Fast Numbers, Slow Language: Bridging Quantitative and Qualitative Earnings Signals

2026-06-28

AI Summary

This study addresses the longstanding disconnect between financial economics and natural language processing in analyzing earnings-related information by introducing EarningsInOne, a unified corpus integrating earnings press releases, conference call transcripts, and corresponding stock price data for S&P 1500 constituents. Employing a consistent long–short portfolio strategy (long top 20%, short bottom 20%) and the Q5–Q1 return spread metric, the paper systematically compares the market impact of quantitative earnings surprises and qualitative linguistic signals. It reveals, for the first time, a “fast numbers, slow words” pattern of information absorption: earnings surprises are rapidly incorporated into prices on the announcement day, whereas sentiment signals from conference calls peak in predictive power the following day, generating significant tradable abnormal returns. Notably, conventional mean squared error (MSE) evaluation obscures the directional forecasting ability of these linguistic cues.

Abstract

Earnings announcements release two types of information sequentially: quantitative surprise (numeric earnings-per-share (EPS)/revenue versus analyst estimate) arrives first in press releases and financial news, processed by algorithmic traders within minutes; qualitative language (management tone, guidance, question-and-answer (Q&A) credibility) arrives 30-90 min later in the earnings conference call transcript (ECT), requiring human interpretation overnight. Financial economists have studied quantitative surprise for 50 years; natural language processing (NLP) researchers have studied qualitative ECT signals for a decade. Despite studying the same event, the two communities used incompatible frameworks: different targets (return vs. volatility), trading setups (long top-decile and short bottom-decile vs. trade-all), and metrics (return spread between top and bottom 20% (Q5-Q1) vs. mean squared error (MSE)), making direct comparison and connection challenging. We bridge these communities with EarningsInOne, the first corpus aligning earnings news, ECTs, and intraday and next-day prices across SP 1500 (broad U.S. equity universe, 2022-2025). Applying unified trading and evaluation tools to both signal types, we confirm a clean speed separation, fast numbers, slow language: quantitative surprise peaks at announcement and is largely eliminated by the next market open; qualitative ECT sentiment peaks on the next trading day, real and tradeable, but hidden under prior transcript-based evaluation that optimised sign-agnostic volatility with pointwise MSE.

Problem

• earnings announcements
• quantitative surprise
• qualitative language
• incompatible frameworks
• signal comparison

Innovation

• earnings announcements
• quantitative vs. qualitative signals
• natural language processing
• market efficiency
• EarningsInOne corpus