What is an investment signal?
Reported financial statements are a lagging record of decisions made months earlier. Between the moment a trajectory shifts and the moment it is legible in a quarterly print, hiring freezes, capital commitments, altered disclosure language, insider transactions, and competitor moves leave an observable public trail.
This investment signal framework organises eight families — management, operating and labour, capital allocation, filings, insider, competitive, industry and macro, and regulatory — scored on lead time, reliability, noise, and verifiability.
1. Executive Summary Filings-language change is underpriced, accruing over 6–18 months. Postings and opportunistic insider trades lead; capex and tone need two-sided reading. Convergence only counts after shared-driver controls. Strongest support: Lazy Prices textual-change findings and job-posting studies. Cases: Blockbuster vs Netflix (slow decline) and SVB (speed).
2. Signal, Lead and Prediction A signal is a measurable observation. A leading indicator occurred before an outcome historically. A predictive signal adds out-of-sample forecasting power under stated conditions. Most commentary skips from correlation to prediction — this framework refuses that jump, grading each family as peer-reviewed, directional, or mixed.
3. The Eight Signal Families Management (call tone, guidance); Operating/labour (postings, accounting turnover); Capital allocation (capex, R&D — two-sided); Filings (YoY textual similarity, Lazy Prices); Insider (opportunistic trades only); Competitive (pricing, launches, M&A); Industry/macro (production, utilisation, credit); Regulatory (approvals, restrictions). Families 1, 2, 4, 5 carry strongest peer review.
4. Which Signals Appear First Labour/postings: 1–3 quarters, near-real-time, strong. Capex: 2–4 quarters, lumpy, mixed. Filings language: 2–6 quarters, strong. Management tone: 0–2 quarters, strong. Insider: 1–3 quarters, conditional. Competitive/industry/regulatory: weeks to quarters, directional. Weight fast families for warning, slow families for confirmation.
5. When Signals Converge — and When Convergence Lies One budget approval lifting hiring, capex, and tone is one cause counted thrice, not three confirmations. Defensible convergence needs different decision processes, no shared seasonality, and 2–3 independent families each with base rates. Weak = single observation; Moderate = one documented family or two independent alignments; Strong = 3+ independent families after controls.
6. When Signals Fail Hiring gains can precede layoffs. Capex surges preceded weaker returns in some samples. Most insider selling is routine/plan-based. Optimistic tone may reflect budget confidence. Language shifts can stem from new counsel. Rival moves often proxy sector cycles. Survivorship overstates hit rates — validate with point-in-time archives including failed firms.
7. Blockbuster vs Netflix: Slow Public Decline 9,094 stores and 84,300 staff (2004 peak) to 2010 Chapter 11. Public trail: passed on $50M Netflix buy (2000), Viacom spin leaving ~$905M debt, late 2007 streaming entry, Redbox/Netflix streaming pivot. Leads ran 3–10 years. Signals showed a debt-constrained incumbent ceding share — never the filing date.
8. SVB: Framework Under Speed $42B single-day outflow (Mar 10, 2023); stress visible only via Mar 8 raise. Governance led 2–3 quarters: no permanent CRO through 2022, insider loans tripling to $219M in Q4 2022, executive sales, 80%+ uninsured deposits. Read was elevated risk — never a 48-hour countdown.
9. Five Documented Weaknesses 1. No growth or failed-signal case — cannot calibrate hits vs disciplined rejection. 2. Competitive/industry/regulatory rows rest on thin peer review. 3. Sector splits are prose, not analysis — software capex pooled with manufacturing. 4. ~22 sources leave weakest rows thin. 5. Index unweighted and unvalidated — bands are hypotheses, not a model.
10. Five Concrete Improvements 1. Add one growth + one failed-signal case with entry dates and windows. 2. Targeted pass on competitive/regulatory literature (patents, channel, approvals). 3. Standalone sector sidebar: tech, banking, manufacturing, consumer. 4. Disclosed pilot: one family × 20–30 names × one quarter, point-in-time. 5. Upgrade forthcoming citations to final DOIs on publication.
11. Conclusion Filings language and postings carry strongest support; opportunistic insider trades add conditional evidence; capex stays two-sided. Signals were public — sustained joint attention was not. Way forward: point-in-time archives, sector-separated walk-forward tests, disclosed miss rates, versioned methods.
Where Brane Fits: continuous layer — versioned filings for similarity scoring, postings/pricing feeds where licensed, cross-source surfacing vs baselines. Framework and pilots published, not scores, until validation completes.
FAQ Q1: What is an investment signal? A measurable change that becomes evidence only once history, base rate, and context are established. Q2: Leading vs predictive? Leading occurred before an outcome; predictive adds out-of-sample power under stated conditions. Q3: Does this predict prices? No — organises early fundamental evidence, issues no buy/sell calls. Q4: Strongest evidence? YoY 10-K textual change (Lazy Prices) and daily postings; opportunistic insider trades conditional. Q5: Why test convergence? Hiring, capex, tone often share one budget cause — three moves may be one event counted thrice.
Sources: Jiang, Lee, Martin & Zhou (2019); Price et al. (2012); Gutierrez et al. (postings, forthcoming); Khavis et al. (2026); Cohen, Malloy & Nguyen — Lazy Prices (2020); Cohen, Malloy & Pomorski (2012); Fed SVB Review (2023); Business Standard (SVB $219M); Bailey et al. — Seven Sins; Kadoa (2026).
This article is educational and does not make a recommendation about any security. Evidence as of September 2026.