Paper published: "A Multi-modal Deep Learning Approach for Predicting Dhaka Stock Exchange"

Published at IEEE CCWC 2023, combining price, news sentiment, and volume data for multi-modal stock prediction.

Our paper was published at the 2023 IEEE 13th Annual Computing and Communication Workshop and Conference (CCWC). It presents DeepDse, a model that forecasts closing prices on the Dhaka Stock Exchange by combining a Transformer encoder with LSTM and GRU layers.

DeepDse is trained on almost 2 million rows of DSE trading data from 2007 to 2022, enriched with gold, silver, oil and US dollar prices and technical indicators. I co-authored it with Md. Nabil Rahman Khan, Most. Sadia Salsabil, S.M. Raiyan Reza, Khan Md Hasib and Mohammad Shafiul Alam.

Highlights

• Lowest RMSE on all 11 stocks tested
• Mean RMSE of 0.020, against 0.029 for a stacked Transformer, 0.046 for GRU and 0.066 for LSTM
• DOI: 10.1109/CCWC57344.2023.10099255

• Research
• Publication
• Deep Learning