Quantarded Weekly Signals #011 — Week 10, 2026

This week closed positive, but with the usual reminder that concentrated baskets can get most of their outcome from one or two names. The path mattered less than the structure: when one position carries a large share, you should expect the weekly print to be dominated by that single contributor, for better or worse.

For context, this week Quantarded processed 51 House trade disclosures filed during the week, 269,769 Reddit comments analyzed and 24,636 stock ticker mentions detected and classified. As always, what matters is not volume alone, but how agreement evolves as volume grows.

Geopolitical risk is still on the tape. The ongoing Middle East escalation has pushed energy markets into the spotlight, and if oil remains volatile it can spill into next week’s trading via risk-off positioning, inflation expectations, and sector rotation (energy up, duration-sensitive names down). Oil and gas prices surge as Iran war disrupts Middle East supply and shipping.

Reddit picks

As a reminder, a ticker is labeled BUY or SELL only when it clears a minimum imbalance threshold. High visibility alone is not sufficient; divided sentiment is explicitly penalized.

This week’s basket is clearly top-heavy, with the leading signal carrying roughly a quarter of the allocation and the rest forming a descending cluster behind it. In practice that means the portfolio outcome will be highly sensitive to the largest name in the basket.

$HIMS: BUY, 27% share

$HIMS leads the basket this week with the largest signal share and the strongest directional imbalance among the picks. The BUY side clearly dominates the SELL side, which is exactly the profile required for a high-conviction entry in the system.

Participation is smaller than the mega-cap names in the basket, but the agreement inside that participation is unusually strong. Signals like this often reflect concentrated enthusiasm: fewer participants, but a very clear directional consensus.

In the background, the conversation around $HIMS has been orbiting the GLP-1 opportunity set and the platform’s distribution and partner dynamics. Novo and Hims to sell obesity drugs together as feud ends, Bloomberg News reports, Hims expansion may not come in time for risky GLP-1 business, Strive Pharmacy says it will not resume compounded oral semaglutide sales.

$MSFT: BUY, 26% share

$MSFT enters as the second largest position and provides the broadest participation footprint in the basket. The imbalance is clearly positive, but less extreme than the leading signal, which tends to correspond with more stable sentiment dynamics.

Discussion volume appears across multiple threads rather than concentrated in a single spike. That breadth is important in this framework: sustained attention generally produces signals that decay more slowly than those generated by one-off posts.

The wider context for $MSFT right now is still the same familiar mix: data center scale, power constraints, and the OpenAI adjacency that continues to shape how people frame the name. Microsoft exploring using advanced power lines to make data centers more energy-efficient, Microsoft to keep buying enough renewable energy to match all its electricity needs, OpenAI is developing alternative to Microsoft's GitHub, The Information reports.

$ORCL: SELL, 21% share

$ORCL is the only SELL signal in the basket, and it carries a meaningful share. The imbalance clearly tilts negative, which is necessary for a SELL to pass the threshold in the system.

Confidence is moderate relative to the BUY signals, and participation appears spread across several discussions rather than dominated by a single viral post. That combination produces a signal that is directional but not extremely aggressive.

Context-wise, $ORCL discussion has been heavily anchored to AI infrastructure capex narratives and the cost side of scaling data center buildout, which is exactly the kind of theme that can polarize sentiment fast. Oracle plans thousands of job cuts as data center costs rise, Bloomberg News reports, Oracle and OpenAI drop Texas data center expansion plan, Bloomberg News reports, Oracle shares gain as $50 billion raise eases data-center spending concerns.

$MU: BUY, 14% share

$MU clears the BUY threshold with a positive imbalance profile but lower confidence compared with the top of the basket. Signals in this category often reflect more balanced debate within the comment set.

The participation footprint remains healthy, which supports the signal despite the lower conviction score. In practice this makes the position more of a secondary contributor than a structural driver of weekly performance.

The broader framing for $MU remains the same cyclical-meets-AI story: memory pricing, capex, and how durable the demand tailwind is. Exclusive: Micron to announce memory chip manufacturing investment in Singapore, sources say, Micron forecasts blowout earnings on booming AI market, raises 2026 capex plans.

$NFLX: BUY, 12% share

$NFLX rounds out the basket as the smallest allocation this week. The imbalance is clearly positive, although the confidence score sits below the leaders.

Discussion appears across both submissions and comment threads, which helps stabilize the signal despite the smaller share. In a week where the basket is top-heavy, positions like this mostly function as diversification rather than primary drivers.

For $NFLX, the background narrative has been a mix of product and deal chatter, with attention on strategy rather than a single discrete catalyst. Netflix acquires Ben Affleck's AI film-tech firm, Netflix declines to raise offer for Warner Bros.

