Forecasting stock market trends is a challenging task in computational finance due to the volatile, nonlinear, noisy, and time-dependent behaviour of financial time-series data. Stock prices are influenced by interactions among historical price movements, trading volume, momentum variations, investor sentiment proxies, and rapidly changing market conditions, which reduce the effectiveness of conventional statistical and machine learning models. To address these limitations, this study proposes a deep learning–driven framework for stock market trend classification through a comparative analysis of Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU) networks using a multi-indicator technical signal fusion strategy. The framework employs historical Open, High, Low, Close, and Volume (OHLCV) data integrated with key technical indicators such as Simple Moving Average (SMA), Exponential Moving Average (EMA), Relative Strength Index (RSI), Moving Average Convergence Divergence (MACD), Bollinger Bands, stochastic oscillators, and volatility-based features to capture trend strength, reversal behaviour, overbought or oversold signals, and market instability. A structured preprocessing pipeline involving normalization, sliding-window sequence generation, and feature alignment is used to construct robust temporal inputs for recurrent learning. The model classifies market behaviour into bullish, bearish, and sideways trends. Experimental evaluation using accuracy, precision, recall, F1-score, and classification stability demonstrates that multi-indicator feature fusion improves predictive reliability, while LSTM and GRU effectively capture temporal dependencies, enabling robust support for algorithmic trading, portfolio optimization, and financial decision-making.