[Weekly Recap] CTA Net Position Hits Extreme Low, Nasdaq Valuation at 4-Year Low: The Golden Age of Semiconductors Is Over

$SPDR S&P 500(SPY.US) $VG S&P 500(VOO.US) $Direxion Semicon Bull 3X(SOXL.US) $AMD(AMD.US) $NVIDIA(NVDA.US)

Finance Talk "Continuous Rotating Commentary" Nov 13, 2026:
Sam's Live Stream Return: Institutions have sold to the extreme; Nasdaq 100 valuation has reached its cheapest level in four years; the broader market may not be as panicked as imagined; but the good days for semiconductors, hardware, and upstream AI are over, making it inadvisable to maintain heavy long-term positions.
Following the midterm elections and the NVIDIA controversy, today's live stream discussed hedge fund and systematic fund positions, post-election seasonal statistics, the semiconductor cycle, and a Q&A session. The overall conclusion is clear: institutions have sold to the extreme, and Nasdaq valuations have reached their cheapest range in years. The broader market may not be as panicked as people imagine; however, the good days for semiconductors, hardware, and upstream AI have passed. It is no longer advisable to hold heavy long-term positions. Careful selection is needed, and position sizing and risk management must be strictly controlled.

Hedge Funds: Tech and Industrial Sectors Both Reduced
The first part covers institutional positions. Prime broker data shows that during the week of Oct 30 to Nov 5, net selling of global tech stocks accounted for approximately 3% of the sector's total market cap. The red line marks the largest net sell-off in the past decade, described as extremely crazy. The continuous decline in tech stocks that everyone feels is driven by these funds selling. In the same week, the US industrial sector was also significantly reduced. This wasn't due to issues with a single company, but rather a broad reduction across the entire sector.
Within the industrial sector, familiar names include power stocks and power equipment stocks such as VST, CEG, and Caterpillar. After the midterms, these power-related stocks were washed out: the political variable of Democrats taking power was seen as impacting investments in data centers, which in turn affects power stocks. These hedge funds had already reduced positions during the week of Oct 30 to Nov 5 before the election.

CTAs: Selling Pressure Exhausted, Buying Power Awaiting
After active funds sold off, systematic funds followed. CTAs (Commodity Trading Advisors) rely on algorithms to follow trends, automatically buying and selling, chasing momentum without regard for value or fundamentals. Their current US equity net position Z-score has dropped to -3.3, meaning it is 3.3 standard deviations below the mean. Over the past year, only about 1.9% of the time was more pessimistic than now. The most extreme point this year was the April tariff shock at -4.3; we are now approaching that level. In other words, systematic funds have sold to the extreme.
Goldman Sachs' estimates for CTAs on the S&P 500 depict the next path. The dark blue portion has already occurred: CTAs have sold approximately $55bn in the S&P 500. If the market ranges from now until early December without further major drops, or even rises by one standard deviation, the red line indicates CTAs would be forced to buy back nearly $45bn. If it rises another two standard deviations, buying pressure will be even greater. Conversely, the market would need to drop more than two standard deviations for them to continue heavy selling. Unless another sharp drop like last Thursday occurs, whether the market ranges or surges, significant buying will be required.
This is the so-called asymmetric positioning: downward selling pressure has been largely exhausted, while upward buying power is waiting in the wings. For the outlook, this is a relatively positive signal.

QQQ Short Interest and Market Sentiment
For over two months, QQQ has underperformed SPY because tech stocks are weak. The number of short shares behind this has grown significantly: rising from a low of ~43m shares mid-year to ~82m shares now, nearly doubling and returning to highs seen in 2022-2023. This represents many people shorting QQQ, but it does not mean they are bearish on the market as a whole. As mentioned in the previous episode, these institutions still hold significant long positions and tech stocks. Shorting QQQ is more like buying insurance: holding stocks and using short QQQ positions to reduce overall risk exposure. The rapid rise in short interest indicates they want to reduce risk.
Sentiment has also reached extremes. The Bull Bear Ratio, calculated on a 125-day rolling basis, has dropped to a Z-score of -2, corresponding to the green rectangle on the chart. Historically, reaching this level has marked peaks of panic: the 2022 rate-hike bear market, the April 'Liberation Day' tariff shock, subsequent geopolitical tensions, and this recent tech selloff. Sentiment indicators are contrarian; this doesn't mean an immediate rebound, but rather that pessimism has reached an extreme level, and there is no need to use even more pessimistic assumptions to scare oneself.
On the call, it was noted that many people nearby still believe prices will fall further and haven't bottomed out. Conversely, those who are very optimistic, claiming 'this is definitely the bottom, must bottom-fish,' are few. Especially after some stocks had issues and NVIDIA faced problems, more people became shaken. Combined with CTA and QQQ positioning, sentiment is considered to be almost fully priced in.

