美股投资young
美股投资young
美股科技投资者 ai狂热者 分享每日快讯. all in ai
299Following
666followers
Feed
Feed
$SOXX has fallen for nine consecutive sessions, marking the longest continuous decline since May 2006. A market situation not seen in nearly 20 years is happening in 2026, even as the AI industry continues to grow rapidly. I truly did not expect semiconductors to fall to this extent. $QQQ $NVDA $MU $SNDK
On October 5, $SOXX closed at $589.51, and by October 16, it had dropped to $493.39, a cumulative decline of about 16.3% over 9 trading days. The Philadelphia Semiconductor Index also experienced eight consecutive declines, the longest streak since the 2008 financial crisis.
But what worries me more than the decline itself is the market structure behind this downturn.
From the daily candlestick chart, $SOXX is forming a very typical downtrend. The rebound highs are continuously lower, previous supports are repeatedly broken, and every time buying support appears, it is quickly met with a new round of selling.
Especially on October 15.
That day, $SOXX hit an intraday low of $495.78, then rebounded to close at $514.71, which looked like a decent attempt to stop the decline.
However, the next day it opened lower, broke below the previous day's low during the session, and finally closed at $493.39, near the day's lowest point.
What does this mean?
It means the funds that tried to bottom-fish the day before faced losses again the next day. The market didn’t even give these funds much breathing room.
Even more notable is the trading volume. At the beginning of October, $SOXX’s daily volume was only around 4 million shares, but on October 15 and 16, it expanded to approximately 16.7 million and 14.7 million shares, respectively.
Price continuously breaking support with significantly increased volume, and rebounds failing to sustain—this is not the healthy correction structure I want to see.
Now, about options.
As of the close on October 16, third-party options data models show $SOXX’s Put/Call open interest ratio at about 1.86, with put positions clearly exceeding calls. Some Gamma models also indicate the market is currently in a negative Gamma environment, with the $500 level being a key area of concentrated option holdings.
The price has now fallen below $500.
In a negative Gamma scenario, if market makers need to hedge Delta in the direction of the trend, they may sell more as prices fall, amplifying market volatility. Of course, the actual hedging direction depends on real positions; these models only provide risk references.
But I believe there is something even more important.
There is currently no sufficient evidence that this sharp decline is mainly caused by large-scale new short selling actively dumping shares.
On the contrary, data disclosed by Goldman Sachs Prime Brokerage shows that during the recent global stock market deleveraging, long positions have sold more than shorts have covered, and AI-related stocks have been net sold by hedge funds for six consecutive trading days.
In other words, we are already seeing a very clear withdrawal of existing long positions.
Of course, we cannot conclude that shorts are not active. Public options open interest, short sale volumes, and candlestick charts are insufficient to fully reconstruct institutional real positions.
But if just the unwinding of crowded longs, risk budget contraction, and passive hedging have caused such a severe decline, then if more concentrated new short selling and credit stress spread later, the market could face a second wave of shocks.
This is what I am truly cautious about.
Many friends may not understand why I have always been bullish on AI but now repeatedly emphasize reducing positions?
Because I have always been long, but I never believe being bullish means you must hold full positions at all costs.
I still believe in the demand for AI infrastructure, advanced processes, storage, and optical communications over the next few years. I also do not think there is enough evidence yet to prove the long-term bull market is completely over.
In my judgment, this is still closer to a deep pullback within a bull market.
But a deep pullback in a bull market can still cause severe losses for many investors.
Judging the long-term direction is one thing; managing short-term risk is another.
When the market shows continuous breakdowns, failed rebounds, increased volume, and persistent macro pressure, maintaining previous full-risk exposure just because of a long-term bullish view is not reasonable.
I have repeatedly emphasized before: when it’s time to run, you really need to run; when it’s time to reduce positions, you must reduce positions.
This running does not mean everyone should liquidate at any cost, nor does it mean I am turning bearish.
It means when your position exceeds your risk tolerance, and short-term options and leverage could turn an ordinary pullback into a permanent loss, you should proactively reduce risk.
Even if $SOXX subsequently rebounds 5% or more due to overselling, I will not immediately conclude the correction is over.
