#BTC surged then pulled back, options expiry amplifies the key level battle $BTC surged from around 62,000 to 81,000 in one go this week, reaching a high of 81,200 on August 25 before pulling back. Yesterday it surged again to around 80,800 but couldn't hold, then retracted to fluctuate between 80,200–80,500. This is not a crash, but a surge followed by a pullback. Today the real volatility is in options. On Deribit, about 81,700 contracts with a notional value of $630–640 million expire at 4 PM. Calls outnumber puts, with a PCR around 0.83. Positions cluster at 75,000 and 80,000 strikes: calls at 75,000 have a notional of about $236 million, at 80,000 about $157 million, with over $500 million within 5% of the current price. The biggest pain point remains between 68,000 and 70,000, far from the current price, so don’t use that as a reason for a dump. My view: 81,000 is the first solid resistance in this rebound. Bears are unloading here, while call options are stuck at 80,000. Market makers hedging will keep pressing the price repeatedly at this threshold. This is not evidence of a trend ending, but friction due to expiry. To be blunt: tonight is also the Jackson Hole symposium. Options expiry combined with macro speeches is likely to cause swings both ways. If 80,000 breaks, look to 78,500; only a real break above 81,200 confirms this rebound. Don’t max out leverage before expiry; the key level battle is about who cracks first, not who shouts louder.

#交易之声：你的经验值得被听到 Q: When you trade, do you calculate the total risk of your account in advance? Yes. And what I calculate is not the stop loss of a single trade, but the total account risk. A single trade losing 2% looks well-behaved. The real trouble usually comes when three trades lose together: spot, futures, thematic coins in the same direction, plus US stocks and USD turning against you on the same night. At this point, 2% × 3 is not 6%, because the correlation suddenly becomes 1. I personally calculate three things in advance: First, if all positions hit stop loss simultaneously, how much can the account lose at most. I usually cap this number at 6% to 8% of principal; if it exceeds that, I reduce positions. Second, how much is occupied by the same logic. If all bets are on AI, or all on interest rate cuts, or all on altcoin catch-ups, that’s not diversification, it’s one bet split into three parts. Third, whether there is an event tonight. On days like PCE, Nvidia, Jackson Hole, total risk needs to be cut further because slippage and gaps can break through paper stop losses. My view is simple: survive first, then debate whether the view is right or wrong. Position size is the opposite of conviction; the stronger the conviction, the smaller the position should be. Many people only calculate "Can I afford to lose this trade?" but not "If these trades all fail together, can I still trade tomorrow?" The latter is the total risk.

#财报观察员：英伟达领衔，AI回报进入验证期 Nvidia $NVDA Q2 beat again. Revenue reached $96.2 billion, more than doubling year-over-year by 106%; data center revenue was $89 billion, up 117% year-over-year, accounting for 90% of the total. Adjusted EPS was $2.22, while the market expected about $2.10. Gross margin was 75%. The Q3 guidance is around $108 billion ±2%, not including China data center compute revenue, whereas the market had only expected just over $104 billion. What really moved the market wasn’t the quarterly numbers, but the CFO’s statement on the spot: revenue is expected to grow about 70% more by fiscal year 2028. Wall Street had only dared to forecast a bit over 40%. Jensen Huang’s original words were even more aggressive — demand is even greater, constrained by the supply chain. After-hours trading first dropped, then rallied, ending up more than 4% higher. My view: this is not a “barely beating expectations” scenario; it’s solid proof that demand hasn’t peaked yet. Cloud providers are still buying, and enterprise growth is even stronger than hyperscale, indicating AI hasn’t stopped at just the training phase. Q3 will very likely officially cross the $100 billion single-quarter revenue mark. I also have to be blunt. The gross margin guidance dropped to 74%, memory price increases have already started to bite; China revenue this time was directly counted as zero, which isn’t a fully priced-in negative but a lingering risk. The easiest mistake after beating expectations is to treat the 70% growth as a reason to chase recklessly right now. Short-term sentiment will be supported by this earnings report, but mid-term it still depends on two things: whether Rubin ramps up smoothly and whether gross margin can hold. The numbers won, but the valuation battle has just begun.