Nasdaq 100 Valuation: Cheapest in Four Years
The valuation chart shows the Nasdaq 100 (NDX) consensus forecast P/E is currently 19x, down 25% from its January high of 25.3x, and cheaper than the five-year average of 24.2x. The five-year percentile is at 1%, meaning only 1% of the time in the past five years was cheaper than now, and 99% of the time was more expensive. This is the cheapest level since the bear market bottom in Oct-Nov 2022.
This does not imply an immediate rebound. It can get cheaper after getting cheap, or become more extreme. The key point is: this drawdown has reflected many pessimistic expectations, and valuations have released significant risk. Looking short now feels late; shorts are usually established at highs, not continued at lows.
Midterm Election Seasonality: Data is Optimistic, But Don't Take It All at Face Value

The second part discusses elections. Historically, post-election day trends often lean upward. As mentioned in the last episode, regardless of the two-party outcome, markets can usually 'have a good run' after midterms, though how long is uncertain. This time deviates slightly from expectations: past data shouldn't be fully trusted. If one bought heavily before the election, the result now might not be favorable. Indices haven't fallen much, but no upward trend is visible yet. These statistics look at roughly a six-month horizon, offering the market an opportunity to rise, not implying an immediate surge right after the election.
Events are changing. Democrats took power this time, Republicans lost decisively, and Texas results surprised many. The market hasn't fully prepared for such a stark contrast. Institutions hedged and psychologically prepared earlier, but aren't completely bearish. Adjustments in positioning and sectors will be necessary going forward.

Average Returns of Indices and Sectors Post-Midterm Elections (1990-2022) This chart shows the tech sector leading: Nasdaq 100 averages a 12% gain over three months post-election. Not all stocks rise together post-election; growth and tech stocks lead. Taking this literally might suggest going All-in on tech now; however, the live stream explicitly stated that such statistics could be proven wrong. Do not blindly trust historical data from institutions like Morgan Stanley. US tech stocks can still be bought, but with careful selection. For parts related to semiconductors, hardware, and AI infrastructure, reducing exposure seems preferable. The tech sector will remain highly volatile in the future.

Another chart shows: historically, when Republican presidents lost Congress to Democrats, this happened seven times between 1954 and 2006. On average, returns were positive for one, three, six months, and one year post-election. The data looks optimistic, dominated by tech leaders; but the semiconductor cycle ahead is what needs more caution today.

Semiconductor Cycle: Good Days Are Past
It was already discussed in the group chat on Aug 3: storage, semiconductors, and hardware will likely see more volatility, making it hard to say they will hit new highs. The most comfortable 'lying flat and winning' era is over. Heavy long-term holdings or buying in these sectors should be avoided now. Software and cybersecurity are relatively favored; NET (Cloudflare) was suggested as a consideration when appearing on the show for the first time. By mid-November, the view remains unchanged; with Democrats in power, extra caution is needed for the upstream AI segment, whose shine is considered a thing of the past.

This isn't telling everyone to liquidate everything. Keeping one or two positions, or a small portion to bet on further upside, is fine. If you don't want to play anymore, clearing out completely is also okay. These stocks have risen many multiples. Taking storage as an example, entering the market now with MU or Sandisk means facing a downtrend. Buying after a drop from highs is like catching a falling knife: you might miss the rally but catch the decline. Among storage plays, only US company MU is suitable for swing trading: buy low, e.g., around $600 low and $800+ high, potentially earning $100-$200, but know when to exit. Holding at $600+ as a long-term investment is dangerous and unlike the previous era.

The strongest rally started in Feb this year, peaking around June, multiplying many times in four months. Actions in the Korean market in June were seen as a key turning point; initially just a warning to watch Korean stocks, by August it was clear semiconductors should no longer be touched. This is a medium-to-long term view, requiring caution into next year and beyond.

Fund Flows, Market Breadth, and Dot-com Bubble Comparison
On the Semiconductor ETF fund flow chart, the black line is the S&P Semiconductor Index, serving as a proxy for SOX: a double top broke the neckline, and the technical trend is very poor. The index rose from a low of 1,347 in late Aug 2022 to a high of 12,735 in Aug 2026, nearly nine times higher in four years—a massive gain. Even more striking is the fund flow below: the blue and pink bars represent cumulative inflows over the past 13 weeks. Before 2026, 13-week inflows were mostly in the billions; starting Feb this year, they exploded, peaking at nearly $60bn over 13 weeks. Blue is US ETFs, pink is Korean ETFs, indicating global capital flooding in, not just US retail investors—this is classic FOMO.

