Optimization of algorithmic studies in the Tehran Stock Exchange market using feature engineering with a hybrid Autoencoder-LSTM-CNN model and prediction with an LSTM model