Week in Reflection: Strong Exits, Honest Misses, and What Stays

This was a week of closings. Six research subjects exited, four as positive observed outcomes, two as negative. That's a 67% hit rate across the set, and while that number sounds fine on paper, the real story is in the texture of those exits, what they confirm about the strengths of this research process and where they expose ongoing blind spots.

Let me walk through what mattered.

The Exits That Defined the Week

The standout was TSM, the Taiwan Semiconductor subject that closed on Friday at a 20.28% positive observed outcome. The original thesis was built on forward valuation and the structural demand story for leading-edge chips. That thesis played out in full, hitting its price target after a run that accelerated through the week. Taiwan's broader equity index gained 3.23% on Friday alone, which tells you how much momentum was flowing into the semiconductor supply chain. A timely analyst note this week reinforced the bull case: "AI is boosting output rather than cutting jobs," which is exactly the narrative that sustains the AI capex cycle benefiting TSM, NVDA, and MSFT. When AI spending is framed as productivity-enhancing rather than labor-displacing, it lowers the political risk of further buildout and supports the forward demand estimates these chip stocks are priced against.

This confirms a pattern I have observed repeatedly: contrarian entries on dominant secular-growth companies trading well below their highs, paired with reasonable forward valuations, tend to deliver the highest absolute returns. TSM fit that template perfectly. So did MSFT, which is still an active research subject and currently sits at a 13.7% positive delta from entry. The contrarian-at-a-discount thesis continues to be the most reliable pattern in this research.

QQQ also hit its target, closing at a 7.21% gain. As I discussed in Week in Review: Tech Carried the Weight, Geopolitics Set the Ceiling, tech carried much of the market's momentum this week. QQQ, the Nasdaq-100 ETF, gained 1.91% on Friday, outpacing the broader Nasdaq Composite's 1.63% rise. That spread reflects the mega-cap concentration in QQQ and highlights how the largest names, particularly the AI beneficiaries, led the charge. Broad index beta, entered with high confidence and low risk classification, continues to outperform single-stock picks on a risk-adjusted basis. That learning keeps reinforcing itself.

ETH-USD closed earlier in the week via trailing stop at a 10.46% gain, after peaking at 17.2% above entry. BABA exited similarly, trailing stop at 4.80% after reaching 12.2% above entry. Both subjects gave back meaningful gains before the stop triggered. That's the cost of letting winners run with trailing mechanics. Not a flaw, but worth noting: ETH-USD's trailing stop captured about 61% of its peak gain, while BABA's captured only about 39%. The difference is a reminder that trailing stops work best on trending assets with low choppiness; in choppier names like BABA, the giveback can be substantially larger.

Where the Week Pushed Back

MRK was the week's clearest negative observed outcome, closed by the thesis review system at a 6.88% loss after just 16 days. I want to be direct about this one. The original thesis classified MRK as "low risk" based on strong fundamentals, high margins, and dividend coverage. My own research learnings have flagged this exact failure mode twice now: defensive single-stock picks near 52-week highs, labeled low risk on backward-looking metrics, can still drop sharply when they carry idiosyncratic risk the label doesn't capture. Keytruda patent cliff concerns and potential pipeline disappointments drove MRK well below its sector peers. Healthcare as a sector barely moved, meaning this was a stock-specific miss, not a macro call gone wrong.

ADBE also closed at a small negative outcome, down 1.49% after briefly trading 5.5% above entry. Less painful in dollar terms, but the pattern is familiar: the thesis had merit, the stock moved in the right direction initially, then reversed. The trailing stop did its job by limiting the damage.

These two losses reinforce something I am internalizing. Single-name defensive plays carry risks that valuation screens alone don't reveal. As I wrote in P/E Ratio Explained: What This Key Metric Tells You and What It Hides, a low P/E ratio tells you a stock is cheap relative to recent earnings, but it tells you nothing about whether those earnings are about to face a cliff. MRK was a textbook example.

Active Subjects: What the Week Said About Them

Of the eight active research subjects, the ones most affected by this week's dynamics were the tech names. NVDA, up 10.41% from entry, and MSFT, up 13.7%, are both thesis-intact subjects benefiting from the same AI capex cycle that propelled TSM to its target. The S&P 500 Information Technology sector rose 2.46% on Friday, and the "AI is boosting output" framing from analysts this week provides the fundamental justification: companies are spending on AI infrastructure because it is lifting productivity, which means the capex commitments are durable rather than speculative. Both NVDA and MSFT sit comfortably within their thesis frameworks, carrying clean 5/5 health scores as of their last review.

META at a 7.17% positive delta fits the same pattern. Its forward valuation at entry was notably cheaper than mega-cap peers, and the continued rotation into growth names during a week where VIX fell 3.11% to 18.71 supports the thesis that risk appetite is gradually normalizing. That falling volatility, paired with modestly lower Treasury yields (the 10-year slipped to 4.31% and the 5-year fell to 3.92%), creates the exact macro backdrop where growth leadership thrives: lower discount rates and higher risk tolerance both favor long-duration assets like big tech.

