Yesterday, U.S. spot Bitcoin ETFs recorded a net outflow of $469 million, while Ethereum ETFs saw a net outflow of $30.3 million.

According to on-chain analyst Ai Yi (@ai_9684xtpa), the largest short address of Changxin Memory Technologies (CXMT) is still adding to its position. Two hours ago, this address added 2 million USDC in margin to Hyperliquid and continued placing limit sell orders via TWAP. Currently, the address holds 2.072 million CXMT in 1x leveraged short positions, with a notional value of approximately $12.78 million, an average entry price of $6.4434, and an unrealized profit of around $563,000. Additionally, there are still 435,677.8 CXMT limit sell orders at $6.11 that remain unfilled, with a total value of about $2.66 million.

According to monitoring by Onchain Lens, Hyperliquid generated approximately $1.4 million in fees over the past 24 hours, burning 20,640 HYPE tokens valued at around $1.2 million. To date, Hyperliquid has cumulatively burned 47.27 million HYPE tokens, worth roughly $2.76 billion, accounting for 4.73% of its maximum total supply of 1 billion HYPE.

According to on-chain analyst Ai Yi (@ai_9684xtpa), BitMEX founder Arthur Hayes received 644.723 ETH from FalconX 8 hours ago, valued at approximately $1.2 million. The transaction size matches the USDC he transferred three days prior, strongly indicating a buy operation. Since July 15, Hayes has accumulated 3,914.84 ETH via multiple market makers and trading platforms, at an average purchase price of $1,908.86, resulting in an unrealized loss of roughly $113,000.

SK Group Chairman Choi Tae-won stated that South Korea must become an "AI-native country," adding, "The goal is to ensure every person has at least one AI agent." He noted that if this goal is achieved, South Korea will become a global AI testbed where various AI technologies can be tested and deployed. Choi added that SK Group will provide more opportunities for AI development through continuous R&D investment. He also emphasized that reducing AI costs is critical, saying, "Currently, AI costs are too high. We must cut costs through various means so that more people around the world can benefit from AI." Choi further stated that South Korea can lower AI costs by expanding its memory chip production capacity and building more AI data centers, while establishing a global AI data center hub connecting the U.S., Europe, and Asia. (Source: Jinshi)

Renowned trader Killa posted that roughly 80% of cycle top indicators never triggered during the last Bitcoin bull run, and a similar situation could occur with bottom indicators in the future, meaning market bottoms cannot be judged by mechanically relying on historical signals. He noted that the supply of long-term holders currently in a loss state has exceeded levels seen during the FTX collapse and is approaching those of the 2018 bear market. Bitcoin’s realized price is currently around $50,000; in every past cycle, the price has tested the realized price of long-term holders, so a return to this region remains possible. However, Killa said it should not be assumed that Bitcoin will definitely drop back to this level. Many top indicators failed to trigger in the last cycle, and some bottom indicators may also become invalid in the future. Regardless, the current level of market loss is already comparable to that during the FTX incident and the 2018 bear market.

The latest Emerson College poll shows that Democrats hold an 11-point lead over Republicans in midterm election preferences, with 53% to 42%—marking the Democrats’ largest lead in the poll series. Meanwhile, former President Donald Trump’s approval rating stands at 39%, while his disapproval rating is 57%. The poll indicates that the Democrats’ advantage stems largely from female voters, among whom they lead Republicans by 27 points. Newsweek, analyzing via its Uniform Swing Model, stated that if this national advantage translates evenly to House districts, roughly 36 Republican-held seats could flip to Democrats, resulting in a hypothetical 251-184 seat split favoring the Democrats. However, the article emphasizes that this is only a model stress test based on national polls, not an election prediction, and the actual outcome will still be influenced by factors such as candidate performance, local issues, voter turnout, and district mapping. (Jinshi)