A MODIFIED TRADING STRATEGY MODEL COMBINING NEURAL NETWORKS WITH THE BOLLINGER BANDS TECHNICAL INDICATOR

摘要

The Bollinger Band is a widely used technical indicator to measure and display the volatility of securities by showing whether prices are high with the use of an upper band and whether they are low with the use of a lower band. The bands are based on the volatility (standard deviation) of the past price data. This indicator can aid in rigorous pattern recognition and is useful in comparing current price action to possible buy and sell signals, helping to arrive at a self contained systematic trading decision. However, due to its inherent lagging characteristics, the indicator can provide false signals during trading in some markets. The paper proposes a modified model, combining neural networks with the Bollinger Band technical indicator, to predict and trade on the security trend. The benefit of the combined system is that the neural network will help to overcome the lagging aspects of the Bollinger Band indicator by providing a next day forecast, allowing the trader to make the correct trading decisions. The profitability of the model will be tested using data from the S&P 500, Nasdaq Composite, and Russell 2000, along with stocks of varying levels of volatility, such as Cisco Systems, IBM, Amazon, Pfizer, and General Electric, among others.