#美国核心PCE持平上月，沃什杰克逊霍尔讲话如何定调？ US July core PCE year-on-year is 3.3%, exactly the same as last month, month-on-month 0.2%. Overall PCE is still 3.7%. Not exceeding expectations, nor giving any face. My view is simple: this is not a signal to ease. Core stuck at 3.3% for two consecutive months, already the 65th month above 2%. Services are still rising, goods are falling, real consumption is flat, income up 0.4%. Demand hasn't collapsed, inflation hasn't gone away. Pricing for a September rate hike is about 30-40%, which I think is reasonable, neither a must nor off the table. On Friday at the Jackson Hole speech, my view is clearer—don't expect him to say whether September will have a hike. He hates forward guidance anyway, and as the new chair's first time, the safest is to reiterate the 2% target, emphasize risks remain, and keep the door open for rate hikes. This sounds like correct nonsense, but it's enough for the market. What I really fear is that he talks vaguely. All about financial innovation, long-term framework, less intervention, the bond market will take it as him not wanting to bear 3.3%. If the long end stirs, short-end pricing is wasted. My own trading view: don't take sides tonight. If tone is hawkish, watch the dollar and US bonds; if ambiguous, gold is more comfortable, crypto don't rush to front-run. The ball is at Powell's feet, wait for him to speak before pricing. $BTC $ETH

#BTC breaks through $80,000, can it hold the new level? Bitcoin $BTC surged past 80,000 these days, reaching an intraday high near 81,200 for the first time since mid-May. But it didn't hold, now it has fallen back to fluctuate around 78,000 to 79,000. This rally was rapid, gaining more than 20% in a week. The reasons are basically a few things: The US Treasury said it will increase repurchases of long-term bonds, the market sees this as a form of easing, and money is flowing into assets like Bitcoin; the spot ETF attracted nearly $2 billion last week; shorts were squeezed out, and the more it rose, the more people bought. 80,000 is a round number, good for appearances, but it hasn't truly become support yet. There are still trapped positions near the May high, the rise was too fast and overbought, so some taking profits is normal. These days, attention is also on the US PCE inflation data and the Jackson Hole meeting; once news comes out, the price may fluctuate again. Whether it can hold, don't just look at the round number. First see if it can hold around 79,000, then see if ETF money continues to flow in. If it holds, there’s a chance to look at 82,000 and 88,000; if it doesn't, a pullback to 75,000 to 76,000 is very likely. It's still far from last year's high of 126,000, so it's a bit early to call a new bull market. Short-term volatility will be large, don't chase the highs, watch volume and capital flow, which is more reliable than focusing on round numbers.

In the next 30 days, expect $BTC to fluctuate widely, first digesting then leaning bullish, no chasing highs. Current price is about 78,900, just pulled from 63,000 to 81,200, ETFs are still buying, but RSI is overbought, resistance at 80,000–83,000. This week also has Jackson Hole and options expiry. Main range is 75,000–83,000, strong target 85,000–87,000, weak target 72,000. 1 million U: Spot 32%, Dollar-cost averaging 18%, Grid 16%, Dual currency earn 12%, Earn coin 8%, Options 6%, Futures 3%, Flexible 5%. At current price, buy 180,000 spot first, place remaining 140,000 orders at 77,000/75,000/72,800; DCA 20,000 weekly, plus add at 76,000/74,000/71,500. Grid trading between 74,500–82,500, close if closing breaks 83,000 or 74,000. Dual currency earn: buy low with 60,000 U at 73,500–75,000, sell high 40,000 BTC at 84,500–86,000. Options only buy, no naked selling; futures max 3x leverage, stop after losing 30,000. Reduce spot by one-third at 85,500–87,000; if close falls below 72,000, reduce by half and lower risk. Invalid signals: break below 72,000 with continuous ETF outflows, or fail to recover 75,000 after speeches. #OKX星球话题来啦

#Anthropic estimates a $30 trillion market, can the IPO narrative be fulfilled? $ANTHROPIC directly calls out a potential market size of $30 trillion, even higher than $SPCX's previous $28.5 trillion. A company founded five years ago is setting its ceiling close to the entire annual GDP of the United States. The short-term narrative looks great: Q2 revenue has already exceeded $11.5 billion, annualized to $65 billion, with plans to reach $190 to $200 billion by 2028. Telling the story at this growth rate, a $2 trillion valuation and raising $100 billion seem plausible. But this $30 trillion TAM essentially counts "all the work AI could theoretically do." The combined revenue of 191 tech companies in the S&P 1500 last year was only $2.4 trillion; this denominator is so large it's almost impossible to falsify, and very hard to disprove. What really determines whether the narrative can be fulfilled is not how big the TAM is, but two things: first, whether enterprise customers are willing to continuously pay for high-priced APIs, and second, whether gross margins and computing costs can hold up. Growth is fast now, but so is cash burn, and competition is still OpenAI and Google. If the valuation is only supported by painting a big market, once growth slows down after listing, the stock price will look very bad. I personally prefer to see if it can turn the current high growth into relatively stable enterprise revenue and positive cash flow. The story can be told big, but ultimately it has to come down to orders and profits.