The green and pink background bars represent market breadth: green if many stocks rise together; pink/red if the index rises but only a few large caps support it. Most of 2022-2025 was green. In 2026 near the top, red increased, meaning indices hit new highs but fewer stocks kept up—a typical topping or late-cycle signal. Fund flows have recently dropped back to around $10bn; the influx has stopped, and no one is FOMO-ing anymore. From a fund flow perspective, tops occur when inflows are highest, then FOMO stops, and indices drop nearly 30%. There have been some rebounds and inflows recently, but far less than the peak.

Another chart compares AI (2022-2026) with the Internet (1995-2000); the two lines are too similar. The top line is the index; the bottom is the cyclically adjusted P/E (CAPE). The orange line is 1995-2002; the black line is current conditions. Both rose from ~21.3x in early 2025 to 28.2x, comparable to the 29.2x peak then. Valuations have reached the dot-com bubble top region, though not broken through it yet. The next move is down.

Worryingly, many are saying semiconductor valuations are cheap and want to buy the dip. But if the cycle has topped, the valuation compression might be justified. The market has already priced in events one to two years out; stock prices reflect these expectations. Why insist others are wrong? Why buy just because valuations are low?

Caution with Long-Term Underperformers
NVIDIA was used as an example in the live stream: why is it always called undervalued? Investigation news has circulated for over a year, and the stock has been cheap for over a year. Since breaking above $200, it has ranged, while AMD rallied. This isn't a confirmed conclusion but an explanation of price action: don't just buy based on your own analysis; watch price action and how institutions position themselves. Long-term underperformers often have problems, e.g., Lululemon, Nike. Buying solely because they are 'undervalued' can be troublesome; there may be hidden issues. Retail investors might not know what's wrong, but insiders do, hence front-running.
Applying this logic to semiconductors: why are they so low? Nasdaq's low valuation is different; it's a basket of many stocks. The semiconductor index covers fewer stocks, carrying higher topping risk than QQQ. At the individual stock level, long-term underperformers like NVIDIA require extra caution. Don't assume you're right. Retail investors are tiny; even after a year of rumors, understanding may be limited, and even insiders don't fully grasp everything. Don't think you're smarter than institutions. Institutions often have information retail lacks. AI can help analyze fundamentals and earnings, but cannot provide insider info. More time should be spent on visible real data—the market's collective force.

Buyback Window Reopens, Second Leg May Be Opportunity
The final chart shows the distribution of S&P 500 companies exiting their buyback blackout periods, shown previously. At this level, buybacks have resumed as most earnings are reported, providing market buying power. The broader market may not be as panicked as feared; if a 'second leg' down occurs, it could be viewed as an opportunity. NVIDIA reports earnings next week; volatility is expected, so position control is crucial. As mentioned last time, volatility will be high; positions must be managed well, and risk controlled. Have a mental defense level before buying.
Using NVIDIA as an example: if bought at $165 for trading/profit, a stop-loss is essential below, e.g., $150. The main resistance zone above is $180-$183. If holding for the next generation, that's a different thesis.

