Post

峰哥的交易日记
峰哥的交易日记
Show original
570美元的AMD,你追不追? 黄仁勋那边刚出幺蛾子,资金疯了一样涌进AMD,单周暴涨30%——但你今天一追,盘前就回吐了。这波到底是AI芯片的接力赛,还是狗庄借NVDA的东风割一波追高的? 先看表面:利好炸裂,但价格开始犹豫。 过去一周涨超30%,从400出头的低点硬生生拉回570,把10月658高点以来的失地收复了大半。但今天站上570之后,盘前和永续都开始回吐——典型的“强势后获利盘砸盘”剧本。 K线告诉你:50日、200日均线全站上了,MACD柱体转正,RSI 57-58中性偏强——结构没坏,但短线确实热了。 第一件事:这不是AMD变强了,是NVDA出事了。 这波涨,AMD得谢谢黄仁勋。 NVDA董事会那边一有风吹草动,资金立马找“替代品”,AMD、Broadcom全被抬起来。AMD被硬生生包装成“AI芯片的Plan B”。 但你要清醒——AMD还是追赶者,不是领跑者。 不过话说回来,追赶者也有追赶者的好处:NVDA一家独大吃肉,AMD只要喝口汤,放在570这个价位上,就是几十个点的弹性。 什么叫轮动?就是老大打个喷嚏,老二直接起飞。 第二件事:财报是真的硬,不是画饼。 11月3日Q3财报: 营收134亿美元,同比+45% 数据中心业务85亿美元,同比+97%——翻倍了 Q4指引170亿美元,同比+66%,还在加速 这不是“预期”,这是已经落袋的钱。 EPYC服务器CPU在抢Intel的份额,Instinct MI400系列GPU在抢NVDA吃剩的份额,Helios机架方案开始放量。BofA直接把目标价提到700,共识目标658。 AMD现在讲的故事是:CPU我有份额,GPU我有性价比,机架我有一整套。 这叫“端到端AI平台”,不是单卖一颗芯片。 第三件事:宏观在拖后腿,这是最大的雷。 10年期美债收益率5.12%——这个数字对高估值成长股就是一把刀。 AMD的beta大概2.5,宏观一紧,它跌得比大盘狠一倍。AI资本开支的可持续性质疑、数据中心电力审批、中期选举后的政策不确定——这些雷都还埋着。 基本面是发动机,宏观是刹车。现在两只脚同时踩着。 多空对决,你自己看 一边是: Q3数据中心同比+97%,Q4指引还在加速 NVDA动荡,资金轮动逻辑还在 技术面站上所有均线,中期多头结构 分析师集体上调,共识Strong Buy 本周放量上涨,资金参与度高 一边是: 单周涨30%,短线严重过热 570盘前已经开始回吐 10Y收益率5.12%,高估值承压 beta 2.5,大盘一抖它就摔 NVDA要是快速反弹,轮动逻辑直接降温 关键位置 上方阻力:590-600(心理+前高)→ 620-640 → 658(终极前高) 下方支撑:550-560(近期低点/缺口/均线)→ 530(强支撑)→ 500-520 操作策略 短线选手: 等回踩550-560分批低吸,止损535-540下方。第一目标590-600先出一半,突破并站稳590-600再追多,止损回撤到570下方。追高的人,一个回吐就扛不住。 波段玩家: 550-560是主战场,分批建仓,目标620-640,极端看658前高。仓位控制在可承受风险的一半以内,别重仓过夜。 杠杆玩家: 永续杠杆压到5-10倍以内,资金费率接近中性但要盯着持仓变化。AMD的波动率,不是给你上20倍杠杆用的。 AMD现在就像2023年初的英伟达—— 所有人都觉得“AI是泡沫”,结果财报一个接一个打脸,股价翻着跟头往上走。 但2023年追NVDA追在顶上的人,也套了半年才解套。 570这个位置,不是不能买,是不能闭着眼睛买。 $BTC $NVDA $AMD
峰哥的交易日记
峰哥的交易日记
SanDisk at $944, dare to buy? Falling from 2350 to 944, the AI storage leader has been halved twice, with a 7x PE unwanted—yet before the November 20th stock split, a whale quietly accumulated around 870. Is this a golden pit or a guillotine? First, look at the surface: explosive earnings, but the stock price has crashed. SanDisk's latest financial report shows revenue of $11.03 billion, a year-over-year surge of 378%; data center revenue at $4.8 billion, up 61% quarter-over-quarter; gross margin at 85%; non-GAAP EPS of 49.51, all exceeding expectations. Analysts' target price midpoint is over 2000, with Evercore at 2500. And the result? The stock price dropped from 2350 straight down to 944, a 60% retracement. Earnings are explosive, but the stock price is unrecognizable—even to its own mother—this is the current state of the AI storage leader in 2026. First point: The AI storage leader with a 7x PE is being sold as trash by the market. SanDisk's current PE (TTM) is only 7-8x, with forward PE at 3-4x. Compare that to: how many times is NVIDIA trading? AMD? A bunch of loss-making AI concept stocks? A leader supplying NAND flash for AI data centers, with an 85% gross margin, trading at 7x PE. This is not valuation; this is an insult. Why the drop? Because of NAND price cycle concerns, AI sector pullback, and storage stocks catching a cold whenever NVDA sneezes. But look at the fundamentals: enterprise SSD supply is tight, contract prices have surged, long-term contracts cover future output, and CoreWeave's earnings beat expectations driving AI chain rebound. The market trades "sentiment," the company delivers "performance." Second point: The stock split is not a positive; it’s a truth-revealing mirror. On November 20, SanDisk’s 8-for-1 split takes effect. Binance SNDKB and perpetual contracts adjust simultaneously: price divided by 8, position