Paper: arXiv 2611.14325

Authors: Alessio Brini

Abstract

Text encoders turn earnings calls into vectors for search, for language models that answer from retrieved passages, and for forecasting models. A live retriever can use an encoder trained once on all available text, as finance retrievers are, but a backtest needs an encoder that read no text written after the date of each vector. To address this look-ahead bias, from 412,248 calls we train vintages under four pretraining recipes, with yearly cutoffs from 2010 to 2024, and each vintage reads only text dated on or before its cutoff. We then fine-tune each vintage on the question-answer pairs of the calls themselves. The vintages we train on calls alone, with 33.8M parameters, retrieve answers on held-out calls better than the general retrievers we tested in every test quarter. Their vectors also raise by about half the rank correlation between next-day returns and forecasts built from past returns and firm size. When each vintage continues the previous one, a later vintage retrieves better on earlier quarters than the vintage available at the time, and a point-in-time language model that answers from the passages the later vintage retrieves inherits the gain. The 2024 vintage of our families also forecasts earnings surprises better, although the interval of this gain includes zero on calls that no vintage read. It forecasts returns no better, so point-in-time vintages matter most for retrieval and little for forecasts of returns.

Complexity vs Empirical Score

  • Math Complexity: 5.0/10
  • Empirical Rigor: 8.0/10
  • Quadrant: Holy Grail — high math complexity, high empirical rigor

Why this score: This paper presents a novel and empirically rigorous approach to address look-ahead bias in text encoders for financial applications. The methodology is well-defined, and the extensive backtesting across multiple vintages demonstrates strong empirical rigor. The mathematical complexity is moderate, focusing on the application of existing NLP techniques rather than novel theoretical derivations.

Research Flowchart

  flowchart TD
    A[Research Goal: Address Look-Ahead Bias in Text Encoders for Financial Data] --> B{Methodology: Point-in-Time Encoders};
    B --> C[Data/Inputs: 412,248 Earnings Calls];
    C --> D{Computational Processes: Train & Fine-tune Vintage Encoders};
    D --> E[Outcomes - Retrieval: Better Retrieval on Held-Out Calls];
    D --> F[Outcomes - Forecasting: Improved Rank Correlation for Returns & Earnings Surprises];
    E --> G[Impact: Point-in-Time LMs inherit retrieval gains];
    F --> H[Limitations: No better return forecasts for new calls];