In recent years, a method of machine learning technology has been advancing. Getting the distributed representation of words by the Neural Probabilistic Language Model (NPLM) has attracted interest. In this study, we attempt to create a new dictionary on the basis of ones widely used in finance. In this analysis, we use News Feed Direct of Thomson Reuters Corporation provides (NFD) that focus on the market trends of the Japanese stock market in order to make the dictionary. As a result of analysis, we could succeed in creating a new financial dictionary. However, we also found problems to be solved to get better performance.