#JaneStreet holds 5% of SanDisk, AI storage valuation under renewed scrutiny This time, JaneStreet added 7.41 million shares of SanDisk $SNDK, with a market value of about 9 billion, making it their second largest single position. Over the past year, SanDisk has surged over 3000%, then pulled back more than 30% from its peak, showing significant volatility. The quant giant still dares to hold a heavy position now, indicating they at least don't think the AI storage logic has run its course. The short-term room for further large gains is indeed narrowing, but the mid-to-long-term logic remains intact. AI's demand for storage is not a passing trend; both training and inference will continuously consume large amounts of enterprise-grade NAND. What truly matters to watch are long-term orders and profit margins, not just pure demand figures. The storage industry fears strong demand but prices being crushed by new capacity. Now that major clients are willing to sign multi-year contracts locking prices and volumes, profit visibility is significantly improved. Only by locking in high gross margins for several years can valuations stand firm. In terms of positioning, I would prioritize SanDisk and Micron. SanDisk focuses purely on NAND, deeply tied to several US cloud providers, and is pushing high-bandwidth flash for AI inference; Micron is more balanced, with both HBM and NAND, offering stronger risk resistance. SK Hynix leads technologically but has higher geopolitical and client concentration risks, so I would rank it a bit lower. Whether AI storage can truly deliver depends on whether major clients' capital expenditures can continue and whether vendors can lock in high prosperity through contracts. Relying solely on shortage-driven price speculation will eventually fall back; if order locking plus technological iteration can form, this cycle may be longer than previous ones. I won't blindly chase highs and will wait for a comfortable pullback position

#美启动对伊经济孤立，油价为何回落？ On Monday, U.S. Treasury Secretary Janet Yellen announced the launch of an "economic isolation operation" against Iran, expanding sanctions to five major areas including digital assets, gold, and shipping, and warned that countries continuing to do business with Iran could be kicked out of the dollar system. Logically, this would raise geopolitical premiums, yet oil prices have clearly fallen—on Tuesday, WTI dropped more than 3% to $82.36, and Brent fell nearly 4% to $88.58. The reason is not complicated. The market's biggest fear before was military escalation and physical disruption of the Strait of Hormuz. Although economic sanctions may compress Iran's exports in the long term, their short-term impact on actual supply is much less than another war. The U.S. shifting from "hard military" to "soft economic" measures directly reduces the probability of the worst-case scenario, quickly squeezing out risk premiums. Additionally, the sanctions details leave room: major buyers like China were not immediately named, nor was a clear timetable for secondary sanctions given. Coupled with Iran discussing temporary shipping routes with Oman and the U.S. considering sending diplomats back to the Middle East—signals of easing—traders are more inclined to believe the negotiation window is still open. Profit-taking after consecutive rises further amplified the decline. Simply put, the current market pricing logic is: the supply threat from economic pressure is temporarily less than that from military conflict. As long as there is no more severe physical disruption in the strait, oil prices are more likely to oscillate downward rather than surge unilaterally. $CL $XAU $BTC

#交易之声：你的经验值得被听到 In trading, the real challenge isn't taking profits, it's cutting losses. Many people say "I set stop losses and strictly follow them," but when floating losses appear, the first thought is often: "Let's wait a bit longer, it might come back soon." "Just bought and already losing, it must be temporary." "Cutting now means all the effort was for nothing." So small losses turn into medium losses, and medium losses turn into big losses. Taking profits isn't that hard. The profit is there, people feel somewhat satisfied, and after giving back some gains, most are still willing to act. What really traps people is losses. Losses touch the deepest chord of human nature—loss aversion. Admitting you're wrong is much more painful than earning a little less. So we prefer to hold on, using the illusion of "it will come back" to avoid the moment of confirming failure. This is the disposition effect in behavioral finance: you want to cash in winning trades as soon as possible, but you cling tightly to losing trades. Those who truly survive long-term have almost turned "timely stop loss" into muscle memory. It's not about willpower in the moment, but about setting the rules in advance and executing them like a robot. Taking profits can be trained, relying on discipline. Cutting losses is hard because it requires going against human nature. Next time before placing an order, ask yourself: if this trade really hits the stop loss, can you press it without hesitation? If the answer is hesitation, then this trade shouldn't be opened. $BTC #BTC突破80000美元，能否站稳新关口