Pattern Refinement with Model Data Fusion to Predict Exchange Rate Movement

摘要

The exchange rate depends on various factors. Its movement appears as a result of piled influence of many factors. To predict comming movement, the chart analysis method, which is widely used, pays attention to characteristics in local shapes of the movement by that tune. This paper provides a method to specify intuitive characteristics of time series data. Experts can specify their qualitative knowledge with the method. Their qualitative models are refined with actual data into quantitative ones, which can be handled by computers.