hyperliquid funding rate is priced hourly, and expressed in basis points per hour.

Weekend session data analysis

This analysis focuses on three different weekend exchange holidays for CME:

W1: February 28 to March 1, 2026

W2: March 7 to March 8, 2026

W3: March 14 to March 15, 2026

In W1 and W2, Hyperliquid perpetual contracts are constrained, so the mark price cannot exceed the “trading range limit boundary” (DB). When the oracle price is frozen (e.g., when the major reference market (CME) is closed and external price data sources stop updating), the protocol effectively locks the price within a narrow range.

For each weekend window, we report key metrics for Hyperliquid xyz:CL, including price, traded volume, and number of trades. To measure the deviation of the Monday opening spread, for each weekend we measure the price gap between Hyperliquid and CME at three reference points:

3 hours before CME reopens

1 hour before CME reopens

At CME open (T=0)

All spreads are expressed in basis points. Positive values mean Hyperliquid is above CME’s opening price; negative values indicate a discount.

Quantitative analysis

This section first analyzes and compares the liquidity of the Hyperliquid xyz:CL HIP-3 crude oil market versus NYMEX CLJ6 during overlapping weekday sessions.

Liquidity depth: less than 1% of CME

There’s no doubt that on-chain exchanges have liquidity conditions that are completely different from CME. The average liquidity depth of CL on Hyperliquid is less than 1% of CLJ6, and the liquidity depth across price ranges is consistent (109x at ±5 bps). In the mid-price range of ±2 bps, CME’s executable depth is $19 million, while Hyperliquid is only $152k—an approximately 125x difference.

Given Hyperliquid’s novelty as a CL market and the differences in its target user base, this result is not surprising. The main value of on-chain exchanges is to provide permissionless trading channels for users who are traditionally excluded from institutions like CME.

However, as weekend trading volume on DEXs like Hyperliquid grows, people’s perceptions of these platforms are starting to change. Institutional investors are increasingly interested in hedging positions during non-trading hours. Therefore, for Hyperliquid, it becomes increasingly important to cultivate a market environment that suits both traditional investors and retail traders.

For a retail trader with a trade value of $10k, this cost gap is negligible. But for an institutional investor with a trade value exceeding $1M, the on-chain trading costs for CL (and most other markets) remain hard to bear.

In fact, the inherent differences in the user base are reflected in the median trade size during these overlapping sessions.

The 166x difference in median trade size (90,450 vs 543) most clearly demonstrates that these trading venues serve fundamentally different types of users. The median trade size for CLJ6 is comparable to a standard crude oil futures contract (based on the current price, a notional value of about $94k), while Hyperliquid’s median trade size is $543, reflecting crypto-native retail traders placing leveraged directional bets.

We expect that as these markets become increasingly “legitimized” in the eyes of more traditional investors and capital migrates to on-chain platforms, the median trade size in Hyperliquid’s commodities market will reach an inflection point.

To further distinguish between different trade sizes, we ran order simulations with order size caps ranging from $10k to $1M.

For a $10k order, CLJ6 traders have no slippage, which matches expectations, while Hyperliquid users’ median execution slippage is below 1 basis point, at 0.77 basis points. The gap appears at a $100k order size: Hyperliquid users’ slippage rises to 4.33 basis points, approaching the ~5 basis point threshold, while CME CLJ6 has no slippage.

Notably, this is higher than the median trade size of the CLJ6 market ($90,450).

At a trade size of $1M, Hyperliquid’s 15.4 basis points is about 20x CME’s 0.79 basis points, confirming that this venue currently lacks the ability to handle institutional-level orders. Given Hyperliquid’s average trade size, the platform should be able to provide equally good service to users without causing slippage.

CLJ6 orders start to show significant slippage at trade sizes around $500k, which then affects execution.

