Author name Dependable parameters Models Dataset Method Conclusion Madge and Bhatt [30 ] Index momentum, stock price volatility, stock momentum SVM model was computed to check if the efficient markets hypothesis is subjected to prove true as they move on uncertain randomly NASDAQ-100 technology stocks. It constitutes 34 to 39 companies. The time interval is 2007 to 2014 70% data for training and the remaining 30% for testing. Radial kernel was used for the computation The results proved EMH true. Qiu and Song [31 ] Hit ratio, two types of input variables were involved ANN algorithm Nikkei 225 value index of the Japanese stock companies The model was tweaked with the help of G.A. to make it less convergent and more precise The model showed an 81.27% accuracy rate Tsai and Chen [32 ] Average of cluster parameter, probability, standard deviation ANN and decision trees The stock prices of the electron industry in Taiwan. All the stock values were collected from TEJ One hidden layer was considered 1 and 0 could be the ideal choice for the decision model (1 for prices rise, 0 for prices fall) Prediction accuracy of 77%. Only DT gave 65% accuracy, and only ANN gave 59% accuracy Senol and Ozturan [33 ] Coefficient of determination, average mean squared errors, close price Eight ANN models were prepared for seven discrete prediction systems to calculate indicators of P.S. algorithms Istanbul stock exchange (ISE-30) Stock prices decrease when output is greater than or equal to 0 and less than 0.5 which stays the same when output is equal to 0.5 which increases when output is greater than 0.5 and equal to or less than 0 78.47% was the maximum success rate, and 50% was the minimum success rate for a specific stock Nti et al. [34 ] Probability of crossover, probability of mutation, genome length, population size, number of generations The homogeneous ensemble classifier known as GASVM, which is focused on the G.A. This helps choose and optimizing SVM 10-day price movements were predicted by the Ghana Stock Exchange (GSE). Time duration from Jun 25, 2007 to Aug 27, 2019 Models such as preferable RMSE, MAE, AUC, accuracy, or recall were conditionally used to evaluate the dataset efficiency GASVM showed the best results, and the proposed model provided an accuracy of 93.7% Hao and Gao [35 ] Loss function, learning rate, price trend LSTM, SVM, CNN, NFNN, and multiple pipeline models S&P 500 stock prices. Time duration from Jan 30, 1999, to Jan 30, 2019 Multiple LSTMs to learn the time dependency of features of different time scales The LSTM model was proven to be better, and the model had an accuracy of 74.55% over one month Budiharto [36 ] Open, High, Low, and Closing (OHLC) LSTM model P.T. Bank Central Asia Tbk and P.T. Bank Mandiri. 80% for training and 20% for testing The performance of the model was computed based on different time intervals and varying numbers of epochs The optimum results were utilized in a period of 1 year and a total of 100 epochs. The accuracy of 94.59% is achieved Bhupinder and Santosh Closing price, price differences, daily return Hybrid multiple regression models Dataset used in Indian Stock Exchange and S&P 500 (largest exchange fund) The price fluctuate method is implemented using the relationship of independent variables and dependent variables on the basis of Our approach achieved the overall accuracy of 96.3% which is the best of all the above techniques
