Stock forecasting has long drawn people’s attention because the stock market is a crucial source of financing for publicly traded corporations and a sizable investment market. To fully use the evidence from dissimilar typical prices and recover the stock forecasting effect, a Blockchain-based secure stock value forecasting model TL-EMD-LSTM-MA (TELM) is projected. Other methods are selected for prediction according to the oscillation frequencies of the details, and the high-frequency components use the depth of the transfer learning method to train the stacked LSTM. Deep transferable learning-trained stacked LSTMs incorporate data from several equities and have a deeper understanding of the marketplace or commerce, which can significantly lower forecasting mistakes. Furthermore, it is possible to more accurately estimate the low-frequency components and the overall trend of the stock by employing the average movement approach. 500 stocks in the stock market are shown, as well as indices such as the Stock Exchange Index and Stock Exchange Component Index, the outcomes demonstrate that compared with other models, TELM has the least prediction error and the highest goodness of fit. Finally, simulate the stock trading process based on the stock closing price predicted by TELM, and the results show that TELM investment has low risk and high returns.

股票市场是上市公司重要的融资来源，也是一个规模庞大的投资市场，股票预测长期以来一直受到人们的关注。为了充分利用不同典型价格的证据并恢复股票预测效果，提出了一种基于区块链的安全股票价值预测模型TL-EMD-LSTM-MA（TELM）。根据细节的振荡频率选择其他方法进行预测，高频成分使用深度迁移学习方法来训练堆叠LSTM。经过深度可迁移学习训练的堆叠 LSTM 整合了来自多个股票的数据，并对市场或商业有更深入的了解，这可以显着减少预测错误。此外，采用平均变动方法可以更准确地估计股票的低频成分和整体趋势。展示了股市中的 500 只股票，以及证券交易所指数和证券交易所成分指数等指数，结果表明，与其他模型相比，TELM 的预测误差最小，拟合优度最高。最后，根据TELM预测的股票收盘价模拟股票交易过程，结果表明TELM投资风险低、回报高。