Hyperliquid Weekend Price Discovery

How closely do Hyperliquid prices estimate the next futures reopening while traditional markets are closed?

This eight-page report compares 37 NASDAQ 100 weekends and 12 WTI crude oil weekends at four observation times. It includes intraday comparisons, error estimates with confidence intervals, a May 2026 oil-news case study, and calculation details.

Transcript

1. DIGITAL ASSET RESEARCH Weekend Price Discovery Hyperliquid, NASDAQ 100 and WTI crude oil Primary samples: NASDAQ 100, 37 weekends / WTI, 12 weekends NASDAQ: Oct 2025-Sep 2026 / WTI: Jan-Sep 2026 Research as of 25 September 2026
2. THE RESEARCH QUESTION How useful are Hyperliquid prices while futures markets are closed? A current price reference can help value positions during the weekend. We test how closely it estimates the reopening price of NASDAQ 100 and WTI crude oil futures. MARKET FRIDAY CLOSE DURING THE WEEKEND SUNDAY REOPEN Index / oil futures Record the closing price Market closed Observe the reopening price Hyperliquid Corresponding perpetuals Record the starting price Trading continues Measure the price change Use the price just before futures reopen When does the estimate become useful? How close is the price estimate? Compare observations 24 hours, 6 hours, 1 hour and just before the futures market reopens. Apply the Hyperliquid move to the Friday futures price. Compare the error with simply keeping Friday’s price. HL = Hyperliquid. Product references, sample filters and calculation details: p. 8. 02
3. WHEN BOTH MARKETS ARE OPEN Futures and Hyperliquid follow closely matching intraday paths 7 and 14 August 2026, the first two Fridays of the month. Intraday prices sampled every 15 minutes. Futures Change from just before 09:30 (%) US Eastern time. Scale differs by asset. Hyperliquid NASDAQ 100 / 7 August 2026 WTI crude oil / 7 August 2026 +2.0% +0.4% 0% +1.0% -0.4% -0.8% 09:30 0% -0.5% 11:00 13:00 15:00 16:00 NASDAQ 100 / 14 August 2026 11:00 13:00 15:00 16:00 13:00 15:00 16:00 WTI crude oil / 14 August 2026 +2.0% +0.4% 0% +1.0% -0.4% -0.8% 09:30 09:30 0% -0.5% 11:00 129 trading days: hourly return correlation 23 March to 24 September 2026 13:00 15:00 16:00 09:30 11:00 NASDAQ 100 WTI crude oil Daytime 0.9979 / Overnight 0.9969 Daytime 0.9949 / Overnight 0.9962 Source: Hyperliquid and FMP. Each intraday line starts at 0% using its own price just before 09:30. Details: p. 8. 03
4. WEEKEND PRICE DISCOVERY WTI’s largest weekend moves are visible before futures reopen 11 shared weekends in 2026. Returns from Friday close. Dates identify the Friday; vertical scales differ by asset. Futures: Friday close to actual reopening Hyperliquid: Friday close to just before reopening NASDAQ 100 HL estimate closer: 6/11 weekends +1% 0% -1% -2% 3/27 4/24 5/1 5/22 5/29 6/26 7/24 7/31 WTI crude oil 8/21 8/28 9/4 HL estimate closer: 11/11 weekends +2% 0% -4% -8% 3/27 4/24 5/1 5/22 5/29 Shared dates: intersection of the two primary samples. Main estimates use 37 NASDAQ and 12 WTI weekends. Methods: p. 8. 6/26 7/24 7/31 8/21 8/28 9/4 04
5. WEEKEND PRICE DISCOVERY How much does Hyperliquid improve the estimate of the futures reopening price? Compared with using Friday's futures close unchanged. Each dot is one weekend; farther right means a larger improvement. Error reduced Error increased Mean improvement Pointwise 95% confidence interval NASDAQ 100 / 37 weekends WTI crude oil / 12 weekends 2025-10-17 to 2026-09-04 2026-01-30 to 2026-09-04 Friday close is closer -1.0 -0.5 HL estimate is closer 0 +0.5 Friday close is closer +1.0 -2.0 0 HL estimate is closer +2.0 24h before 24h before 6h before 6h before 1h before 1h before Just before Just before Final MAE: 41% below Friday unchanged Final MAE: 90% below Friday unchanged +4.0 +6.0 +8.0 Improvement = Friday-close error - HL error (pp). MAE = mean absolute error. Bars: unadjusted 95% intervals. Horizontal scales differ. Source: Hyperliquid and FMP. E9 Technologies calculations. Methods and scope: p. 8. 05
