The manipulations in Borsa Istanbul were analysed using Long Short-Term Memory (LSTM), an artificial intelligence model, in this study. The goal of the study is to determine the patterns of manipulated transactions over time and to evaluate the LSTM model's effectiveness in predicting them. The analyses were conducted using a dataset comprising 12.549 daily observations from 106 manipulation periods involving 88 publicly listed firms identified as having engaged in market manipulation, as reported in the bulletins of the Capital Markets Board of Türkiye during the period 2020 to 2023. The model's accuracy was evaluated using performance metrics. The results show that the proposed LSTM model achieved high performance on both the training and test data. This study can be presented as a perspective for regulators to predict and truly classify stock manipulation on bourses using the LSTM model.
Fettahoğlu et al. (Tue,) studied this question.