Extreme Movers: Claude Sonnet 5.5's picks for 2026-10-07
Claude Sonnet 5.5 picked CEG, MRNA, STX, MRVL, OPCH for a 1-day horizon. Average directional return: +1.67%. Directions called correctly: 80%.
| Experiment | Extreme Movers |
|---|---|
| Model | Claude Sonnet 5.5 |
| Prompt version | extreme-movers-v11 |
| Asked | |
| Horizon | 1-day |
| Measured from | 2026-10-07 |
| Measured to | 2026-10-08 |
| Benchmark | SPY |
| Status | Settled |
| Average directional return | +1.67% |
| Directions called correctly | 80% |
| Benchmark return | +0.21% |
The picks
| Rank | Symbol | Direction | Price change | Return in the called direction | Confidence | vs benchmark | Called right |
|---|---|---|---|---|---|---|---|
| 1 | CEG | up | +3.93% | +3.93% | 50% | +3.72% | Yes |
| 2 | MRNA | up | +3.93% | +3.93% | 45% | +3.72% | Yes |
| 3 | STX | up | -0.24% | -0.24% | 42% | -0.45% | No |
| 4 | MRVL | down | -0.56% | +0.56% | 38% | +0.35% | Yes |
| 5 | OPCH | up | +0.18% | +0.18% | 35% | -0.03% | Yes |
The model's market view
The Oct 6 session closed at record highs on AI enthusiasm and falling oil. Oct 7 trading opened lower: the S&P 500 was down about 0.4% and the Nasdaq about 0.6% as the 10-year Treasury yield reached about 5.35%, the highest since 2002, and FOMC minutes were due at 2 p.m. ET. The picks lean on catalyst-driven names. Two are contrarian long reversals on moves that looked like profit-taking or sector contagion, one is a continuation on a substantive power-demand catalyst, one is a fade of a stretched semiconductor rally, and one is a deal-anchored stock. Limitation: I could not reach live intraday quotes, so no reference_price, market_cap_usd or avg_dollar_volume_20d_usd is reported for any pick. The previous_change_percent values are the latest session figures I could find (morning or midday, not confirmed closes) from a single live-blog source. All picks are large, liquid US-listed common stocks, but I did not verify the three screening floors with numbers.
Why each pick
CEG
Source session: Constellation Energy rose about 12.3% in the morning and about 14.8% at midday on Oct 6 (official close not confirmed; I used the midday figure) after announcing a 20-year, 890 MW nuclear power purchase agreement with Google. Catalyst: substantive, because it is a long-term contract with a committed $4.3B build-out and it lifted other power producers such as Vistra and Talen. Hypothesis: continuation up, because contract-backed AI power demand tends to attract follow-through buying after large re-rating days. Risks: sharp one-day gains in power names often fade, and rising yields (10-year about 5.35%) and a weaker market on Oct 7 hurt long-duration utilities. Not confirmed: the Oct 6 close, a current quote, the market cap (the only figure I saw was stale, about $98B in July), and 20-day dollar volume, so those fields are omitted. The company is a large Nasdaq-listed utility and clearly well above the floors.
MRNA
Source session: Moderna fell about 6.1% in the morning of Oct 6 (close not confirmed), described as profit-taking after a fresh 52-week high. The stock had rallied more than 7% on news that it will rejoin the Nasdaq-100 before the open on Friday, Oct 9, 2026. Catalyst: speculative and technical; the drop looks like sell-the-news rather than a fundamental change. Hypothesis: reversal up, because index-inclusion buying demand typically peaks into the effective date and the dip came without negative company news. Risks: healthcare was weak broadly on Oct 6, the rising-yield tape is hostile to high-beta biotech, and inclusion flows may already be priced in. Not confirmed: reference price, market cap and 20-day dollar volume (omitted). It is a widely traded large-cap biotech, so the floors are very likely met but not verified with numbers.
STX
Source session: Seagate fell about 8.0% in the morning of Oct 6 (close not confirmed) on continued worry over Toshiba's plan to double hard-disk capacity, after a double-digit drop on Oct 2. Catalyst: largely speculative, since the extra capacity arrives only by fiscal 2027 and Citi, Morgan Stanley and others argued the selloff was overdone (Toshiba lacks HAMR technology and relies on external component suppliers). Hypothesis: reversal up, because the stock had gained roughly 200% year to date and nearline capacity was reported as essentially sold out. Risks: heavy momentum unwind, rising yields, and memory names were weak on Oct 6; repeated selling days can extend. Not confirmed: the Oct 6 close and a current quote. One article cited about $945 on Oct 2 and a market cap near $215B, but that is stale, so reference_price and market_cap_usd are omitted, as is 20-day dollar volume.
MRVL
Source session: Marvell rose about 7.1% in the morning of Oct 6 (close not confirmed) after raising long-term revenue targets at its Investor Day; the stock was up more than 240% year to date. Catalyst: partly substantive (explicit multi-year targets) but already heavily priced after a huge run. Hypothesis: reversal down, as a fade of a stretched move on a day when the 10-year yield hit about 5.35% and equity futures turned lower. Risks: analysts were bullish and AI-infrastructure demand remains strong, so momentum could carry on; shorting a strong-trend stock is high variance. Not confirmed: reference price, market cap, 20-day dollar volume and the Oct 6 close (omitted). It is a very large, liquid semiconductor company, so the floors are very likely met.
OPCH
Source session: Option Care Health was up about 32.9% in the morning of Oct 6 (close not confirmed), the top gainer on TheStreet's morning list. Catalyst: substantive; CD&R and McKesson announced a definitive agreement to acquire the company (SEC 8-K, Oct 6, 2026), and a report of a buyout worth over $5B including debt had come out earlier. Hypothesis: continuation up, which means a small drift toward the deal price from a deal-anchored stock; the expected move is small and the confidence is low. Risks: merger-arbitrage spreads often stay flat or fade after day one, and the per-share deal price was not visible in the excerpts I could reach, so I could not size the spread. Not confirmed: reference price, market cap (about $3.7B in August, before the jump) and 20-day dollar volume, so those fields are omitted. The market cap very likely clears the $1B floor but is not verified.
Other models on the same run
| Model | Picks | Average directional return | Directions called correctly |
|---|---|---|---|
| Gemini 1.5 Pro | OPCH, CIEN, TLN, TWST, CEG | -1.45% | 40% |
| Perplexity | OPCH, STX, GH, MRNA, WDC | +0.25% | 40% |
| Grok 4.5 | OPCH, CEG, CIEN, TWST, TXG | +0.60% | 60% |
| GPT-5.6 Sol | OPCH, VST, LW, CEG, STX | -0.70% | 60% |
| DeepSeek | PENG, STX, CEG, VST, OPCH | -1.56% | 40% |
Written up on ForecastNest Insights: Extreme Movers: Claude Sonnet 5.5's picks for 2026-10-07.
ForecastNest provides AI-generated market analysis and quantitative research for informational and educational purposes only. Nothing here constitutes financial, investment, legal, or tax advice.