---
title: Opening the Black Box: Which News Explain LLM-Based Stock Ratings?
url: https://www.ml-quant.com/papers/ssrn/7665703/
site: ML-Quant (https://www.ml-quant.com)
updated: 2026-10-30
license: Summaries CC BY 4.0; links go to the original sources
index: https://www.ml-quant.com/llms.txt
identifier: SSRN 7665703
source_url: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=7665703
featured: 2026-10-30
citations: unknown
topic: LLMs & Text
---


# Opening the Black Box: Which News Explain LLM-Based Stock Ratings?

Research using FinBERT embeddings finds that firm-specific and fundamental news topics explain LLM stock ratings far better than sentiment alone, doubling the model's explanatory power.

- Source: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=7665703
- Identifier: SSRN 7665703
- Released: 2026-10-24
- First featured: Quant Letter No. 137 (2026-10-30): https://www.ml-quant.com/issues/2026-10-30/
- Citations (Semantic Scholar): not tracked
- Published in: not yet
- Topic: LLMs & Text
- Authors: Victoria Kiszczuk

## Related

- [SYNTHEVAL: Hybrid Behavioral Testing of NLP Models with Synthetic CheckLists](https://www.ml-quant.com/papers/arxiv/2408.17437/): SYNTHEVAL is a testing framework that uses large language models to generate tests for evaluating NLP models, particularly in sentiment analysis and toxic language detection.
- [BioFinBERT: Finetuning Large Language Models (LLMs) to Analyze Sentiment of Press Releases and Financial Text Around Inflection Points of Biotech Stocks](https://www.ml-quant.com/papers/arxiv/2401.11011/): LLMs for Financial Sentiment Analysis: BioFinBERT, a finetuned Large Language Model, is introduced for financial sentiment analysis of biotech press releases and financial texts, which greatly impact biotech stock prices.
- [SliceGPT: Compress Large Language Models by Deleting Rows and Columns](https://www.ml-quant.com/papers/arxiv/2401.15024/): Compressing Language Models: The paper introduces SliceGPT, a post-training sparsification scheme for large language models that reduces the network's embedding dimension, maintains high performance, reduces inference computation, and reveals computational invariance in transformer networks.
- [Can ChatGPT Forecast Stock Price Movements? Return Predictability and Large Language Models](https://www.ml-quant.com/papers/ssrn/4412788/): ChatGPT predicts stock market returns using sentiment analysis, outperforming traditional methods.
- [Efficient Adversarial Training in LLMs with Continuous Attacks](https://www.ml-quant.com/papers/arxiv/2405.15589/): CAdvUL, a new adversarial training algorithm, enhances the resilience of large language models against adversarial attacks by efficiently calculating attacks in the continuous embedding space.
- [Sentiment trading with large language models](https://www.ml-quant.com/papers/doi/10-1016-j-frl-2024-105227/): The OPT model, a large language model, has proven superior in predicting stock market returns using sentiment analysis of U.S. financial news, outdoing traditional methods like the Loughran-McDonald dictionary model.