6. CONCLUSION The final estimate has the lowest mean absolute error in both samples Compare the same reopening price using observations 24h, 6h, 1h and just before reopening. NASDAQ 100: 37 weekends WTI crude oil: 12 weekends 0.333% / 0.195% 2.181% / 0.227% MAE: Friday unchanged / HL just before reopen MAE: Friday unchanged / HL just before reopen 41% lower MAE than Friday unchanged 90% lower MAE than Friday unchanged Only the final pointwise 95% interval is above zero. Adjusting for all four observation times widens it to include zero. At 1 hour before and just before reopening, the 95% intervals remain above zero after adjusting for all four observation times. Use: a current price reference for valuing positions while the futures market is closed. Source: Hyperliquid and FMP. E9 Technologies calculations. Methods and scope: p. 8. 06
7. WEEKEND NEWS: A CASE STUDY WTI moved on Hyperliquid while futures were closed WTI crude oil | 22-25 May 2026 (UTC) | Trump’s posts on negotiations with Iran One of the 12 WTI weekends. The decline began before the first selected post. 1 23 May, 20:30 UTC 2 24 May, 14:10 UTC Deal largely negotiated No rush; blockade stays Move in the posting hour: -4.53% Move in the posting hour: +2.29% Futures open Futures Hyperliquid Futures open FUTURES CLOSED / HYPERLIQUID TRADING 1 Each market’s Friday close = 0% 2 +2% 0% Unchanged error 3.22pp Friday price unchanged (0%) -4% HL error: 0.08pp -8% 22 May 15:00 Fri close 21:00 23 May 12:00 24 May 00:00 24 May 12:00 Sun reopen 22:00 25 May 04:00 UTC Sources: UCSB, Hyperliquid, FMP. Hourly closes + futures open. Yellow bands include minutes before each post. pp = percentage points. Details: p. 8. 07
8. REFERENCE Sources and calculation details Prices NASDAQ 100: xyz:XYZ100 / NQUSD. WTI: xyz:CL / CLUSD. Captured 25 September 2026. HL: API candles and hourly candles rebuilt from official S3 fills. FMP: minute endpoints checked against hourly bars. Intraday examples use HL 15-minute and FMP minute candles; each series starts at its own 09:29 close. Hyperliquid API Open hours Official archive XYZ specifications 129 weekdays, 23 Mar-24 Sep 2026. Per asset: 774 daytime / 1,806 overnight hourly returns. US Eastern 10:00-16:00 and 19:00-09:00; holiday sessions excluded. Nights belong to the following daytime date. CME trading hours Weekend sample FMP documentation Holiday calendar NASDAQ: 37 weekends, 17 Oct 2025-4 Sep 2026; holiday closes, quarterly rolls and 30 Jan mismatch excluded. WTI: 12 weekends, 30 Jan-4 Sep 2026; missing or inconsistent price data excluded. The roll filter leaves 12 of 29 otherwise valid weekends: XYZ monthly roll start through CME expiry is excluded. Jan-Mar XYZ dates apply the current business-day rule retrospectively; Apr-Sep dates use the published schedule. FMP continuous symbols lack contract IDs and a documented roll rule. Samples cover different periods. NASDAQ roll dates XYZ monthly roll schedule CME WTI expiry rule Calculation Friday close 17:00 / Sunday reopen 18:00, US Eastern. Observe 24h, 6h, 1h and just before reopening. Use the completed HL hourly candle before each cutoff. Estimate = Friday futures close × (HL cutoff / HL Friday). Error (%) = 100 × |estimate - actual reopen| / Friday futures close. Improvement = unchanged error - HL error. Equal weekend weights. Paired t 95% intervals; bootstrap checks and Bonferroni correction for four times per asset. Context Blockworks Research examines weekend price discovery and trading during market closures. This study tests a price reference for valuing positions during the weekend. Evaluating trading returns requires executable quotes, trading rules, funding payments and fees. News case: 23-24 May. UCSB post times converted to UTC; event-hour returns include minutes before each post. Blockworks Research UCSB: 23 May posts E9 Technologies. Sources and saved observations as of 25 September 2026. UCSB: 24 May posts 08