Oli.
Oli.
🍓Web3投研 🍑人工智能 🚀《干翻狗庄》系列工具作者
5Following
220followers
Feed
Feed
On October 2nd, the yield on the U.S. 10-year Treasury closed at about 5.28%, and the 30-year at about 5.63%. The employment report was clearly weak, yet the cost of long-term borrowing did not drop significantly.
This divergence hits the housing market harder than it does market software. Mortgage rates do not mechanically follow a single policy expectation change; they are also influenced by the long-term bond market and loan pricing. Traders may feel the pressure to raise rates has eased, but prospective homebuyers might still see heavy monthly payments.
Households with existing low-interest loans may also be reluctant to move. Selling an old home and refinancing could increase interest burdens. Both potential buyers and sellers hesitate, so improved policy expectations rarely translate immediately into active transactions. This is a concrete way high long-term rates affect demand.
Therefore, I am reluctant to summarize financial conditions simply by whether the next rate hike will happen. Long-term corporate financing and household mortgage loans each have their own pricing, and what borrowers actually pay ultimately depends on whether those prices fall.
The same applies to crypto investors: macro news can cause prices to react first, but financing pressures in the economy may not yet be relieved. I consider whether long-term rates can continue to decline as another key observation point, and I won’t dismiss it just because of a weak employment report.
#非农降温难压美债收益率，长期利率压力仍在
This time, the SEC's custody proposal adds options for registered investment advisers and regulated funds: under certain circumstances, they can self-custody, and it also proposes allowing state trust companies to undertake related custody work. Note, this is still a proposal and cannot be taken as all institutions having unconditional approval.
I support giving institutions more choices. Some assets lack suitable custody services, and rigidly applying old rules can indeed cause investment strategies to get stuck in operational steps.
But once institutions manage clients' private keys themselves, many specific issues can no longer be pushed to third parties. Who has signing authority? How to revoke permissions after a responsible person leaves? Can backups truly restore? Can investment managers bypass others to transfer assets? These arrangements must withstand scrutiny and cannot rely solely on "our technology is strong."
Retail investors who manage their own wallets bear their own risks. Institutions that custody clients' money must also explain how authority and responsibility are separated. Technically being able to put coins into a wallet is still far from reliably managing client assets.
What excites me about this news is that institutions can finally design custody processes more aligned with on-chain assets while clarifying responsibilities. Regulators provide operational space, so internal controls must keep pace. When the formal rules and implementation plans come out, I will pay special attention to how independent audits are conducted, not just which institutions announce entry.
#SEC加密资产托管新规，拟放宽机构自托管限制
NVIDIA's stock price hits a new high again, with a market value of about $5.65 trillion on October 2. Seeing this scale, the first thing I think of is: some people believe they have diversified investments, only to find that several accounts are actually betting on the same thing.
One account buys a tech index, another buys a semiconductor fund, and then picks a few AI industry chain stocks. The product names differ, but the funds' dependence on AI capital expenditure may be increasingly similar. To know exactly how much overlap there is, just check the fund holdings; no need to guess based on feeling.
Market-cap weighted indexes give higher weight to big companies, which also means that even if investors are not actively chasing NVIDIA, their portfolios may still be influenced by it. When the company continues to grow, this overlap feels comfortable; but if one day the market starts doubting customer budgets, several investments that were thought to be diversified may suffer together.
I admit NVIDIA's business performance is strong, and its rise to this level has a solid operational foundation. But a company's excellence and a portfolio's suitability for oneself are two different issues. You can't assume that every tech asset you hold is safer just because the leader hits a new high again.
Rather than obsessing over whether $6 trillion is the limit, it's better to first check your holdings: if AI industry growth expectations are downgraded, how many assets will be affected simultaneously? This answer is more useful than guessing the next round number threshold.
#英伟达股价再创历史新高，市值逼近6万亿美元
The G7 plans to release up to 100 million barrels of crude oil and refined products within four months. Roughly averaging over 120 days, this equates to about 830,000 barrels per day. The actual arrangement will release some diesel earlier; this calculation is just to help us understand the scale and should not be taken as a daily execution plan.
Seeing "100 million barrels" can indeed be reassuring, but the market consumes fuel every day. To judge whether this measure is sufficient, the total volume needs to be converted into release speed and then compared with the supply gap. Focusing only on a large number can easily lead to overestimating how long it can sustain.
