AN EVALUATION OF ML TECHNIQUES ON NASDAQ DATASET FOR STOCK MARKET FORECASTING

Abstract

The uncertainty of stock pricing has popularized stock market prediction as a common practice. Forecasting prices in the market are viewed as problematic, as the hypothesis of efficient markets (EMH) explains. According to the EMH, all accessible information is represented in market prices, and price variations are only the consequence of newly available information. The approach to prediction forecasts the market as either positive or negative based on a variety of input parameters. A combination of derived, fundamental, and pure technical data is utilized in stock forecasts to project future stock prices. Algorithms for machine learning (ML) are made to find patterns in data and utilize those patterns to forecast future events. K Nearest Neighbour (KNN) can process relationships between the numerical data, it is particularly effective in numerical prediction problems for predicting changes in stock value the following day. KNN categorizes freshly input data according to how similar it is to previously taught data, and it does this by clustering the data into coherence subsets or clusters. Using the KNN approach in conjunction with technical analysis, the closest neighbour search strategy yielded the desired outcome. KEYWORDS: Hypothesis, K Nearest Neighbour, Prediction, Stock prices,