The aim of our work was to analyze the point-fight kickboxing matches for the techniques resulting in the highest score and present a possible prediction model of the points earned/matches won. We evaluated 28 matches. We observed the successful techniques carried out through punches and kicks. A technique was annotated by means of 4 letters depending on the following criteria: use of hand/leg; lead/rear; offense/counter-offense; body/head. In terms of occurrence, the techniques with the highest score are HROH-21%, HLCH-15% and HLOH-10%. In stepwise regression, the number of points achieved by the match winners was selected as a dependent variable and the independent variables were represented by the sum of the individual successfully implemented techniques in the match. 3 statistically significant models were selected out. The highest value of reliability R-2 was discovered in the model with three variables in which we can predict the number of winning points with a 64.5% probability and the error rate of 1.955 points using 35.59% HRCH, 17.08% HLOH, and 11.84% HLOB. The results point to the possibility of quantifying the match, and can be subsequent used in the training process.
