Asian Journal of Research in Computer Science

In recent decades, the prediction of financial markets based on artificial intelligence has expanded a lot, which has led to the emergence of non-parametric models in this field. Models based on historical data provide traders with high accuracy predictions and do not require simplifying assumptions such as the absence of arbitrage in the market. Machine learning and especially deep learning are one of the newest topics in this field, which has been used the most in studies in recent years. According to the risk in the capital market, the use of derivative instruments, especially the option contract, is necessary for investment risk management. Forming a portfolio with the lowest available risk and with a return close to the return of the entire market is something that many investment companies are looking for. As a result, the need for a tool to predict the price of these contracts is felt, and the most important variable for pricing option contracts is implied volatility. This research is looking for a model to predict the implied volatility of option contracts using deep learning techniques, so that the prediction of this model can be used to estimate the price of option contracts. For this purpose, a modeling of target points and loss limits in the New York trading market is considered. In the proposed model, first, a probabilistic neural network with spokes, clustering operation and then classification at the data level are performed, and then the time series method based on the Brownian curve, based on control theory, can reduce dimensions, select and extract features. Based on the proposed approach, it has been shown that based on the Brownian curve, it has the ability to optimize the results of the probabilistic neural network, and the results of the combined approach, in addition to having the problem of high computational complexity, have more optimal results in terms of evaluation criteria, including accuracy.