What I want to see more is the price stabilizing again in key areas, rebounds no longer quickly swallowed by selling, improved volume structure, and credit markets and macro environment gradually stabilizing.
Before these signals appear, I tend to use rebounds lacking sustained buying support to continue adjusting overly high positions.
This decline has made me more certain of one thing:
Being right about a bull market does not mean you can withstand every pullback within it.
I still believe in AI’s long-term development and in companies with real technological barriers, profitability, and cash flow having great investment opportunities ahead.
But I will not ignore the risks already present just because I believe in the future.
The market may eventually rise again, but the premise is that you have enough capital and position space to participate in the next wave.
In this kind of historically extreme market, surviving is more important than proving you were always right.
After this round of semiconductor sell-off, I will start focusing on companies that have maintained relative strength throughout the sector's decline.
Many investors like to buy the stocks with the biggest drops, thinking that a company that fell 30% must have more rebound potential than one that fell 10%. But from a practical trading perspective, I don't fully agree. $SOXX $QQQ $NVDA $AVGO
A company dropping 30% might be due to previously overstretched valuations or because the market is starting to lower future earnings expectations. Another company dropping only 10% might indicate that investors are still willing to hold onto its long-term growth and cash flow.
These two price behaviors have completely different implications.
I now categorize semiconductor stocks into three groups.
The first group consists of companies with solid fundamentals, deep competitive moats, and that show strong relative strength during the decline.
For example, $TSM, $ASML, $KLAC, $AVGO.
These companies hold important competitive advantages in advanced manufacturing, lithography technology, process inspection, and custom chips.
Their long-term growth does not fully depend on market sentiment but is supported by commercialized products, customer relationships, and profit bases.
If the entire sector rebounds, these companies' gains may not be the most aggressive, but I prefer to study whether they can form a sustained upward trend.
The second group includes highly elastic sectors like memory and optical communications.
$MU, $SNDK, $LITE, $COHR.
These companies may benefit from industry demand and profit growth but are also more susceptible to cyclical expectations, valuation compression, and trading behaviors.
Once the market shows a clear risk appetite recovery, their rebound elasticity could be greater. But if subsequent adjustments are not over, these stocks may also face more pronounced secondary declines.
The third group consists of companies highly dependent on future growth expectations, financing conditions, or that have yet to stabilize earnings realization.
Even if these stocks rebound 20% after a crash, it does not necessarily mean fundamental risks have been resolved.
I will not automatically consider the biggest decliners as the most worthy to buy.
When judging rebound quality next, I will focus on relative strength.
If $SOXX rises 5%, a company rises 8%, and then when the sector pulls back, it can hold a higher low, this at least shows stronger price resilience in this rebound.
Conversely, if $SOXX has started to recover but a company continues to underperform the sector and even hits new lows, I will re-examine its earnings expectations, valuation, and market positioning.
Of course, relative strength does not necessarily mean institutional accumulation, nor can a few candlesticks alone determine a long-term trend.
I still tend to believe the October adjustment is not over, and the subsequent trend will likely alternate between rebounds and declines.
So the real observation window may precisely come after the next rebound.
Who can break previous highs? Who stops making new lows on pullbacks? Who can re-establish above key moving averages? Whose earnings expectations remain stable?
These questions are more important than simply comparing who fell the most.
For cash-rich investors, I prefer to wait for quality companies to reach reasonable valuations and better price structures before gradually building core stock positions.
For those who have already heavily invested in high-elasticity stocks, if the rebound lacks sustained capital support, I still tend to reduce risk by selling into strength.
I am optimistic about the semiconductor rebound and still bullish on AI's long-term development, but I am not prepared to treat all fallen stocks as golden opportunities.
The next meaningful uptrend to study will be those companies that, after market adjustments, can still maintain competitive advantages, realize earnings, and gradually regain buying support.
There is a very noteworthy data point in this round of semiconductor decline: retail investors are buying heavily at the bottom, but institutional funds are becoming increasingly cautious. $SOXX has fallen for 9 consecutive trading days. The problem may not be that no one is buying at all, but that the buying volume is insufficient to absorb the selling volume. $QQQ $NVDA $MU $SNDK
According to the latest disclosure from Citadel Securities, during one statistical week on its platform, retail investors net bought $9.1 billion in U.S. stocks, the highest level in 15 weeks, with $2.6 billion net bought in semiconductors and $700 million net bought in AI cloud infrastructure-related stocks.