There is also an easily overlooked aspect this time: the G7 has committed to avoiding restrictions on energy exports among members. If reserves are released on one hand while each member shuts down exports on the other, the relief the global market receives could be significantly reduced. Whether inventories can smoothly reach areas with fuel shortages also affects prices.
I think this arrangement has value, especially in easing short-term fuel tightness, but it will not immediately turn energy costs optimistic. What happens after four months and when the consumed reserves will be replenished still need to be faced.
For traders, the inventory release announcement can quickly change sentiment; for transportation companies, pressure only truly eases when procurement bills actually decrease. What is worth tracking next is the actual volume released—don’t let the same "100 million barrels" headline repeatedly create a sense of novelty.
#美伊局势持续紧张，G7将释放最多1亿桶储备
The increase in total crypto market capitalization does not mean that the same amount of money has just entered the market. This misunderstanding is especially prone to arise when ETFs show outflows: on one hand, people say institutions have withdrawn, while on the other hand, the market cap has clearly increased, and both sides feel they hold ironclad evidence.
Market cap is calculated by multiplying the latest price by the circulating supply; when the price changes, the value of existing assets is revalued. ETF net flow statistics, however, track subscriptions and redemptions—these two tables answer completely different questions.
Dates must also be aligned. Farside shows that on September 30, BTC and ETH spot ETFs indeed had simultaneous net outflows; on October 1, BTC had already returned to net inflows, while ETH was still experiencing outflows. On October 2, some product data in the table was missing, so the temporary totals cannot be taken as final results, and old titles should not be used to draw conclusions for the latest trading day.
My requirements for this kind of capital flow news are actually quite low: first, clearly state the statistical date. If the date of the money is not even matched, starting to explain institutions' long-term beliefs is a bit premature.
The cooling of capital flow deserves attention, but to judge whether the market trend can continue, we still need to see if new buying can continuously absorb selling. An increase in market cap can describe the market becoming more expensive but cannot alone prove that more cash has been invested. Without clarifying this, it is easy to mistake price increases for a financial safety cushion.
#BTC、ETH现货ETF同步转流出，资金热度降温
Nonfarm payrolls in September increased by only 29,000, with the healthcare sector contributing 17,000 of those. Putting these two figures together, the feeling is colder than just looking at the unemployment rate: other industries fluctuate, and the remaining new jobs are already few.
The BLS also mentioned that the financial sector lost about 7,000 jobs, and most major industries saw little change in employment. I don't really like directly translating such reports as "bad news for the economy, good news for crypto." A slowdown in hiring primarily means it's harder to find jobs and households are more cautious about income expectations. Policy expectations might push assets up for a while, but whether ordinary people dare to increase consumption is another matter.
Healthcare demand is relatively stable; it can provide jobs but cannot prove that all industries are expanding. If employment growth continues to concentrate in a few sectors, even if the total nonfarm number improves, we need to look more closely at the composition.
This report makes me cautious but not to the point of calling a recession. For the crypto market, the excitement from a smaller rate hike and concerns about economic weakness may coexist. Those eager to leverage on good news should first think clearly: are they betting on looser policy or genuinely stronger demand? These two judgments may lead to different future trends.
#美国9月非农仅增2.9万，失业率升至4.2%
🟢 Oli Daily Brief｜2026.10.04
Core coins continue a mild rebound, but the market has not yet entered a full Risk-on phase. In the past 24 hours, BTC, ETH, and SOL have all risen simultaneously, with SOL showing the strongest performance; however, at the same time, the total crypto market cap still declined, and overall trading volume noticeably contracted. Currently, the market looks more like: weekend low-volume oscillation + BTC capital dominance + relative strength in SOL + rotation in RWA/AI payment narratives. 📊 BTC remains near $85,000, SOL retests 120 As of 04:43 HKT: BTC: $84,854, 24h +0.74% ETH: $2,686.67, 24h +0.90% SOL: $119.90, 24h +1.82% Total crypto market cap: $2.903 trillion, 24h -1.83% BTC market dominance: 58.62% Fear and Greed Index: 67, Greed Previous: 72. Approximately $58.7 million liquidated across the network in 24 hours Compared to previous liquidations in the hundreds of millions, leverage liquidation pressure has clearly eased. But the most notable divergence today is: BTC, ETH, and SOL all rose, yet the total crypto market cap still fell. Meanwhile, overall market trading volume significantly decreased. This means the current rebound is mainly concentrated in core assets and a few hot coins, not a broad altcoin rally. Among the top 60 non-stablecoins by market cap: PUMP: +18.54% became the strongest performing asset
After the NEAR Intents attack, the team stated that they have identified the attacker, provided a 48-hour window to return the funds, and promised full compensation to affected users. The preliminary disclosed loss is about $3.8 million. The information is more complete than when the incident first occurred, but identifying the attacker, recovering the funds, and completing compensation are still at different stages.
