Abstract
In the financial markets, because real-time transactions directly relate to profit, it is important to process and analyze data on a real-time basis. In practice, decisions influenced by experts’ experiences from fundamental and technical analysis occur frequently compared to decisions using prediction algorithms. A domain-specific data mining framework was proposed recently to reduce related cost. Therefore, this study proposes a novel data mining framework suitable for financial markets according to expert knowledge. The proposed framework predominantly considers the following three perspectives as the standards for the effectiveness of research: interpretability, proper prediction metrics, and reporting methods. We applied our framework to the real-world financial prediction problems, such as the 3–10 year treasury spread forecasts. Consequently, we achieved an 84% prediction performance on the spread prediction and used hierarchical information to provide additional insight. In addition, we obtained practical knowledge and synergies through extraction of critical variables that can be used as a quick and accurate data-driven decision making support tool by active agents in the real world.
1. Introduction
Through several different channels via diverse means, financial markets play important roles in the well-being of enterprises and the overall economy. A financial market transfers real economic resources, offers dividends or interest to the market participants, creates liquidity, and enables trade among the investors in the market. Therefore, it is important to analyze factors that are linked organically to the financial market and to derive latest information from the huge data available. Many investors use several sources of information to predict target values and develop strategies to gain an edge in competitions.
5. Conclusion
In this study, we proposed a data mining framework designed specifically for the financial market with the following three characteristics: interpretation, prediction evaluation metrics, and reporting methods. Firstly, a prediction model should provide predictive power and interpretability. We proposed that such an objective can be achieved by employing wrapper approaches if developing a prediction model, wherein the extent of contribution of each variable in the analysis can be observed. Secondly, the financial prediction model requires evaluation using appropriate metrics. For example, it may be important to predict the range of a target outcome, rather than the exact value, to incorporate tolerance. In this case, PARE is an appropriate measure of evaluation. Finally, presenting the results from the analysis in several reporting modes is required, particularly to aid active traders in the market to instantly decide with detailed explanations.