At the same time, Vanda Research observed completely different behavior from institutional funds. Hedge fund equity exposure has declined, institutional options buying has weakened, and funds are shifting from growth stocks to defensive sectors.
This set of data basically aligns with my previous judgment on the October market.
I have always believed that the main feature of this adjustment is that existing AI bulls are reducing their positions, rather than bears having fully exerted their power. Although retail investors are willing to buy on dips, institutional investors who previously crowded positions are reassessing risks.
After the short squeeze in September ended, some stock valuations remain high. Now, with U.S. Treasury yields staying elevated and rising oil prices increasing inflationary pressure, the financing costs for AI data centers are also being questioned.
Until these issues are resolved, institutions may very well continue to reduce their risk exposure.
Retail investors buying $2.6 billion sounds like a lot, but this is data from a specific trading platform and cannot represent the net capital flow of the entire market. The continuous price decline indicates that the current buying and selling forces are still imbalanced.
Moreover, I believe this capital structure carries a risk.
If retail investors keep bottom-fishing but institutions remain unwilling to increase their positions again, every rebound may face selling pressure from previously trapped funds.
So even if semiconductors see a significant rebound next week, I would not conclude that the adjustment is over.
What really deserves attention is whether institutional funds will increase their exposure to tech stocks again, and whether $SOXX can form a higher low after the rebound.
If the capital structure does not improve and U.S. Treasury yields and oil prices remain high, I still tend to use rebounds to reduce positions.
However, there is no need to be overly pessimistic.
Current weak institutional buying does not mean they will never return. As long as macro pressures ease, valuations return to reasonable ranges, and corporate earnings expectations remain stable, capital reallocation could still drive a strong rally.
I still tend to believe the bull market is not over, but this October adjustment may need more time to digest.
What we need to be wary of now is the situation where retail investors keep buying while institutions keep reducing exposure.
Before capital truly flows back, controlling position size is more important than guessing the rebound date.
$SOXX has fallen for nine consecutive days, with a cumulative drop of nearly 12% this week, while $QQQ fell 5.5%, and $SPY only about 2%. Many people are now starting to talk about a golden buying opportunity, but my attitude remains: opportunities are indeed increasing, but there is no need to rush to exhaust all cash.
I still tend to believe this is a deep pullback within a bull market, not that the long-term growth logic of AI has collapsed. The industrial positions of companies like TSMC and ASML have not fundamentally changed due to the October decline. Some stock pullbacks are more due to previously high valuations and existing bulls continuously reducing risk exposure.
But good companies do not necessarily mean good prices right now.
Currently, the U.S. 10-year Treasury yield remains around 5%, oil prices are high, the November midterm elections are approaching, and risks in AI infrastructure financing are beginning to surface. Even if semiconductors rebound by 5% or more next week, these issues will not automatically disappear.
Especially since $SPY’s current decline is significantly less than that of semiconductors. If the subsequent adjustment continues to spread to large tech platforms, $SOXX could very well hit new lows again after a technical rebound.
So I will not assume the market has little room to fall just because of nine consecutive days of decline.
For those holding ample cash, I actually think now is a very worthwhile stage to study companies.
$TSM, $ASML, $AVGO, $GOOG, $MSFT, $KLAC — these companies’ moats come from technological accumulation, customer switching costs, ecosystems, and long-term profitability.
If their future earnings forecasts do not significantly deteriorate, but stock prices adjust further due to macro pressures, then long-term expected returns may gradually improve.
My approach is to build stock positions in batches, rather than trying to catch the absolute bottom all at once.
When prices enter a reasonable range, you can start building partial positions. If prices continue to fall afterward, reassess valuations and fundamentals before deciding whether to increase investment. If the market rebounds early, at least you already hold some positions and don’t need to chase prices.
But I will always keep enough cash.
Because what’s truly painful in a bull market is not just stock declines, but when quality companies fall to clearly more attractive prices and you no longer have funds to buy.
What the market lacks most now is sustained buying pressure. For those forced to deleverage, declines mean pressure; for those with ample cash, declines may increase choices.
