Using Fuzzy Logic to Predict Winners in Horseraces at the Champ de Mars

In this paper we have used a Fuzzy Logic approach to predict winners at the Champ de Mars race course. We have built a fuzzy logic system with 5 input variables and 1 output. The Fuzzy Inference System (FIS) also consists of 35 fuzzy rules based on expert experience. These rules are analysed turn by turn to evaluate the input values and enable the FIS to determine the output which is an aggregate of the individual results from the evaluation of each rule. The FIS is based on Mamdani algorithm which uses the centroid technique during the defuzzification process to produce a single value result. To our knowledge, this is the first real application of fuzzy logic in horse racing. The system was tested for 3 consecutive race meetings with a 77.7 % success of predicting horses finishing in top 3 places for 2 out of the 3 race meetings tested and an average of 23.1% horses winning for the 26 races used for testing. The system could be enhanced by adding more valuable input parameters and the addition of more rules which could provide greater flexibility and coverage. The knowledge gained from this research might be useful for the application of fuzzy logic to other areas like decision making in business, selection of models in construction firms, automating controllers and others.