Modeling was the core of data warehouse building, and the dimensional modeling technique was generally adopted. After analyzed the advantages and drawbacks of star and snowflake schemas in dimensional modeling, the securities business was taken as an example, the concept of a derived sub-dimension schema was proposed from the star according to the features of financial data warehouse, stated the conditions, algorithms and advantages of the structure. In the typical star schema, for the large dimension table, the relative independent field with low cardinality was broken into sub-dimension tables, which were associated with the super-dimension and fact tables by surrogate keys. The structure benefits data loading, improving query performance of data warehouse, and decreasing of data redundancy to some extent. It is practical valuable in multi-subject financial data warehouse modeling.