I still expect a strong rebound later, but I won’t bet all my positions on an immediate rebound.
If U.S. debt, oil prices, and financing conditions do not improve significantly, I even tend to think there will be a second round of adjustments after the rebound.
Control risk in the short term, keep cash in the medium term, and look for truly undervalued quality companies in the long term.
If October ultimately produces a golden buying opportunity worth positioning for, I hope to have enough cash rather than being fully invested halfway up the mountain.
There is no shortage of stocks now; what’s lacking is patience and the right buying price.
If $SOXX suddenly rebounds 5% or even higher next week, I wouldn't find it surprising. But given the current price structure, a single large bullish candle is not enough for me to conclude that semiconductors have bottomed out. We must distinguish among three completely different upward drivers: short-covering, options hedging, and sustained new capital inflows. $QQQ $SPY $NVDA $MU
As of October 16, $SOXX closed at $493.39, down for nine consecutive trading days. Third-party options data shows the Put/Call open interest ratio is about 1.86, while estimated Gamma exposure is negative.
There is a common misunderstanding here.
High Put holdings do not mean everyone is short. Institutions holding large amounts of the underlying stock may also buy Puts for protection. Open interest alone cannot directly reveal the true net short position.
Negative Gamma also requires careful interpretation. It is a trading structure estimated based on options holdings and assumptions, not a full disclosure of market makers' actual positions.
But if market makers are indeed in a negative Gamma state, hedging trades may amplify price volatility.
When prices fall, some market makers may need to continue selling the underlying to hedge; when prices rise, they may be forced to buy the underlying to adjust Delta exposure.
Therefore, even a sharp short-term rebound may contain a large amount of mechanical hedging buying and does not necessarily mean long-term funds have re-entered.
How to judge the quality of the rebound?
I first look at the candlestick structure.
Currently, $SOXX still shows lower highs and lower lows during rebounds. The previously broken platform has gradually become resistance.
If the rebound next reaches above $500, or even approaches the $510–$520 range again but faces selling pressure and then breaks below the previous low, the downtrend remains intact.
Only if the rebound breaks the previous high, forms a higher low on the pullback, and can reclaim major moving averages will I start to consider the trend is repairing.
Volume is also important, but volume expansion alone cannot prove institutions are buying.
If huge volume spikes occur only on one or two trading days followed by rapid volume shrinkage and price decline, this is more characteristic of short-term position adjustments or trading rebounds.
If the rise can be sustained continuously, selling pressure weakens on pullbacks, more semiconductor component stocks simultaneously reclaim short-term moving averages, and $QQQ and $SPY also improve, then the credibility of sustained buying returning increases.
Regarding options, I observe whether downside protection demand weakens, Put holding structure changes, implied volatility declines, and whether new risk appetite improvements appear during price rises.
But these indicators must be judged in conjunction with spot trading volume and price structure; a mere decline in Put/Call ratio does not confirm short squeezes.
I have emphasized before that this round of adjustment looks more like existing AI longs exiting and insufficient incremental buying from the market perspective. There is still insufficient evidence to prove shorts have built large concentrated positions like in July.
So while I believe a considerable rebound is possible after continuous extreme declines, I do not preset a large-scale short squeeze will definitely occur.
Moreover, U.S. Treasury yields, oil prices, AI financing risks, and uncertainties around the November midterm elections still exist. Even if semiconductors rebound short-term, macro pressures will not disappear.
If prices cannot break previous highs during the rebound, and volume and market breadth do not improve sustainably, I still tend to reduce positions on rallies.
If prices stabilize above key moving averages, pullbacks do not make new lows, options market risk hedging demand eases, and large tech stocks regain relative strength, then I will consider increasing positions.
Short-term oversold conditions can drive rebounds, and short-covering can amplify gains, but only sustained buying can change the trend.
I still believe the bull market has not been definitively disproven, but this October adjustment has not confirmed its end either.
The next rebound is worth looking forward to, but until capital truly flows back, I will continue managing positions according to a volatile downtrend.
