transcript exec tone divergence
What it checks
Compares CEO tone vs CFO tone on the same earnings call. If the CEO sounds optimistic but the CFO sounds cautious, they're not on the same page — historically bearish.
Mechanism
When the CEO and CFO speak with different sentiment tones on the same earnings call, post-earnings drift skews negative. We compute |ceo_tone − cfo_tone| per call from FinBERT scores on speaker-tagged transcript segments, then z-score against the firm's own history. A high absolute divergence implies the strategic narrative (CEO) and financial detail (CFO) are misaligned — a bearish setup. Negative z (unusually aligned tones) acts as a confidence proxy.
Signal rule
SHORT when |ceo−cfo tone gap| own-z >= +1.0 (mgmt not aligned). LONG when own-z <= -1.0 (unusually aligned).
Data dependencies
daily_pricesAdjusted-close OHLCV for every US-listed ticker; primary price feed.
transcript_finbert_scoresWorker data table — see services/worker schema.
Expected edge
- Paper alpha
- ~3-4% ann. L/S (Davis-Ge-Matsumoto-Zhang 2015 follow-up)
- Paper window
- 1995-2010
Modern FinBERT-driven replications report ~3-4% annual long-short on the absolute tone-gap, z-scored against the firm's own history.
Illustrative pattern only
NOT a backtestIllustrative pattern only — see /app for live backtests and the actual current equity curve.
Related families
transcript finbert qa dispersionText-NLPHuang-Wang-Yang (2023 CAR) "FinBERT: A Large Language Model for Extracting Information from Financial Text." FinBERT-tone outperforms Loughran-McDonald wordcount on every post-call drift measure tested (3-day CAR R² roughly doubles). Follow-up Chen et al. (2024 JFE) reports ~5% annualised L/S on the Q&A-dispersion sub-signal. Own-history z of per-call Q&A sentiment dispersion (stdev of per-sentence pos-neg score): short when z ≥ +1 (uncertainty), long when z ≤ −1 (confident answers).
transcript cfo qa defensivenessText-NLPWhen a CFO oscillates between positive boilerplate and negative-tone clarifications during analyst Q&A (high cfo_qa_dispersion in FinBERT scores), they are typically defending weak fundamentals — predicting negative 60-90 day drift independent of the miss/beat itself. The signal isolates *defensive non-answers*: sentence-level CFO tone variance during the Q&A section, z-scored against the firm's own call history.
transcript analyst frustrationText-NLPAnalyst tone during the Q&A section — the *questions* themselves, not the answers — carries forward-looking information independent of the news in the call. When analysts sound skeptical or frustrated (negative FinBERT tone, sharp questions), 60-day drift skews negative. We z-score per-call mean analyst_qa_tone against the firm's own history of analyst sentiment.
Explore transcript exec tone divergence on alphactor.ai
See which tickers this family is currently firing on, with live signals and rankings.