Extreme Movers: Claude Sonnet 5.5's picks for 2026-10-15
Claude Sonnet 5.5 picked CRWV, NBIS, DELL, BSP, ORCL for a 1-day horizon. Average directional return: +3.06%. Directions called correctly: 60%.
| Experiment | Extreme Movers |
|---|---|
| Model | Claude Sonnet 5.5 |
| Prompt version | extreme-movers-v11 |
| Asked | |
| Horizon | 1-day |
| Measured from | 2026-10-15 |
| Measured to | 2026-10-16 |
| Benchmark | SPY |
| Status | Settled |
| Average directional return | +3.06% |
| Directions called correctly | 60% |
| Benchmark return | +0.88% |
The picks
| Rank | Symbol | Direction | Price change | Return in the called direction | Confidence | vs benchmark | Called right |
|---|---|---|---|---|---|---|---|
| 1 | CRWV | up | +7.03% | +7.03% | 55% | +6.15% | Yes |
| 2 | NBIS | up | +7.96% | +7.96% | 50% | +7.08% | Yes |
| 3 | DELL | down | +0.62% | -0.62% | 45% | -1.50% | No |
| 4 | BSP | down | +1.02% | -1.02% | 40% | -1.90% | No |
| 5 | ORCL | up | +1.96% | +1.96% | 45% | +1.08% | Yes |
The model's market view
The source session (2026-10-14) was a risk-off day. A hot September CPI (core 2.6% year over year versus 2.5% expected) lifted the odds of an October Fed hike, the S&P 500 fell 1.11% and the Nasdaq fell 1.88%, and semiconductors and AI infrastructure names were hit hardest. The sell-off looks macro- and rate-driven rather than tied to company news, so I lean toward one-day mean reversion in the most oversold AI-infrastructure names. I pair that with one analyst-downgrade continuation short and one fade of a one-day sentiment spike. The risks are that rate-hike fears keep pressuring long-duration tech and that Thursday's retail sales report moves yields. I could not reach live quotes, market caps or 20-day dollar volumes for any pick, so those fields are omitted and recorded as unverified. The previous_change_percent figures come from a single closing recap, not an exchange feed.
Why each pick
CRWV
CoreWeave fell about 15.0% in the source session, the worst of the large AI infrastructure names. The drop was a sector-wide rate and AI-financing sell-off, not company-specific news, which makes it a candidate for an oversold bounce. A reversal means direction up. The risk is that its bonds were reported at distress-signalling levels and further rate fear could extend the decline. I could not confirm a current quote, market cap or 20-day dollar volume, so those fields are omitted and the price, market-cap and liquidity floors are unverified. Earlier snapshots put the market cap in the tens of billions and the price well above $5, but I did not use them as verified figures. The -15.0% previous change comes from a market recap citing Reuters closing figures.
NBIS
Nebius fell about 10% in the source session, extending a rough week (down about 11% over the prior week at the October 12 midday quote). The catalyst was speculative and macro: the hot-CPI sell-off in AI infrastructure and AI-financing concerns, with no company-specific news. I expect a partial rebound, so this is a reversal with direction up. The risk is continued selling in neocloud names. I could not confirm a current quote, market cap or 20-day dollar volume, so those fields are omitted and the screening floors are unverified. The -10% previous change comes from the same single recap.
DELL
Dell fell about 11.4% in the source session after Barclays cut it to Equal Weight from Overweight, citing elevated valuation and limited enterprise AI-driven growth so far. The catalyst is substantive (a valuation downgrade) on top of the macro AI sell-off. I expect follow-through selling, so this is a continuation with direction down. The risk is a sharp dip-buying bounce in oversold hardware names. I could not confirm a current quote, market cap or 20-day dollar volume, so those fields are omitted. The -11.4% previous change comes from the closing recap. The downgrade reason comes from an intraday live blog that reported -8.28% at that time.
BSP
Bending Spoons rose roughly 41% in the source session after Berkshire Hathaway invested another $330 million. A midday report had it up 34%. The stock also jumped about 24% on October 7, so this is a second large spike in about a week. The catalyst is a real strategic investment, but I expect some give-back after such a spike and a weak small-cap tape, so this is a reversal with direction down. The risk is that the Berkshire endorsement attracts more buyers and the move continues. The recap's wording on the percentage is approximate. I could not confirm a current quote, market cap or 20-day dollar volume, so those fields are omitted, and this recently listed name's liquidity floor is unverified.
ORCL
Oracle fell about 8.8% in the source session as a bellwether for AI infrastructure and AI-debt concerns, with no company-specific news found. It is a very large, liquid name, but I could not confirm a quote, market cap or 20-day dollar volume, so those fields are omitted. A reversal means direction up. The risk is a further rate-driven de-rating of leveraged AI buildout stories. This pick is correlated with the CoreWeave and Nebius picks, so the portfolio is concentrated in one theme. The -8.8% previous change comes from the closing recap.
Other models on the same run
| Model | Picks | Average directional return | Directions called correctly |
|---|---|---|---|
| Muse Spark 1.1 | ACN, SNPS, CRWV, APLD, CIEN | +2.58% | 80% |
| Grok 4.5 | LQDA, CRWV, DELL, HALO, UNH | -1.89% | 20% |
| Gemini 1.5 Pro | UNH, BMY, KR, GILD, PRGO | +0.60% | 60% |
| GPT-6 | CRWV, DELL, BSP, DOCN, LQDA | -2.19% | 0% |
| DeepSeek | CHWY, UNH, DLTR, BMY, SKYD | -0.28% | 40% |
Written up on ForecastNest Insights: Extreme Movers: Claude Sonnet 5.5's picks for 2026-10-15.
ForecastNest provides AI-generated market analysis and quantitative research for informational and educational purposes only. Nothing here constitutes financial, investment, legal, or tax advice.