Quantarded Weekly Signals #006 — Week 5, 2026
This was a week where the tape looked fine until it did not.
The basket-level return built steadily into Thursday, then gave back a large chunk on Friday. That is not a subtle pattern, and it matters because short-horizon signals do not get the luxury of “waiting it out” when the market decides to reprice something quickly.
We should treat Week 5 as a reminder that the strategy’s edge (if it exists) is about net drift, not about being insulated from tail moves.
For context, this week Quantarded processed 49 House trade disclosures filed during the week, 337,825 Reddit comments analyzed and 49,857 stock ticker mentions detected and classified. As always, what matters is not volume alone, but how agreement evolves as volume grows.
Reddit picks
This week the ordering is unusually sharp. One name clearly dominates, and everything else is borderline.
$TSLA: SELL, 30% share
$TSLA sits well above the rest in implied bearish skew. The narrative driving this is tension between near-term fundamental pressures and extremely long-term optionality. Recent earnings revealed a -17% drop in profits year-over-year and a decline in deliveries even as revenue topped expectations, and leadership outlined a strategy pivot toward autonomy and robotics that carries both promise and execution risk. See the earnings recap with Musk’s robotics and autonomy emphasis here: Business Insider
Meanwhile, conventional EV deliveries are slowing and competition is intensifying; Tesla narrowly beat revenue estimates but saw automotive gross margin compress, suggesting near-term stress in the core business: Reuters
This juxtaposition — declining traditional results versus expensive, speculative future pivots — likely underpins a broad retail bias toward selling: a narrative that isn’t “fundamentally broken” but lacks clear conviction and has lots of moving parts. That complexity tends to dilute consensus, which is precisely what shows up here.
$MU: BUY, 20% share
$MU is the clearest BUY this week, and its news backdrop is coherent and reinforcing rather than speculative. The memory chip maker continues to benefit from structural supply tightness, driven by AI infrastructure demand outstripping near-term capacity. Analysts and company leadership have highlighted that the global memory chip supply is selling out well into 2026, with hyperscale data center demand a critical driver: Intellectia
Additionally, Micron’s broader capital investment plans — including multi-billion-dollar fab expansions in Singapore and the U.S. — reflect a long-term belief in persistent memory tightness: Barron's and Morningstar
$SNDK: BUY, 20% share
$SNDK’s inclusion sits on the oversized impact of a blowout earnings beat rather than a broad consensus. SanDisk’s latest quarterly results showed revenue and earnings well above expectations, driven by enterprise and AI-related storage demand that appears fully priced into recent valuation gains: The Motley Fool
The company also extended a long-term joint venture and is projecting very strong near-term growth, which has captured attention from both analysts and traders: Sandisk
But the sentiment score reflects narrow enthusiasm, not robust agreement: this is a name with plenty of attention, a clear price move, and some conviction — but not deep, sustained directional sentiment.
$NVDA: BUY, 15% share
$NVDA is one of the most prominent tickers in retail and institutional tech narratives, and that visibility is why it qualifies. Real-world context supports continued relevance: analysts and commentators alike call Nvidia a core AI infrastructure play — with some top brokerages positioning it as a leading AI pick for 2026: Investing
At the same time, the stock’s sentiment signal is shallow, suggesting ongoing debate about valuation, cyclical demand, and competitive dynamics. That tension matches the news picture: broad institutional optimism balanced against concerns about pricing and product cycle timing.
$MSTR: SELL, 15% share
$MSTR sits at the bottom as a weak bearish theme. Its story this week is tied most directly to its ongoing Bitcoin strategy and shifts in analyst coverage: Cantor Fitzgerald recently initiated coverage with an overweight rating and a higher price target, highlighting potential upside but also signaling high dispersion of views given the stock’s history and volatility: MarketBeat
Separately, markets are digesting Strategy’s large crypto holdings and its capital structure, which adds complexity to any directional thesis. That complexity shows up in sentiment as shallow negative imbalance — not a clear sell consensus.
Even when they do not qualify as picks, some names still dominate attention. $SLV and $UNH were heavily discussed this week.
House trades — one real cluster, everything else looks like noise
House disclosures this week had a familiar shape: plenty of filings in the background, but only a small subset reads like position building rather than incidental activity.
The trades that actually passed the Quantarded filters fall into two buckets: a single, concentrated “portfolio construction” pattern from one filer, and one standalone Senate SELL that is real in size but not reinforced by anyone else.