In this round of AI crash, I think the easiest mistake to make is treating all stocks that have dropped 20% as golden opportunities. The market is re-pricing AI assets, but different companies have completely different balance sheets, business models, and competitive moats. $TSM $ASML $NVDA $GOOG $CRWV
Let's first look at TSMC.
$TSM's latest quarterly revenue is $46.72 billion, with a gross margin of 66.4% and net profit growth of 68.7% year-over-year. Advanced process technology, yield rates, advanced packaging, and customer ecosystem form very high competitive barriers. Even if short-term AI capital expenditure growth slows, competitors will find it difficult to quickly replicate its manufacturing capabilities.
$ASML is similar. The technical threshold for EUV lithography equipment is extremely high. The company's latest quarterly revenue is €11.66 billion, with net profit close to €5 billion. This industry position is not maintained by continuous capital market financing but is built on long-term technological accumulation and real customer demand.
Now let's look at the other end.
$CRWV's bond yields have risen to about 15%, and the market is beginning to reassess its debt burden and future cash flows. The company has many long-term customer contracts but needs to invest huge upfront capital to build data centers and purchase equipment, then gradually recover the investment through future revenue.
If financing costs continue to rise, the return on investment for new projects will be compressed. Once financing conditions deteriorate, the company may even be forced to adjust its expansion plans.
The risks of these two types of companies are completely different.
For companies like TSMC and ASML, short-term stock price declines may mainly come from valuation compression and capital withdrawal. As long as long-term profitability does not significantly deteriorate, price drops may actually increase potential future investment returns.
But for companies highly dependent on external financing, stock price declines sometimes accompany worsening financing ability. Higher financing costs may affect future expansion and cash flow, creating a negative feedback loop.
Of course, a deep moat does not guarantee the stock price is cheap. Even globally leading companies may continue to adjust if their previous valuations have overextended future years' growth.
Therefore, when I screen AI stocks now, I focus on four things: whether the company truly has a competitive advantage, whether operating cash flow can support capital expenditures, whether the balance sheet can withstand high interest rates, and whether the current valuation still demands excessively high growth.
I still believe October is closer to a deep correction within a bull market, but I will not indiscriminately buy all companies that have crashed just because I am optimistic about AI in the long term.
In the short term, I expect a technical rebound in semiconductors; in the medium term, I remain cautious of further pullbacks. Therefore, those with ample cash need not rush to fully invest at once.
If the market continues to decline, I prefer to research $TSM, $ASML, $GOOG, $MSFT, $NVDA, $AVGO in batches—companies with deeper moats and mature business models.
For companies with high leverage, high debt, and earnings still highly dependent on future expectations, I will require a higher margin of safety.
October may be creating long-term investment opportunities, but not every fallen company is worth catching.
In the coming years, companies that can truly recover and continuously create shareholder value need to prove themselves through earnings, cash flow, and competitive advantages—not through the next round of market hype.
The yield on the US 10-year Treasury has fallen back from previous highs, but $SOXX has still declined for 9 consecutive trading days. This is a phenomenon worth paying close attention to this week. If the semiconductor adjustment was only caused by the rise in Treasury yields, then after the rate decline, tech stocks should at least receive some support. But the reality is not so.
As of October 16, the US 10-year Treasury yield remains around 5.1%, and Brent crude oil is close to $110. Although Treasury yields have fallen this week, high financing costs, energy prices, and uncertainty about AI investment returns still limit the valuation recovery of tech stocks.
It is important to understand one issue: a Treasury yield around 5% versus around 3% means a completely different pricing environment for high-growth companies' valuations.
For companies relying on earnings growth over many years, the higher the discount rate, the lower the present value of future cash flows. Even if the market continues to be optimistic about AI's long-term development, it may not be willing to pay the previously high valuation multiples.
A more direct impact comes from AI infrastructure financing.
Many data center projects require large upfront investments in GPUs, servers, network equipment, and must bear construction and electricity costs before revenue is realized.
If financing costs remain high, projects will need to generate more cash flow in the future to achieve the same investment returns as in low-interest-rate periods.
CoreWeave is a case worth watching. Recently, some of its bonds' secondary market yields have risen to about 15%, indicating that the credit market is beginning to demand higher risk compensation.