Several tickers showed visibility without conviction this week. Names such as $NVDA and $TSLA appeared frequently in discussions, but sentiment remained too divided to pass the imbalance threshold required to enter the basket.

House trades

This week produced no qualifying signals from House trade disclosures, because no new disclosures were ingested in the weekly window. With zero new filings, there is nothing to cluster, no recurrence to evaluate, and therefore no meaningful signal to extract.

That said, the most recent batch of filings visible in the extended feed (the prior week’s activity) is a good illustration of why the filter exists. The activity was concentrated by filer, not by ticker: 9 total trades in that batch, with John Boozman (4 trades) and Sheldon Whitehouse (4 trades) accounting for the majority, and Shelley Moore Capito (1 trade) rounding it out.

Crucially, tickers did not repeat. Each of these appeared once, which is exactly the profile that gets filtered out as noise rather than signal: $VZ (SELL), $PG (SELL), $NTAP (BUY), $KR (BUY), $AMAT (SELL), $HD (SELL), $MA (SELL), $PEP (SELL), and $ITW (SELL).

There is nothing actionable on the House side this week, and even the most recent observed activity is dominated by one-offs rather than repeat patterns.

Performance review

Last week’s results

Gains were led by one outsized mover, while the rest of the basket mixed small advances with steady drawdowns into Friday.

Dispersion defined the week. One name posted a very large weekly move, and that single contributor dominated the aggregate result. This is the core mechanical property of a concentrated basket: even if multiple positions are directionally correct, the realized outcome is often decided by the largest deviation from the mean.

The other four names were a mix of modest positives and negatives. That matters because it shows the portfolio was not "uniformly right" or "uniformly wrong" at the ticker level. It was structurally dependent on the largest swing.

Portfolio tracking

• End of 26W9 return: +2.38%
• YTD (2026) return: +13.72%
• Cumulative return since inception: +25.29%

As of 2026-03-08, the portfolio stands at $12,528.65 starting from $10,000 on 2025-12-21.

Knowing the algorithm

One of the most common pieces of feedback I get is: "Why weekly signals and not daily picks?".

The short answer is that daily leaderboards are usually a great way to maximize activity, but a poor way to maximize signal quality. When you compress the horizon to 24 hours, you get more churn, more sensitivity to one-off bursts, and more false positives driven by transient attention.

A weekly window, by design, asks a different and more investable question: which names sustained directional agreement long enough to survive beyond the initial spike.

There is a clean statistical intuition behind this. In most real-world time series, the fastest-moving component contains a lot of noise relative to the slower-moving component, which is why forecasting and filtering methods typically down-weight older observations while still smoothing the path. The goal is not to be slow for its own sake, it is to avoid letting a single shock dominate the estimate. This is the same family of ideas behind exponential smoothing and other recency-weighted filters: responsiveness, but with inertia. If you want an academic-friendly reference for the core concept, see: Forecasting: Principles and Practice (Hyndman and Athanasopoulos).

There is also an execution and robustness angle that matters even if you ignore transaction costs. Short-horizon strategies pay a hidden price in turnover: more frequent re-ranking means more frequent position changes, and that amplifies the gap between what a model sees and what a portfolio can realistically hold. The portfolio rebalancing literature frames this as a trade-off between tracking error and trading frequency, where the optimal cadence often lands in the range of days to a week depending on correlation structure and costs. A useful reference on the rebalancing frequency problem is: Classification of the optimal rebalancing frequency.

Weekly signals are also easier to interpret and compare.

A daily pick list is often a reaction function, highly sensitive to what happened to be salient that morning. A weekly list behaves more like a standardized experiment: the window is long enough for the market and the crowd to express disagreement, and long enough for persistence to separate from noise. This is conceptually aligned with how risk estimates are often made more stable by weighting recent observations while avoiding excessive sensitivity to the last datapoint. A canonical reference for the intuition of exponentially weighted estimators in risk practice is: RiskMetrics Technical Document.

None of this says daily signals are useless. They can be informative as a real-time thermometer. But the newsletter is trying to be a weekly process you can reason about, backtest, and read as a coherent narrative. Weekly cadence is a deliberate bias toward persistence, comparability, and lower churn, which is exactly what most quant readers end up wanting once they have seen enough daily leaderboards whipsaw.

Disclaimer

This newsletter is not financial advice.

All content is provided for informational and educational purposes only. Markets involve risk, including loss of principal. Past performance does not guarantee future results. Always do your own research.

Links

Website: https://quantarded.com

Subscribe: https://quantarded.kit.com/products/monthly-subscription?step=checkout