DeepTrust: Assessing Social-Media Evidence for Extreme Stock Moves
Summary
DeepTrust is a framework for finding and assessing information that may explain extreme equity price changes. Its workflow has three parts: machine-learning detection of anomalous moves from historical prices, retrieval of related unstructured information from Twitter using queries that adapt to search conditions, and assessment of the retrieved content’s reliability. The assessment considers tweet characteristics, signs of generated language, argument structure, subjectivity, and sentiment to narrow results to a concise set of potentially credible posts.
The framework was evaluated on self-annotated Twitter and Facebook stock anomalies from April 2021. The reported optimal setup exceeded a baseline classifier on F0.5 score and precision for both cases. The authors also examine the retrieval and reliability modules separately and identify factors that limit their performance. Evidence is limited to two annotated anomalies in the description, so it does not establish effectiveness across other events, assets, or periods. The method supports information triage; credible social posts do not by themselves prove the cause of a price move.
Key ideas
- DeepTrust combines price-anomaly detection, Twitter retrieval and information-reliability assessment.
- Dynamic query conditions are used to retrieve social-media material related to unusual price moves.
- Reliability assessment draws on tweet features, language-generation traces, argumentation, subjectivity and sentiment.
- Evaluation on two annotated stock anomalies reports better F0.5 scores and precision than a baseline classifier.
- The small described evaluation limits conclusions about generalization, and retrieved posts do not establish causality.
Tags
Full text
# DeepTrust: A Reliable Financial Knowledge Retrieval Framework For Explaining Extreme Pricing Anomalies # DeepTrust: A Reliable Financial Knowledge Retrieval Framework For Explaining Extreme Pricing Anomalies Extreme pricing anomalies may occur unexpectedly without a trivial cause, and equity traders typically experience a meticulous process to source disparate information and analyze its reliability before integrating it into the trusted knowledge base. We introduce DeepTrust, a reliable financial knowledge retrieval framework on Twitter to explain extreme price moves at speed, while ensuring data veracity using state-of-the-art NLP techniques. Our proposed framework consists of three modules, specialized for anomaly detection, information retrieval and reliability assessment. The workflow starts with identifying anomalous asset price changes using machine learning models trained with historical pricing data, and retrieving correlated unstructured data from Twitter using enhanced queries with dynamic search conditions. DeepTrust extrapolates information reliability from tweet features, traces of generative language model, argumentation structure, subjectivity and sentiment signals, and refine a concise collection of credible tweets for market insights. The framework is evaluated on two self-annotated financial anomalies, i.e., Twitter and Facebook stock price on 29 and 30 April 2021. The optimal setup outperforms the baseline classifier by 7.75% and 15.77% on F0.5-scores, and 10.55% and 18.88% on precision, respectively, proving its capability in screening unreliable information precisely. At the same time, information retrieval and reliability assessment modules are analyzed individually on their effectiveness and causes of limitations, with identified subjective and objective factors that influence the performance. As a collaborative project with Refinitiv, this framework paves a promising path towards building a scalable commercial solution that assists traders to reach investment decisions on pricing anomalies with authenticated knowledge from social media platforms in real-time.
Shown in full with attribution under the source's licence. Licence: abstract CC0
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.