Of course, this does not mean all AI companies face the same issues. Large tech companies like Microsoft and Google have mature businesses and strong internal financing capabilities, so they should not be conflated with data center operators heavily reliant on external financing.
But the market worries that if financing costs for some AI infrastructure projects continue to rise, future new capital expenditures may be delayed, and this change could gradually affect upstream semiconductor orders.
Oil prices are another source of pressure.
Brent crude near $110 increases energy costs and inflation uncertainty, potentially limiting the Federal Reserve's room to further ease monetary policy.
For data centers, electricity costs are only part of project economics; for the whole market, oil prices more importantly affect inflation expectations and long-term interest rates.
However, I believe it is too simplistic to attribute all declines to macro factors.
This week, Treasury yields have fallen, but tech stocks continue to be under pressure, indicating the market is simultaneously reassessing previous AI stock earnings growth expectations, valuation premiums, and crowded positions.
The rise in September already reflected a lot of optimism in advance. When funds start exiting high-valuation trades, stock prices can continue to adjust even if the company's fundamentals have not significantly deteriorated.
This also explains why despite strong earnings from TSMC and ASML, semiconductors still cannot stabilize.
In the short term, I still tend to believe $SOXX has significant technical rebound potential. The continuous decline has pushed the market into a clearly weak state, and as long as selling pressure eases, a rapid price recovery may occur.
But I remain cautious about the sustainability of the rebound.
If the 10-year Treasury yield remains high, oil prices do not fall significantly, and AI credit spreads do not narrow, then after the rebound, the market may still face financing cost and valuation issues again.
The upcoming US midterm elections in November also add uncertainty to policy expectations.
Therefore, I will focus on observing three changes: whether Treasury yields continue to decline, whether the credit financing environment for AI-related companies improves, and whether tech giants can demonstrate through earnings reports that massive capital expenditures are forming sustainable revenue and cash flow.
If these conditions gradually improve, semiconductors will have a better chance to shift from a technical rebound to a sustained rise.
If it is only a short-term candlestick repair without changes in macro and financing conditions, I still prefer to use the rebound to reduce overly high positions.
I do not deny AI's long-term growth because of the October decline, nor do I believe the bull market has ended at this point.
But long-term industry logic holding true does not mean short-term valuations are already reasonable.
The market is still in a phase of risk repricing. A rebound is worth looking forward to, but until there is substantial improvement in the macro environment and capital structure, I will not easily change my view on the mid-term adjustment.
A 10% or even 20% pullback during a bull market is actually much more normal than many people imagine. The nine consecutive declines of $SOXX this round are indeed severe, but I don't think the AI bull market has ended just based on the sharp drop in October. $QQQ $SPY $NVDA $MU
Let's first look at a set of historical data.
JPMorgan has compiled the performance of the S&P 500 from 1980 to 2025. Over the past 46 years, the average maximum drawdown each year for the S&P 500 reached 14.2%, yet in 35 of those years, the index still ended with a full-year gain.
In 2023, the S&P 500's maximum drawdown during the year was about 10%, with a full-year gain of 24%.
In 2024, the maximum drawdown was about 8%, with a full-year gain of 23%.
In 2025, the maximum drawdown approached 19%, yet the full-year gain was still 16%.
These data show that even in years with significant final gains, the market can go through quite painful adjustments along the way.
Why do pullbacks keep occurring during a bull market?
Because stock prices often rise faster than corporate earnings growth. Especially in high-growth sectors like AI, the market tends to price in demand for several years ahead, with capital continuously chasing gains and valuations expanding.
But companies cannot exceed the market's ever-increasing expectations every quarter, and the macro environment cannot always remain favorable.
Once U.S. Treasury yields rise, financing costs increase, or institutions start reducing risk exposure, previously crowded trades can quickly reverse. Profit-taking, deleveraging, and options hedging can further amplify the declines.
This also explains why stocks can keep falling even when corporate earnings reports remain strong.
So why are these pullbacks, in a sense, a good thing?
Because they allow the market to digest overly high valuations, reduce leverage and crowded positions, and bring stock prices closer to the companies' actual earning power.
For long-term investors, if a company's competitive advantages have not deteriorated and future earnings expectations remain stable, a price drop means buying the same equity at a lower cost, potentially improving long-term expected returns.
