The adequacy of interbank capital is affected by many factors, including the tightness of the entire financial market and the ups-and-downs of interest rates. It has great significance for commercial banks and other non-bank institutions to diagnose the disturbance factors of interbank capital, and predict future capital adequacy level in advance to deploy appropriate countermeasures accordingly. This paper attempts to analyze the relevant factors affecting the interbank capital and to predict the adequacy level of interbank capital based on structured and unstructured financial data. For unstructured data, we crawl the texts from Sina Financial News and then make pre-processing, including word segmentation, emotional word extraction and word-to-vector transformation. For structured data the preprocessing includes padding missing value, data normalization, feature selection and data dimensionality reduction. The prediction models we tried include GBDT, XGBoost, LSTM, SVM, and Perceptron. Experiments show that two-category (loose and tight) average accuracy of the overall adequacy level of interbank capital can achieve more than 94.5%.