Predicting M&A Targets Using ML: Unlocking the potential of NLP based variables
The full article can be found on LSEG’s Developer Portal. It was written by Haykaz Aramyan.
How we can predict Mergers and Acquisitions (M&A) targets using Machine Learning (ML) and Natural Language Processing (NLP) techniques? With ML/NLP libraries, and the London Stock(LSEG)’s Refinitiv Data Platform (RDP) REST API of course!
The overview of the Article
In this article, we extend the M&A predictive modeling that has been published earlier, by incorporating the NLP-based news sentiment variable. The original model used only financial variables and utilized Logistic regression for M&A target identification and showcased if that produces an abnormal return for investors. Instead, the purpose of this article is to test if news sentiment derived by NLP has any significant contribution to M&A predictive modeling. To do that, the significance of news sentiment is tested on different ML models, including logistic regression, random forest, and XGBoost models. When it comes to the NLP model, Finbert and BERT-RNA developed by LSEG Labs models are tested to calculate sentiment on the news preceding M&A announcement.
The Article is utilizing the depth and width of Refinitiv data and APIs, particularly RDP API to access both financial data, and textual (news) data. Additionally, we utilize the RDP Search function to access M&A data and PEER Screener to access Peer data for non-target control dataset.
The conclusion of the Article
The evaluation results on different ML techniques allow claiming the importance of the NLP-based news sentiment variable for the M&A predictive analysis. Although the predictive power of the model isn’t high enough to use this model for an actual prediction and trading, the prediction workflow along with the data ingestion, feature engineering, and model evaluation techniques can be useful for training robust models on much larger datasets and achieve much higher accuracies for actual trading.
While reading this article, do not hesitate to comment on this Medium page if you have any questions.
References
- Getting Start with Refinitiv Data Platform
- RDP Documents
- Summary of Tokenizers
- Transformers from Hugging face
- FinBERT: Financial Sentiment Analysis with BERT
- Financial Language Modelling by LSEG Labs
Downloads
- GitHub — Predicting M&A Targets Using ML: Unlocking the potential of NLP based variables
- Explore API Samples on GitHub
Related Articles
- Predicting M&A Targets Using Machine Learning Techniques - September 21, 2021- Haykaz Aramyan
- Building Search into your Application Workflow - December 21, 2020 - Nick Zincone
- Introduction to News Sentiment Analysis with Eikon Data APIs a Python example -May 05, 2018 - Jason Ramchandani
- RDP News Metadata with JSON Viewer - October 22, 2021 - Zoya Farberov
- Information Demand and Stock Return Predictability (Coded in R): Part 1: Data Retrieval and Analysis - January 05, 2021- Jonathan Legrand
- Finding Expired Options and Backtesting a Short Iron Condor Strategy- November 12, 2021- Haykaz Aramyan
- TensorFlow Variational Quantum Neural Networks in Finance-
November 23, 2021- Dr. Marios Skevofylakas - Explore all Articles