The weekend price discovery ability of Hyperliquid perpetual stock contracts.

Written by: Ponyo (@ponyo_fp), Researcher at Four Pillars, Core Contributor to Hyperliquid

Translated by: AididiaoJP, Foresight News

TradeXYZ launched perpetual futures contracts for Samsung Electronics, SK Hynix, Hyundai Motor, and EWY on Hyperliquid in February, enabling investors to trade exposure to Korean stocks on weekends for the first time.

I reviewed the performance of stock contracts every weekend since the launch, covering 62 observations across 4 assets, comparing the movements of each contract from Friday's close to Sunday's close with the opening auction direction of the underlying assets on Monday. For Korean stocks, I used the KRX opening price at 9:00 AM (KST); for EWY, I used the NYSE opening price at 9:30 AM (ET).

Hyperliquid accurately predicted the opening direction in 45 out of 62 cases, achieving an accuracy rate of 73.8%. This figure sounds good, but the overall data masks the highly uneven performance across different assets.

Samsung Electronics showed a clear outlier. Out of 16 weekends, Hyperliquid correctly predicted the direction 15 times. The distribution of price changes for Samsung Electronics on Monday is even, so there is no issue of a biased basic direction raising the hit rate. The p-value from the binomial test (with a baseline of 50%) is 0.0003, indicating that the probability of such a strong result solely from random guessing is only 0.03%. Although the sample size is still small, the result of 15 correct predictions out of 16 is hard to ignore.

Hyundai Motor also performed well, getting 13 out of 16 correct, with an accuracy rate of 81%. However, after considering the slight downward bias of this stock on Monday, this result did not reach statistical significance. During the sample period, the probability of Hyundai Motor's opening price dropping on Monday was 62%, meaning that a strategy simply predicting "down" would itself have a high baseline win rate.

SK Hynix was less convincing. It was correct only 10 times out of 16, slightly better than a coin flip. Some mistakes were quite severe. Over the weekend of June 5-7, the Hyperliquid contract closed up 0.11%, while SK Hynix gapped down 10.34% when KRX opened the following Monday.

EWY had the worst results. It was correct only 7 times out of 13 weekends, even below baseline. On these 13 Mondays, EWY had an opening price increase 10 times, accounting for 77%. If one simply predicted "up" every week, performance would exceed Hyperliquid's signals. Hyperliquid's weekend contracts closed down 9 times out of 13, repeatedly directing traders in the wrong direction.

The differences between Samsung Electronics and EWY can largely be explained by the market time difference. The KRX opening time is KST 9:00, which is UTC 00:00, almost perfectly aligned with Hyperliquid's daily line reset time. Therefore, the time difference between Hyperliquid's last trade on the weekend and the KRX opening auction is only a few minutes.

In contrast, trading for EWY on the NYSE does not begin until ET 9:30, which is about 14 hours after Hyperliquid's Sunday candle close. By then, KRX has completed the entire Monday trading session, the European market has opened, and pre-market trading in the U.S. has incorporated new information. The information that decides EWY's opening on Monday does not even exist at the time of Hyperliquid's weekend candle close.

A reasonable doubt is that even Samsung Electronics' results may reflect last-minute convergence, rather than true price discovery. Informed traders might adjust their positions on Hyperliquid just before the KRX opens, making such "predictions" almost tautological.

To verify this, I switched to using Saturday's close instead of Sunday's close for repeated analysis, creating about a 24-hour lead time instead of just a few minutes. The results showed that Saturday's signal predicted the direction of Samsung Electronics' opening on Monday with an accuracy of 75% (with a baseline of 50%). Sunday trading improved the results but did not create this signal from scratch.

In comparison, Samsung Electronics' price movement on Friday predicted Monday's direction with an accuracy of only 62%. This indicates that Hyperliquid's weekend market provides information beyond the simple continuation from the previous KRX trading period.

Among the four assets tested, only Samsung Electronics produced a statistically significant directional signal. The sample size is still small, and the results have not been validated out of sample. Nevertheless, this may still provide reference value for any investor trading Samsung Electronics or the broader KOSPI market.

Despite the average weekend trading volume being only around $12,000, Hyperliquid correctly predicted the opening direction of a company valued at about $300 billion with an accuracy rate of 94% and a p-value of less than 0.001. For anyone trading Samsung Electronics on KRX, checking the weekend closing price on Hyperliquid before the Monday opening auction seems highly worthwhile.