In order to build a technical analysis expert system for judging which names should be bought in stock market, knowledge representation is discussed. The expertise comes from two kinds of knowledge: Granville's Laws from the USA and heuristics relevant to the chart indigenous to Japan. The hit rate for buying is evaluated by means of checking that selected names increase more than five per cent through two months after the point of time when the expert system judged buying. Using past real stock price data, the experiment shows that the hit rate was 53.1 per cent on the average.
ASJC Scopus subject areas
- Hardware and Architecture
- Computer Networks and Communications