Welcome to this week’s Cmind Earnings Edge.

The heatmap is flashing a familiar late-season regime: dispersion > direction. The tape is rewarding clean execution and clean guidance while punishing anything that widens the forward range—especially in names where expectations are already elevated. That matters this week because the calendar is anchored by one macro node: NVDA.

When NVDA announces, it doesn’t just move “semis.” It acts as a capex/ROI referendum that ripples across hyperscalers, AI infra (power, cooling, data centers), and second-order suppliers. In other words: this is a week where the reaction function is the trade—not just the EPS line.

INTERACTIVE HEATMAP

Feature of the Week

NVDA (Feb 25, post-market): “AI central bank” meets perfect-priced expectations

Earnings Release Date (est.): 2026-02-25 (post-market)
Cmind latest beat probability: 93%
Consensus EPS: $1.45

What changed since the last NVDA print 

Since the last earnings release, the market’s scoring rubric has tightened. It’s no longer “AI at any price.” It’s show me the payback timeline:

• CapEx scrutiny intensified. Investors are increasingly separating spend from returns, forcing management teams to map spend → revenue → cash flow with more precision.

• Custom silicon + competition narratives grew louder. Even if NVDA remains dominant, the market wants evidence that pricing power and margins are durable as alternative supply ramps.

• Options and positioning are more reflexive. This is the “priced for perfection” setup: NVDA can beat and still trade poorly if the guide is anything less than tight.
  

  Why NVDA at 93% is Meaningful

Cmind’s daily probability path has been volatile—not a straight-line ramp—ranging from a ~51% trough (Jan 8) to a ~95% peak (Feb 14) before settling back to ~93% into Feb 20. That volatility matters: it’s the signature you often see when the model digests conflicting inputs and then re-converges late.

Two key “air pockets” showed up:

1. Early January (the probability broke to the low-50s), and

2. Early February (the signal briefly compressed into the mid-60s).
   The most recent move is the tell: a sharp snap back into the low-to-mid 90s, i.e., late-window convergence—the model is increasingly underwriting a beat.

What a strong beat and outlook would mean for the AI industry

• It revalidates the AI infrastructure spend cycle and supports continued underwriting of second-order beneficiaries (HBM, packaging, power, cooling, data-center build).

• It lifts the “payback narrative”: if NVDA can beat and tighten guidance, it de-risks the market’s fear that AI capex is pulling demand forward without durable conversion.

• It sharpens dispersion: a bullish NVDA read-through tends to widen the gap between “real AI throughput” winners and names still selling narrative without measurable economics.

NVDA checklist:

• Guidance: does it tighten the forward range or widen it?

• Margins: any hint of throughput pressure vs sustained pricing power?

• Demand signal: language that confirms order/lead-time strength vs normalization

• Competitive tone: custom silicon / mix / pricing posture

Top Beats / Misses - Week of February 23

✅ Top 6 Beats

• NTAP (IT, Large) — 95%

• BIDU (Comm Services, Large) — 94%

• NVDA (IT, Large) — 93%

• IMAX (Comm Services, Small) — 92%

• AMPH (Health Care, Small) — 92%

• EBAY (Consumer Discretionary, Large) — 90%

⚠️ Top 6 Misses

• UFPI (Materials, Mid) — 2%

• BIOX (Materials, Small) — 3%

• AMR (Materials, Small) — 6%

• IE (Materials, Small) — 8%

• CVEO (Industrials, Small) — 10%

• BYND (Consumer Staples, Small) — 15%

Heatmap  

This week is a two-speed board:

• Right tail clusters: Communication Services + Information Technology show the cleanest high-probability density (BIDU/IMAX on the comm side; NTAP/NVDA on the tech side). Health Care also remains a consistent “execution premium” pocket (AMPH, plus multiple names in the high-80s/low-90s).

• Left tail pockets: Materials is where miss-risk is most extreme (UFPI/BIOX/AMR/IE), while selected SMID names in Industrials and Consumer also sit in the gap-risk zone.

The practical implication for systematic books: build baskets off clusters; hedge off outliers. Don’t overfit a sector narrative when dispersion is doing the work.

Market Cap Exposure 

Mega / Large Cap

Large caps carry the cleanest “tradable liquidity + conviction” setups: NTAP (95%), NVDA (93%), BIDU (94%), EBAY (90%). But dispersion still shows up in isolated large-cap left-tail flags (e.g., DPZ ~26%), where “good numbers aren’t enough” if the guide disappoints. Net: large-cap tape is selective risk-on, not blanket risk-on.

Mid Cap

Mid-caps are where the board gets more asymmetric: fewer names at extremes, but higher reaction sensitivity because the guide can re-price the distribution quickly. Watch mid-cap Materials and Energy where probabilities sit closer to the middle but the tail names are ugly (UFPI as a standout miss-risk). This is often where pairs and beta-neutral spreads screen best.

Small Cap

Small caps are the widest reaction cones: you’ve got a right-tail winner like IMAX (92%), but also deep left-tail names (BIOX/IE/CVEO/BYND). In this cohort, liquidity amplifies gaps—so sizing discipline matters more than being “right” on direction.

Sector Exposure 

• Information Technology: top-end anchored by NTAP/NVDA, but with scattered low-probability pockets—classic “winners separate from met-but-questioned.”

• Communication Services: unusually strong right-tail density (BIDU, IMAX) supporting a cluster/basket lens.

• Materials: the clearest left-tail risk this week (multiple single-digit to teens probabilities).

• Health Care: continues to screen as an execution pocket (AMPH plus a broad high-80s band).

Top Movers (±10 pts - Week over Week)

Below is the Top Movers table for Week of Feb 23, 2026, ranked most positive → most negative.

Top Movers 

✅ Largest Positive Movers 

⚠️ Largest Negative Movers 

Quick interpretation:
This week’s movers are tight and idiosyncratic —with the biggest “confidence ramp” (MANU +37) and the sharpest “fragility reset” (NGVT −26) showing where the model is repricing expectations most aggressively into the next catalyst window.

• Upward ramps = late-window tightening as market is underwriting execution

• Downward resets = guide sensitivity rising

• In dispersion regimes, movers are often more actionable than level.

Setup for next week 

Post-NVDA, the market will immediately try to answer one question: did the guide tighten the AI payback narrative—or widen it? A strong NVDA print with a clean forward range tends to lift the entire AI infra chain (power, cooling, DC build, second-order semis) and compress left-tail risk. A “beat-but-widen” outcome tends to increase dispersion and punish anything with vague conversion timelines.

About the Model

Cmind AI’s EPS predictions are powered by a machine learning model built for accuracy, objectivity, transparency, and daily updates with the latest market information. We ingest over 150 variables across five data modalities—including real-time 10-Q filings, earnings transcripts, governance metrics, and peer signals—to provide early, company-specific EPS forecasts.

Our EPS signals update daily across 4,400+ U.S. stocks using a multi-input ML model (filings, transcripts, price/earnings dynamics, governance, and peer signals). The goal isn’t to predict headlines—it’s to quantify where dispersion is most likely so you can build better baskets, hedges, and sizing into catalyst windows.

📩 To learn more, contact us at [email protected].