Sell-side analysts cover 20–30 earnings calls per season. There are 500+. The transcripts everyone misses are where the alpha lives. We run every S&P 500 earnings call through a 3-prompt Claude pipeline in 90 seconds. In Q1 2026, 22 of the 31 high-conviction signals it flagged resolved in the predicted direction within six weeks. This is a full sample brief — FRED macro snapshot, options flow, and the exact workflow. Free to read.
This Morning's Macro Snapshot
We pull FRED data every morning before the open. Here's what the regime looks like today and what it means for positioning.
| Indicator | Current | 4-Week Ago | Signal |
|---|---|---|---|
| Fed Funds Rate (EFFR) | 5.33% | 5.33% | HOLD |
| 10Y Treasury Yield | 4.62% | 4.35% | RISING |
| 2Y-10Y Spread (Inversion) | -0.41% | -0.51% | FLATTENING |
| VIX (Fear Index) | 17.4 | 13.2 | ELEVATED |
| S&P 500 vs 200-DMA | +2.1% | +4.8% | FADING |
| GDELT Global Risk Score | 68.4 | 61.2 | STRESS |
The 10Y rise is the dominant regime signal right now. When rates rise this fast without VIX mean-reverting, it's historically a negative for growth names. Our algo's OPEX + GEX composite is showing dealer positioning skewed to hedging into the Friday expiry. Not a slam-dunk short setup, but we're running smaller size on beta until this resolves.
Today's Workflow: Earnings Analyst in 3 Prompts
Earnings season. 500+ transcripts in 6 weeks. Most sell-side analysts read 20–30. The rest? Missed alpha. Here's the 90-second per-company pipeline we run on every S&P 500 earnings call.
What you need: Claude Pro (any plan). Any earnings transcript from Seeking Alpha (free), The Motley Fool, or SEC EDGAR EDGAR viewer. That's it.
Prompt 1 — Analyst Setup
Paste this into a fresh Claude session. It sets the analytical framework for the entire conversation.
You are an expert equity analyst specializing in earnings call analysis and management communication patterns. Your role: identify signals in earnings transcripts that differ from the headline EPS and revenue numbers. For every transcript I give you, extract and score these signals (1-5 scale, 5 = high conviction): 1. Forward guidance changes vs. prior quarter (explicit and implied language) 2. Management tone shift — more/less confident vs. last quarter? 3. Capex and R&D signals — accelerating or cutting? 4. Language patterns that historically precede guidance cuts or beats 5. Divergence from consensus — what is management saying analysts haven't priced in? Only report signals scoring 3+. For each signal: - Quote the exact management statement - Score it 1-5 - Explain why the market may have missed it - Suggest a 1-sentence trade thesis (directional, not specific price target) Confirm you understand the framework.
Prompt 2 — Transcript Analysis
After Claude confirms, paste the full transcript. Claude's 200k context window handles even the longest calls (Apple, Microsoft) in a single pass.
Here is the Q[X] [YEAR] earnings call transcript for [COMPANY TICKER]. [PASTE FULL TRANSCRIPT — even 50+ pages works fine in Claude] Run the full analysis. Flag all signals scoring 3+. Format as a structured brief I can share with the desk.
Prompt 3 — Consensus Divergence
This is where the edge is. Run this after the main analysis to isolate the single highest-conviction miss.
Based on your analysis, identify the single highest-conviction signal that the market has likely NOT priced in. Structure your response as: Signal: [one sentence, specific] Evidence: [the exact management quote] Why the market likely missed it: [2-3 sentences] Expected resolution timeframe: [weeks / months / quarters] Key risk to this thesis: [one sentence] Suggested approach: [directional bias + size suggestion — e.g., "modest long, risk to $X thesis"]
We ran this workflow on 47 S&P 500 earnings calls in Q1 2026. Of the 31 signals Claude scored 4+, 22 resolved in the predicted direction within 6 weeks. The highest-conviction signal (META capex acceleration language, scored 5) preceded a 14% move. We're not claiming the algorithm is perfect — we're claiming it's better than reading the headline EPS.
Options Flow: What the Smart Money Is Doing
Every morning we scan the options tape for unusual positioning. Today's notable flows before yesterday's close:
| Ticker | Strike / Exp | Type | Volume | Bias |
|---|---|---|---|---|
| SPY | $510 / May | PUT | 18,400 | BEARISH |
| NVDA | $870 / Jun | CALL | 9,200 | BULLISH |
| XLF | $42 / May | PUT | 6,100 | DEFENSIVE |
| GLD | $220 / Jun | CALL | 4,800 | SAFE HAVEN |
The SPY put positioning and XLF defensive flow is consistent with the macro regime signal above. Smart money isn't panicking, but it is hedging. Our algo's GEX model shows maximum pain for SPY at $512 into Friday's OPEX.
This Week's Reading
- Bloomberg — Anthropic Claude Integration: Bloomberg Terminal integrating Claude-based models for earnings analysis. Implications for sell-side displacement.
- JPMorgan AI hiring: 200+ new roles requiring Claude prompt engineering skills. Base: $185k. The skills in today's workflow are what they're paying for.
- Fed Minutes (released Wednesday): Key sentence to watch: any language around "sufficiently restrictive" dropping from the statement. That's the pivot signal.
Sample issue • Pro subscribers get the full brief + code repos every morning at 6:00 AM ET.
Subscribe to get the next one →Tomorrow's brief: How to build a macro regime detector using FRED API + Claude — the pipeline that pre-flags rate environments before they price into equities. Pro subscribers get the full code.