Nancy Pelosi: concentrated BUY exposure across large-cap tech plus one income tilt
The only genuinely structured pattern in the filtered set is the batch of Nancy Pelosi filings: multiple sizeable BUYs clustered in time, all pointing in the same direction.
This is not “one trade”. It is a set of aligned allocations, which is exactly the kind of behavior the House-trades model is designed to prioritize.
It is also a public narrative right now: multiple outlets have covered the same disclosure as a “big tech / AI leaning reposition”, precisely because the volume of moves is high for a single filing window and the names are canonical. See Pelosi just doubled down on AI again.
How to read this: not as “Pelosi predicts next week”, but as a signal of intentional positioning (multiple size-weighted moves, clustered, thematically coherent). That’s meaningfully different from one-off sales.
The other trade that passed filters is a Senate disclosure:
- Tina Smith disclosed a SELL of $BRK.B (≈ $100k, traded 2026-01-27, disclosed 2026-01-30)
On paper, this is a legitimate trade size and worth noting. But it remains a single-filer, single-name signal with no visible clustering around Berkshire from other filers in the same window, which limits how far we should lean into it.
Performance review – what actually happened
Week 4 turned out to be a difficult one, and notably the second negative close so far. Up until Thursday, performance was broadly constructive. Friday changed the picture entirely.
Broad markets were already fragile going into the end of the week, and that fragility mattered. Over the same five-day window, the $NDAQ ended the week meaningfully lower, while the $SPX also finished down, reflecting a risk-off turn rather than a clean trend. Late-week selling pressure dominated tape behavior.
This was not a slow bleed. It was a regime shift into Friday, which is exactly the environment where fixed-window weekly baskets are most exposed.
Last week’s results
The defining feature of the week was $SLV. Through Thursday, the position was behaving normally and contributing positively. On Friday, silver sold off violently, wiping out prior gains and overwhelming the rest of the basket. That single move explains most of the weekly drawdown.
Portfolio tracking
Using the running history you provided, the portfolio now stands at:
- End of 26W4 return: -1.99%
- YTD (2026) return: +9.51%
- Cumulative return since inception (weighted): +20.65%
A note on stop-losses
Weeks like this force a hard question.
A simple dynamic stop-loss, for example at -10%, would have materially reduced the damage from $SLV. At the same time, introducing stop-losses into a weekly sentiment system creates a new trade-off: forced exits during temporary dislocations that later reverse, increasing path dependency and potentially lowering long-term signal capture.
There is no free option here.
What this week makes clear is that not having a stop-loss is itself a choice with real consequences. A reasonable middle ground may be to treat a stop-loss not as a trading tool, but as a catastrophe brake: something that only activates on extreme, late-week moves when conviction across the basket is already thin.
This is an area that deserves deliberate evaluation, but at this points it seems a good recomendation.
Knowing the algorithm — how we consolidate House trades
House trade disclosures are fundamentally different from sentiment data. They are sparse, delayed, and constrained by reporting rules. Treating them like a fast or reactive signal is a mistake.
What we are trying to do here is simple to state, and hard to execute:
Identify disclosures that look intentional and non-random, rather than incidental.
Over the last few weeks, we iterated on the House-trades logic to better reflect how lawmakers actually behave in the data.
From isolated trades to deliberate behavior
Early versions treated each disclosed trade largely in isolation. In practice, that missed an important pattern: lawmakers often build exposure over time.
We adjusted the model to recognize when the same person trades the same ticker, in the same direction, across multiple filings. That does not turn those trades into a single “position”, but it does change how we interpret intent.
A single trade can be noise. Repeated behavior is harder to dismiss.
What we now reward (and what we do not)
The current logic gives modest additional weight to:
- repeated trades by the same person in the same direction
- sustained capital commitment over a short rolling window
- breadth across different lawmakers rather than repetition by one
At the same time, we remain deliberately conservative:
- individual trades are never hidden or merged
- late disclosures naturally decay in relevance
- no single feature can dominate the score
What the signal means
The resulting confidence score is not a timing tool and not a prediction. It answers one narrow question:
Given what we know today, how credible is this disclosure as evidence of intentional positioning?
House trade data invites overinterpretation. The algorithm is designed to do the opposite: reduce noise, reward consistency, and remain explicit about uncertainty.
If something surfaces here, it is because the behavior itself looks deliberate. If it does not, the correct conclusion is often that there is nothing meaningful yet.
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