Live Q&A: AMD, QQQ, Bitcoin, Big Tech, and SOXL
An audience member asked if AMD, having just risen to its 50-day moving average, is a buy. The logic is funds flowing out of NVIDIA into AMD, creating a seesaw effect. During the live stream, NVIDIA was rising while AMD fell, making chasing the rally unadvisable. AMD rose ~$200 from $400; chasing now feels late. Wait for a pullback to ~$530; if it holds, consider buying. But remember the risk of the semiconductor cycle topping. Swing trading and long-term holding are different. For the semiconductor index and individual stocks, long-term participation is no longer desired; once a top is identified, long-term investing in this sector stops. Large swings can still be traded, e.g., AMD $400 to $600, or MU, but clarity is needed. 'Buy and Hold' is no longer suitable. Buffett has exited the market; this method is considered obsolete. Next year, escaping the top will be a priority; concerned about 2027 cycle-wise, will monitor closely.
If trusting institutional choices, institutions have significantly exited semiconductors, with no sign of large-scale return yet. Semiconductors, hardware, and AI infrastructure are being traded by short-covering and hot money—buy fast, sell fast. TSMC is relatively better with stronger defense, but why buy something at a cycle top? Unless you disagree with the topping view, it's a cognitive issue. Not participating won't cause missed opportunities; other sectors exist: software, cybersecurity, AI applications, pharma, and healthcare. Healthcare carries higher risk; Eli Lilly (LLY) is too expensive, skip for now. Hardware has risen nicely; healthcare is correcting. Capital rotates between these two sectors without a clear anchor for broad semiconductor bullishness. Institutions are positioning in healthcare.
QQQ contains more than just semiconductors. If wanting less volatility and regular investing, SPMO or VOO (similar to SPY) are better, more balanced ETFs. SPMO is the S&P Momentum Select ETF, rebalancing semi-annually to select high-momentum stocks. Long-term observation shows it performs slightly better than QQQ and SPY, but with larger drawdowns due to the momentum factor. Without leverage, buying VOO or SPY directly is fine; VOO has lower fees. Technically for QQQ, defend 680 short-term. A break below 680 targets ~655. Resistance is at 706-707; a three-day close above confirms the 'right side' technically. Currently in the middle, waiting for direction.
Bitcoin needs to reclaim $76,000 to have a chance; otherwise, it will continue falling. Key support is ~$64,000-$65,000, possibly dipping to $60,000. Support near $70,000 is weak. For right-side/momentum trading, re-taking $76,000 is necessary as most chips are there. Crypto is seen as part of the Trump trade; with his power waning, consider the 'anti-Trump trade.' Midterm results disrupted plans, plus Chief of Staff changes, personnel reshuffles, and statements from Iran. The US will be volatile for 1-2 years; cooperation with Democrats is unlikely. Making America Great Again again is deemed difficult. Handling Iran poorly sparked public anger; this failure is self-inflicted, underestimating the opponent's strength, and a poor team. Either don't act, or act thoroughly; the current state is the worst.
An audience member asked about Apple, whether software stocks include Microsoft, and if SOXL is buyable. Big Tech overall is still okay due to record cash flows. But watch for next year: Big Tech CapEx may decrease next year, making the AI hype more dangerous, especially upstream chips, infrastructure, and power. Markets often criticize Google for negative free cash flow; they may decide collectively not to spend so much, or demand is sufficient. They may not announce CapEx cuts individually but coordinate first, then announce together. Big Tech's stability lies in their ability to increase or decrease CapEx simultaneously. Increasing CapEx is already priced in and negative for stock prices; decreasing it has little impact on them but is positive as shareholder returns increase, possibly via buybacks. Their core businesses include others, e.g., Google ads, Meta ads, not just AI, which is auxiliary. Investment is nearing saturation; cuts may come next year. Prepare for this; don't assume full-speed ahead like this year. A major engine is tied to NVIDIA; if questions arise there and NVIDIA slows, the whole sector is affected.
Key sectors to watch remain software and AI applications, not upstream construction; upstream is considered topped, including 'new cloud' types, suitable only for swing trades. Swings are manageable, e.g., at 120, 150, but know when to exit. Don't buy to hold for kids forever, or it could drop from 120 to 20, or vanish. Unless you disagree with the top view, take responsibility for your cognition. Views may not be correct, but a top is believed, especially in June. Will US indices hit new highs? Definitely. This means QQQ and SPY rebalancing will add new good companies and remove fading ones. This rapid iteration mechanism allows for confident long-term holding. VOO is for confident long-term holding; Buffett said VOO yields ~10-12% annually. Individual stocks face risks like NVIDIA.
The view on semiconductors applies to the whole sector, including equipment, TSMC, NVIDIA, AMD. Many still favor TSMC for its defense, but the monthly chart has made two tops and declined. Can buy, but not heavily. Don't go all-in on TSMC, NVIDIA, AMD. Change the viewpoint. Check again after Democrats take office in Jan next year. If skeptical, wait until Jan/Feb to see semiconductor performance and trust institutions and price action. Why has NVIDIA ranged around $200+ for so long, with recent news? Mentioned last time: hedge funds started reducing in Sept. They are late movers; why not cut in June? They surely knew something. If seeing large institutional inflows into semiconductors, will inform everyone to reconsider; currently, no shadow of institutional return. Technically, it's a rebound, strong recently; short-term traders can follow. E.g., regarding SOXL, watch $50: if it holds, rebound chances exist, but beware of decay in 3x leverage; if $50 breaks, be cautious. $50 is the short-term watershed, not long-term; don't buy for long. Distinguish clearly between short and long-term attitudes.