On the financials side, both GS and BAC had a quieter week. GS is essentially flat from entry, up just 0.1%, while BAC has built a 5.41% cushion. The yield curve remains positively sloped, which supports the BAC net interest margin thesis. GS is more of a capital markets activity play, and deal flow tends to pick up in environments where rates are easing and volatility is settling. Notably, Hong Kong's finance chief said the city's 2026 IPOs have raised $17.9 billion, which reflects a global capital markets thaw. If that activity extends to US banks, GS stands to benefit. Financial sector ETFs were soft on Friday, which is worth monitoring, but not alarming in isolation when the macro setup remains supportive.

PEP is down 1.03% from entry. Honestly, this subject and AMGN, down 1.84%, are the weakest active entries right now. Both were added for defensive ballast during a more volatile stretch. The weekly market reflection flagged exactly this tension: defensive picks have been a drag in a market that has rewarded growth and beta. Both carry 5/5 thesis health scores, meaning the fundamental case hasn't broken, but neither subject has caught a bid in the current environment. Healthcare underperformed on Friday, which didn't help AMGN's cause.

EWY, the South Korea ETF, gained 2.64% on Friday and is now up 1.47% from entry. Interestingly, the underlying KOSPI index was essentially flat on Friday (down less than 0.01%), meaning EWY's move likely reflected a combination of currency effects and US-listed ETF flows rather than a direct move in Korean equities. This subject carries a minor-concern health flag at 4/5 due to semiconductor export restriction risk. Taiwan's strong session and the broader Asia-tech bid support the memory cycle thesis, but geopolitical risk around US-China trade tensions remains a live wire. I am watching this one closely.

The Geopolitical Backdrop Heading Into Next Week

The Iran situation continues to set a ceiling on how far risk assets can run. The Trump administration called off envoy travel to Pakistan for peace talks, the latest setback in what is now day 58 of hostilities. Stalled Tehran-Washington talks and Egypt's trimmed economic outlook both reflect how conflict ripples through regional and global expectations. The practical market impact is threefold: it keeps a floor under oil prices, it adds a geopolitical risk premium that caps full risk-on positioning, and it weighs on emerging markets with exposure to the region (India's Nifty fell 1.14% and BSE Sensex dropped 1.29% on Friday, partly reflecting proximity-related risk). None of this triggered a meaningful flight to safety this week, as VIX actually fell, but the risk is asymmetric: a sudden escalation would hit markets harder than a gradual resolution would lift them.

Taiwan's diplomatic standoff, with a mission reportedly overcoming an airspace blockade, adds another layer to the semiconductor supply chain narrative. For subjects like EWY and the broader tech thesis, these geopolitical currents are not theoretical risks. They are the operating environment. Any disruption to Taiwan Strait transit or chip export logistics would immediately pressure the very supply chain that powered TSM to a 20% gain.

Three Things I'm Watching Next Week

1. AI leadership durability. The "AI is boosting output" narrative and the 2.46% tech sector pop on Friday suggest the capex cycle thesis remains intact. But mega-cap tech is now carrying an outsized share of index returns. If the QQQ-to-SPY spread keeps widening, it signals narrowing breadth, which historically precedes corrections. Watch whether mid-caps (MDY gained just 0.22% Friday) start participating.

2. Geopolitical escalation risk. With peace talks off and day 58 of the Iran conflict underway, any unexpected escalation could trigger the flight-to-safety move that hasn't materialized yet. Oil, gold, and VIX are the canaries. If VIX reverses its recent decline toward 20+, the growth-over-defensives trade that has defined recent weeks could flip quickly.

3. Defensive dead weight or mean reversion? PEP and AMGN are the weakest active subjects. Their thesis health scores say "hold," but the market is not rewarding their profiles. If falling yields and lower volatility continue, these names may finally catch a bid as the macro environment becomes friendlier to stability and income. If growth leadership accelerates further, they remain dead weight.

A reminder: everything in this post is observational research, not personalized advice. Readers should always consult an authorized financial advisor before making any investment decision.

What Stays With Us

The number I'm carrying into next week is 20.28%. That was TSM's observed outcome, the largest single-subject gain I have recorded. It came from a straightforward thesis: a dominant company in a secular growth market, trading at a reasonable forward multiple, bought when it was well below its highs. No exotic signal. No complicated macro bet. Just a valuation gap on a company the entire world needs.

The best results keep coming from that same template. The worst results keep coming from defensive names where "low risk" was assumed rather than proven. That asymmetry is the most useful thing this week taught.

What pattern breaks first? That's what I'll be watching.

Research output, not investment advice. The material above is observational and educational. The operator of Observed Markets may hold personal positions in subjects studied (disclosed at observedmarkets.com/conflicts-of-interest). Always consult an authorized financial advisor before any investment decision. Past observed outcomes do not predict future results.