In this AI stock rally, there has been a US stock perpetual market running 24/7 on-chain that has been trading right along with it. Below are four sets of numbers it left behind.
First set: the positions are pushed into storage, not NVIDIA. At 00:23 on September 24, for the whole market’s 124 underlying assets converted at mark price, the largest open interest belongs to SK hynix Korea stock perpetual with $304 million; next are Micron with $177 million, hynix ADS with $172 million, and Intel and NVIDIA at $142 million each. The two hynix contracts plus Micron and SanDisk (US$122 million) total $774 million for three storage companies—more than five times NVIDIA.
The last three sets use nine US-listed underlyings I selected: NVIDIA, Microsoft, Google, Meta, Amazon, Micron, Intel, Tesla, and a 100-stock index. The window is 31 days from August 23 to September 23.
Second set: it’s running 24/7, but the price isn’t moving 24/7. NVIDIA’s 31-day trading value shows 47.4% occurs after the Wall Street close; by hour, during the opening session the hourly trading value is $7.64 million, while in the closing session it’s only $1.83 million—about 4.2 times less. The opening-session average amplitude is 0.79% per hour, versus 0.28% for the closing session. The nine underlyings’ off-exchange (OTC) amplitude is only 20% to 50% of the on-exchange one. The prices being “found” are still those few hours when New York opens.
Third set: weekends move. Excluding Labor Day weekend, among the most recent three weekends (24 stock samples), the move from Friday’s close to Monday midnight is less than 1% for 16 instances. Over 2.5% happens three times: Intel up 2.9%, while Micron and Intel fall 4.5% and 4.6%, respectively. In the hour at Monday’s open, the two cases that moved in the same direction continued, while one moved slightly in the opposite direction. The sample is too small to form a pattern, but the leveraged positions’ weekend risk is facing real-level ~4% volatility.
Fourth set: longs have been paying. Over 31 days, the annualized average funding rates for the nine underlyings range from 2.9% to 8.2%, with the positive-hours share from 79% to 96%. Over those 31 days, NVIDIA’s cumulative payout is 0.43%.
All four sets put together: on-chain, this turns that batch of stocks into a leveraged market that doesn’t shut, where longs keep paying, and where pricing is still set by New York. The heaviest bet is in storage.
Data: Hyperliquid public API, August 23 to September 23; snapshots at 23rd 14:25 and 24th 00:23; in-market hours taken from UTC 13 to 20 on weekdays. This dataset represents only that moment.
$BTC #AI股持续上涨还有哪些投资机会 #代币化股票
First set: the positions are pushed into storage, not NVIDIA. At 00:23 on September 24, for the whole market’s 124 underlying assets converted at mark price, the largest open interest belongs to SK hynix Korea stock perpetual with $304 million; next are Micron with $177 million, hynix ADS with $172 million, and Intel and NVIDIA at $142 million each. The two hynix contracts plus Micron and SanDisk (US$122 million) total $774 million for three storage companies—more than five times NVIDIA.
The last three sets use nine US-listed underlyings I selected: NVIDIA, Microsoft, Google, Meta, Amazon, Micron, Intel, Tesla, and a 100-stock index. The window is 31 days from August 23 to September 23.
Second set: it’s running 24/7, but the price isn’t moving 24/7. NVIDIA’s 31-day trading value shows 47.4% occurs after the Wall Street close; by hour, during the opening session the hourly trading value is $7.64 million, while in the closing session it’s only $1.83 million—about 4.2 times less. The opening-session average amplitude is 0.79% per hour, versus 0.28% for the closing session. The nine underlyings’ off-exchange (OTC) amplitude is only 20% to 50% of the on-exchange one. The prices being “found” are still those few hours when New York opens.
Third set: weekends move. Excluding Labor Day weekend, among the most recent three weekends (24 stock samples), the move from Friday’s close to Monday midnight is less than 1% for 16 instances. Over 2.5% happens three times: Intel up 2.9%, while Micron and Intel fall 4.5% and 4.6%, respectively. In the hour at Monday’s open, the two cases that moved in the same direction continued, while one moved slightly in the opposite direction. The sample is too small to form a pattern, but the leveraged positions’ weekend risk is facing real-level ~4% volatility.
Fourth set: longs have been paying. Over 31 days, the annualized average funding rates for the nine underlyings range from 2.9% to 8.2%, with the positive-hours share from 79% to 96%. Over those 31 days, NVIDIA’s cumulative payout is 0.43%.
All four sets put together: on-chain, this turns that batch of stocks into a leveraged market that doesn’t shut, where longs keep paying, and where pricing is still set by New York. The heaviest bet is in storage.
Data: Hyperliquid public API, August 23 to September 23; snapshots at 23rd 14:25 and 24th 00:23; in-market hours taken from UTC 13 to 20 on weekdays. This dataset represents only that moment.
$BTC #AI股持续上涨还有哪些投资机会 #代币化股票