Stock’s selection and trend prediction using technical analysis and artificial neural network

Abstract

Stock trading offers potential profits when traders buy low and sell high. To maximize profits, accurate analysis is essential for selecting the right stocks, timing purchases, and selling at peak prices. The authors propose a new method for selecting potential stocks that are highly likely to rise in price. The method has two stages. First, technical analysis, using moving averages and stochastic oscillators, filters stocks with downward trends, anticipating a reversal and subsequent rise. Second, for selected stocks, future price trends are predicted using artificial neural networks, specifically long short-term memory (LSTM) with adaptive moment estimation (Adam) optimizer. The second step ensures that only stocks with increasing prices will be chosen for trading. This study analyzes five hundred Fortune 500 stocks over three different periods, with 250 days of data each. Simulations conducted in Python showed that technical analysis could filter 5 to 6 candidate stocks. Subsequently, the LSTM model predicted that only 4 of these stocks would have an upward trend. Validation shows that trend predictions are correct, resulting in an average profit of 5.51% within 10 working days. This profit outperforms the profits generated by other existing methods.