With the development of Internet finance, the original method of financial asset value assessment has been challenged, on this basis based on the application of blockchain technology and machine intelligence models, which provides new ideas for the innovation of the financial market. In this study, a set of data analysis models combining deep neural networks and blockchain are designed as a method for financial product valuation, in which a multilayer perceptron is used to realize the operation for high-dimensional input vectors, in addition to the combination of LSTM-Adaptive Attention model for the value assessment of digital financial products. The above model is empirically tested using large-sample, cross-market data, and a variety of digital currencies such as bitcoin, ethereum, and decentralized financial protocol tokens are selected for comparative study, which statistically significantly proves that the method has a high degree of accuracy and effectively manages risks. In summary, the model proposed in this paper can better capture the value characteristics embedded in digital currencies, improve the effectiveness and reliability of risk control while enhancing the accuracy of price prediction, and provide a powerful support for the healthy development of the financial market of the digital economy as well as the prevention and control of systemic risks.
