Overcoming Context Bottlenecks in Financial Time-Series Forecasting via Dynamic External Memory Augmented LSTMs

Abstract

This paper introduces the Dynamic External Memory LSTM (DEM-LSTM), a novel deep neural architecture designed to address the hidden state information bottleneck and temporal context decay inherent to standard LSTMs in financial time-series forecasting. By decoupling sequence processing from persistent state storage via an addressable external memory matrix ($M_t$), DEM-LSTM dynamically reads, erases, and updates market context across long sequences without corrupting internal hidden representations. Evaluated across four distinct asset classes—Foreign Exchange (EUR/USD), Commodities (XAU/USD and USOIL), and Cryptocurrencies (BTC/USD)—DEM-LSTM consistently outperforms standard LSTM baselines across all metrics, achieving up to a 41.4% reduction in RMSE on Gold spot prices while maintaining superior stability across high-volatility market regimes.