When we extend the order-size analysis to weekend sessions, slippage for all order sizes declines, especially for $100k and $1M orders, indicating that the market has matured. Over the three weeks analyzed, the simulated order slippage decreases as follows:

$10k: -16%

$100k: -75%

$1M: -86.9%

Funding rate

CL’s funding rate fluctuates more during CME’s close trading sessions, but fluctuates less during delivery sessions. This helps us uncover the internal pricing dynamics of the market during non-trading hours. Since the weekend is open, the CL market can take advantage of internal price discovery mechanisms (supported by the DB and other risk-reduction mechanisms). Therefore, the funding rate is expected to be more volatile, as highlighted below.

During active trading sessions, Hyperliquid’s xyz:CL closely tracks CME’s CLJ6. However, as oil prices rise, a structural discount emerges and widens, which is likely caused by funding-rate pressure from accumulated long positions. During weekend sessions, with CME closed, Hyperliquid’s price discovery is further constrained by the price range mechanism (DB). In the absence of a live reference market, this mechanism limits the volatility of the mark price.

Separate weekend-only analysis: Hyperliquid has already achieved price discovery capability

These three weekends show Hyperliquid’s rapid maturation:

W1: February 28 to March 1, 2026 (Iran attack event)

Price on Hyperliquid rises from a level near CME of about $67.29 to around $70.80, accounting for about 45% of the Monday final gap up to $75 (+1,146 bps).

It’s important to note that, due to the ±5% price range limit mechanism (DB) mentioned above for trade.xyz, price discovery during this weekend was constrained. This explains why the curve in the chart is relatively smooth and why there is a gap up on Monday. Even so, in the first second after the paired data release, the difference between Hyperliquid xyz:CL ($73.89) and CME CLJ6 ($75) is within 1.5%.

This is not a “mistake” or a “failure,” but risk protection achieved through market design. Therefore, from a data perspective, the correlation is lowest on the first weekend; but it demonstrates that xyz:CL responded to the initial shock from the Iranian airstrikes, while also recognizing the importance of DB as a weekend price discovery mechanism—especially for emerging markets.

W2: March 7 to March 8, 2026

The second weekend is the real test, because xyz:CL touched the range boundary price toward the market close. CLJ6’s opening price is $98 (up 737 bps from the $91.27 closing price), while xyz:CL peaked at about $95.83 and only captured 68% of the upside.

On the second weekend, xyz:CL captured the market trajectory better and was closer to CME’s opening price than on the previous weekend.

W3: March 14 to March 15, 2026

The third weekend’s data show that in a relatively calm market environment, Hyperliquid can more reliably predict CME’s final opening direction.

This weekend shows the best convergence between xyz:CL and CLJ6: up 226 bps versus CME’s close, slightly higher than the 62 bps gap versus Monday’s open. CLJ6’s Friday close is $99.31, and its open is $100.93 (up 163 bps), while xyz:CL’s open is $101.56.

Overall, these three snapshots show structural changes in the xyz:CL market on Hyperliquid: the market transitions from an emerging market constrained by DB price discovery (weekends 1 and 2) to increasingly free price discovery, with overshoots and pullbacks (weekend 3).

When analyzing the price discovery deviation error across different time segments before CME’s open (3 hours, 1 hour, 0 hours), we find W3’s data is the most reliable. During the first two weekends, the xyz:CL market was affected by DB. In W3, xyz:CL’s errors about 3 hours before CME’s open and 1 hour before are approximately +70 and -139 bps, indicating better price discovery capability than the previously analyzed weekends.

Other metrics

We also provide additional metrics from the weekend summary analysis, including trading volume, total number of trades, and average trade size. These metrics vary across weekends and have continued to grow for several consecutive weekends.

The total trading volume in the xyz:CL market grows from $31 million to more than $1 billion over the three weeks, reflecting an increase in the number of users and the market’s final maturation.

In addition, the total number of trades increases from 26k on the first weekend to more than 700k on the third weekend.

Notably, the average trade size during weekends actually grows from the median mentioned earlier to $534. The same growth trend is observed across all three weekends, which may indicate more institutional capital flowing into the market.

The average trade size on the first weekend is $1,199, increasing to over $1,500 by the third weekend.

This may indicate that the user base using the platform on weekends has changed: retail users decrease, while more traders need to obtain crude oil trading exposure before Monday. Therefore, weekend trading is closer to hedging demand rather than speculation.