The sharp price fluctuations in the cryptocurrency market, particularly in Bitcoin (BTC), create significant risks while simultaneously offering speculative profit potential for investors. Traditional analytical methods are often ineffective in detecting non-linear patterns present in stochastic financial time series data. This study proposes the application of a Deep Learning model utilizing the Long Short-Term Memory (LSTM) architecture to project the directional trend of Bitcoin prices (whether upward or downward) for the upcoming one-hour period. In the model's development, historical price data is integrated with a set of crucial technical variables, including the Relative Strength Index (RSI), Moving Average Convergence Divergence (MACD), and Exponential Moving Average (EMA), which serve as input attributes to enhance accuracy. Market data is retrieved in real-time via the Binance API, covering the last 1000 candlesticks. Experimental results using a Stacked LSTM architecture demonstrate that the model achieves an accuracy rate of 51.08% on the test data. Although this classification accuracy is considered moderate, a simple backtesting simulation indicates a positive profitability potential of 2.88% with a win rate of 48.39%. The output of this research also includes a web-based system prototype that integrates a Python backend with a visual interface for real-time monitoring of prediction signals.