Currently disclosed issues involve the interaction between Omni deposit and withdrawal infrastructure and the Intents contract, which cannot be directly expanded to mean the entire NEAR underlying network was compromised. Pinpointing the problematic link helps assess the risk; however, for affected users, the most urgent concern is when their funds will be restored.
I support the team’s initial commitment to compensation, as this at least gives users a responsible party to hold accountable. But after the commitment, the scope of compensation and execution timeline need to be clarified. Whether the attacker cooperates should not be a reason for users to wait indefinitely.
This incident also made me reconsider the convenience of cross-chain products. The more the front end integrates operations smoothly, the easier it is for us to forget which systems a single transaction actually passes through. Users see one confirmation, but the backend may involve multiple interaction steps; security assessments cannot stop at familiar brands alone.
It is a bit early to debate whether the coin price has fallen too far. When services will resume, how the repair report explains the problem, and whether compensation has been received are all more effective at restoring trust than shouting at the attacker. Hopefully, the next update will provide clear execution results.
#NEAR生态协议遭攻击致币价下跌近10%
OpenAI plans to raise at least $30 billion at a valuation of about $1.4 trillion, with discussions still in the early stages. Yesterday everyone was calculating the valuation; today I want to look at the sources of funding.
Reuters reported in September that SoftBank launched dollar and euro bond financing, planning to use part of the funds to invest in OpenAI. This does not mean that this round of financing will necessarily follow the same arrangement, but it reminds us that investors in AI companies may also need to raise money from the capital markets.
This adds another layer of cost to investment decisions. The company itself must prove that the business is worth investing in, while investors have to bear their own financing costs. If long-term interest rates remain high, even if AI is still promising, one must calculate how much interest will be paid during the waiting period for returns.
I am a bit cautious about this detail. Seeing large institutions willing to invest makes it easy to think that funds are sufficiently abundant; but subscription willingness, fund arrival, and investors' subsequent financing arrangements all have their own constraints. The brand is big and will not make borrowing costs disappear.
This also makes me more curious to know whether, after the new capital injection, the speed at which the business generates cash can keep up with expenditures. Capital can help a company get through the expansion period but cannot permanently replace operating returns.
I recognize the potential of AI products, but the larger the industry chain, the more necessary it is to include the sources of funds in the discussion. Simply looking at the amount of financing can easily lead to seeing only that investment is increasing, without seeing who is bearing the long-term costs of these investments.
#OpenAI拟1.4万亿美元估值融资300亿美元
Atkins has included crypto asset fundraising rules, custody, and on-chain transactions in the regulatory agenda, and I support continuing to advance this. But this time I want to look at it from the investors' exit perspective: once fundraising channels become smoother, can the money invested be withdrawn under the conditions previously understood?
The fact that a token can be transferred between wallets does not alone indicate there is sufficient buying demand. Even if transfers are allowed by rules, investors may still fail to find willing buyers. Issuance channels and secondary market trading are two things that need to be developed separately.
I hope future rules and product descriptions will state these matters more plainly. Whether holders face transfer restrictions and whether early exits are arranged should be visible before subscription. Specific guidance has not yet been published, so we cannot pre-judge future provisions.
On-chain tools can indeed reduce certain issuance and record-keeping costs, but if purchasing becomes very easy while exit conditions require digging through a lot of information, it is not friendly to ordinary investors. The smoother the process, the easier it is to overlook long-term commitments in just a few clicks.
My expectation for on-chain fundraising is to make it easier for suitable projects to find funding and for contributors to clearly understand what they are accepting. Making the subscription page look good is not difficult; the real challenge is how to handle the product when it is not well received. I hope this part will not be hidden again in fine print that no one wants to read.
#SEC主席Atkins称将推进链上募资规则明确化