A Stock Trend Prediction Model Based on Wavelet Transform and TCN Combined with Market Sentiment

Based on the existing stock trend prediction methods that combine market sentiment, combined with wavelet transform and TCN time convolution model, to improve the accuracy of stock trend prediction. [Method]Download stock numerical data through the interface, mine and analyze potential numerical data, and then use the GRU network model to perform sentiment analysis on the obtained stock news data, obtain sentiment indicators, and use them as training data. Train the stock price prediction model using wavelet transform and TCN time convolutional network model.[Results] After the introduction of wavelet transform and TCN network model, the accuracy of Gree Electric Appliance's stock trend prediction increased by 6.1%, ZTE's stock trend prediction increased by 5.92%, and AAPL's stock trend prediction accuracy increased by 7.45%. [Limitation] The stock market trades according to working days, and in the case of weekends or holidays in the middle, the previous data may not have a significant impact on the current data. [Conclusion] Stock fluctuations are affected by market sentiment. The use of wavelet transform to process stock data can minimize the impact of extreme situations on stock price trends, and the combination of TCN time convolution model can better predict the trend of stock data.