Summary: | 碩士 === 華梵大學 === 資訊管理學系碩士班 === 100 === In this study, used the recurrent neural network(Elman Recurrent
Neural Network) and the back-propagation neural Network
(Back-Propagation Neural Network, BPN) to build the model, the Taiwan
Securities Exchange(TWSE) cooperation with the Financial Times and
Stock Exchange(FTSE) compiling Taiwan 50 Index for the subject, to
explore its every other day for the closing of the forecast, then two
predictions of the model and operating strategy of a combination of
simulated trading research can profit.
Research data to the Taiwan Securities Exchange(TWSE) cooperation
with the Financial Times and Stock Exchange(FTSE) compiling Taiwan 50
Index, since September 1, 2008 until September 30, 2010 only, a total
of 522 document data. Taiwan 50 Index data, and international stock
markets, after-hours IT TAIEX combined total of 24 technical
indicators, as neural network input variables.
From September 1, 2008 to September 30, 2010 total of 522
documents data as training and test data . Elman and analyzed by
the BPN, in days, for testing, training, explore the neural network to
forecast results and operating strategies combined with the possibility of
profit from the stock market.
According to experimental results show that the use of neural network
prediction predict the Taiwan 50 index closed the next day with the trading
strategies simulated trading can be profitable; the use of neural network
with the trading strategy in the Taiwan stock market, profit is feasible.
|