FINANCIAL DISTRESS PREDICTION FROM TIME SERIES DATA USING XGBOOST: BIST100 OF BORSA ISTANBUL

Abstract

This study utilized financial and non-financial data from 233 companies listed in the Borsa Istanbul BIST SINAI Index from 2010 to 2020. The XGBOOST machine learning algorithm was employed to predict whether these companies would encounter financial distress. The machine was trained using supervised learning, with 80% of the data used for training and 20% for testing purposes. Financial ratios were utilized as independent variables in predicting financial distress. The 25 financial ratios can be categorized into four main headings: Liquidity, Financial Structure, Activity, and Profitability Ratios. Furthermore, the model allowed for individual analysis of each company. In predicting whether companies would experience financial distress, the maximum F1 score (85.1%), recall (84.5%), precision (85.7%), and accuracy (91.6%) were achieved.