For the overall market, after valuation adjustments, future gains can be supported by a more solid earnings foundation rather than relying solely on investors' willingness to pay ever-higher price-to-earnings ratios.
Of course, not all pullbacks are healthy. If corporate earnings continue to deteriorate and credit risks spread, the market may shift from a normal correction into a prolonged downturn.
Back to the present, I still tend to believe that October was a deep correction within the bull market, but I don't think the correction is over.
Risks remain from U.S. Treasury yields, oil prices, AI financing pressures, and the uncertainty of the November midterm elections. Even if semiconductors see a clear short-term rebound, they may pull back again under macro pressures.
So my strategy remains unchanged: control position sizes in the short term, use rebounds lacking sustained buying support to reduce risk; and in the long term, continue researching companies with true moats, profitability, and healthy balance sheets.
There is no need to completely dismiss investment opportunities for the next few years just because of one extreme pullback.
Many historical bull markets have experienced 10% or even 20% corrections. As long as corporate earnings can continue to grow and the macro environment eventually improves, the market still has a foundation for renewed gains.
Bull markets are never about going up every single day.
What truly matters is whether companies can continue to create value after valuation adjustments.
I remain optimistic about AI's long-term development, but what is needed now is patience, cash, and reasonable position sizing—not rushing to use leverage to recover losses from this round.
$SOXX has fallen for 9 consecutive trading days, but the risks to watch out for next may come from $QQQ and $SPY. Semiconductors have already experienced a significant pullback, while the overall adjustment in the S&P 500 remains limited. If the sell-off in tech stocks continues to spread, the market could enter a second phase of correction.
As of the close on October 16, $QQQ was at $709.80, down 5.52% for the week; $SPY was at $763.25, down 1.97% for the week. Meanwhile, the Dow Jones index actually rose 0.7% this week.
This divergence indicates that the current decline is still concentrated in the previously most crowded tech growth sector. Semiconductors, AI infrastructure, and some software companies have borne the main selling pressure, while sectors like energy and consumer staples have held up relatively well.
From a technical perspective, $QQQ has already fallen below its 50-day moving average of about $726.8, and there is still some room before reaching the 200-day moving average at about $674.9. $SPY has also dropped below its 50-day moving average of about $768.5, but it is still quite far from the 200-day moving average at about $725.
The short-term trends of both indices have weakened, but the long-term trends have not yet confirmed a reversal.
Market breadth is also deteriorating. As of Friday, only about 37% of Nasdaq 100 components are above their 50-day moving averages. In other words, the weakness in the index is no longer just a problem for a few semiconductor companies; more large tech stocks are also starting to lose trend support.
I have previously emphasized that this October correction looks more like a risk reduction by existing AI bulls after the short squeeze rally in September ended, with new funds unwilling to step in. There is currently no sufficient evidence to prove that bears have fully exerted their power, but the lack of buying itself is enough to keep prices falling.
There are two key scenarios to watch going forward.
If $SOXX stabilizes first, $QQQ climbs back above its 50-day moving average, and $SPY can hold its medium-term trend, then October could still be just a deep correction concentrated in the AI sector.
But if semiconductors experience a technical rebound while $QQQ continues to form lower highs, and $SPY subsequently breaks more key supports, then the risk could shift from localized valuation compression to a broader index correction.
This is why I am not in a hurry to conclude that the market has bottomed out.
I still believe the bull market is not over and that semiconductors have a clear rebound demand after continuous extreme declines. However, U.S. Treasury yields, oil prices, and the AI financing environment have not yet shown clear improvement, so the possibility of a pullback after a rebound still needs to be considered.
Especially since semiconductors have already released a lot of risk in advance, it cannot be ruled out that large tech companies may start to catch down, dragging the entire AI industry chain down again.
My strategy remains to control position size. If $SOXX shows a clear rebound but $QQQ and $SPY do not improve in sync, I would tend to use the rebound to reduce excessive risk exposure.
True trend repair requires the index to stop making new lows, form higher lows again, and have more components participate in the rally.
Until these conditions appear, I will continue to treat the market as oscillating downward.
What needs attention now is not only how much further semiconductors can fall, but also whether their weakness will further transmit to the entire U.S. stock market.