multipliers change, contract sizes change, spot/leverage have pause windows, and perpetual contracts have adjustment periods. Many get excited hearing "stock split": "Liquidity improves, retail investors enter, good news!" Wake up. A stock split doesn’t change company value, only the per-share price. Its real effect is: amplifying volatility, attracting short-term funds, triggering bots to rebalance, and causing leveraged players to liquidate. Before the split (Nov 18-20), liquidity windows and forced liquidation risks are real. If you go heavy naked through this minefield and get blown up, don’t blame no one warned you. Third point: Technically oscillating between 870-1000, 944 is slightly above the midpoint. Falling from 2350, the daily chart forms lower highs and lower lows, a bearish oscillation. But in early November, the 800-870 zone showed strong support (demand zone + Fibonacci retracement + trendline convergence), rebounding to 950-1000 before facing resistance and pulling back. Currently at 944, right at the upper midpoint of the range. Support: 900-920 → 870-880 (24h low zone) → 800-820 (strong demand/previous low) Resistance: 950-980 → 1000-1050 → 1100-1150 Volume expands on rebounds, remains active on pullbacks, Binance perpetual daily volume often exceeds tens of billions. Open interest is high, leverage participation is strong, funding rates near neutral—bulls and bears relatively balanced but sentiment sensitive. Breaking above 950 and holding is bullish repair; falling below 870 tests 800. Bull vs. bear, you decide: On one side: AI data center NAND demand explosion, tight enterprise SSD supply, contract prices soaring Revenue up 378% YoY, 85% gross margin, 7x PE, forward 3-4x Analysts strongly buy, target price midpoint 2000+ Long-term contracts cover future output, strong cash flow, buybacks support Whale accumulation traces around 870-900 On the other side: 60% retracement from 2350, bearish technical structure NAND cycle concerns, potential oversupply after capacity expansion NVDA/AI chain volatility short-term transmission November 20 stock split adjustment window, volatility and forced liquidation risk Relatively weak consumer mobile/PC demand Trading strategy Short-term players: Buy lightly in batches on pullbacks near 900-920 or 870, stop loss below 850, targets 950-1000/1050. Reduce or hedge lightly at 970-1000 rebound, stop loss above 1020, target back to 900. Reduce positions before the split (Nov 18-20), don’t gamble on the adjustment window. Breakout confirmation: Daily/4H close above 980-1000 with volume → add long, target 1100+. Effective break below 870 → wait or short lightly, target 800. Mid-term players: If bullish on AI storage supercycle, build base positions in support zone (spot/low leverage), after split use adjusted price for dollar-cost averaging, target 1500-2000 (pre-split equivalent). Use trailing stops to protect profits. SanDisk now is like NVIDIA in early 2023— Earnings have exploded, but the stock price is still playing dead. Everyone says "AI bubble will burst," but what happened? At 2350 you thought it was expensive, at 944 you don’t dare to buy. After the split, when retail enters, institutions cover, and NAND contract prices rise again— You’ll say again: "I should have bought at 870." $BTC $NVDA $SNDK

Disclaimer: OKX Orbit content is provided for informational purposes only. Learn more

Replies

No comments yet. Be the first to reply!