The traditional assessment method of volleyball teaching and training effect can't adapt to the informationization and modernization of modern education. On this basis, a new assessment method is put forward, that is, using the SVM model. This method makes full use of the high efficiency of small sample learning of SVM, achieves excellent learning effect, reduces the divergence error and subjective factors in traditional assessment methods, and is more in line with the macro-oriented assessment conclusion. The research results can be combined with information entropy, fuzzy mathematics and other research methods to further enhance the accuracy of data fitting. This method has certain reference significance for improving teachers' teaching level and promoting the improvement of instructional quality. Starting from deepening the effect of volleyball teaching and training, and combining with the characteristics of volleyball events, this paper evaluates the significance, function and methods of volleyball teaching and training. It is pointed out that the assessment of volleyball teaching and training effect is very important for the management of college volleyball teaching and the control of volleyball instructional process. Finally, it is analyzed that the prediction accuracy of DBT-SVM algorithm is higher than that of SVM and multi-class classification algorithm (BT-SVM) algorithm. The accuracy of SVM is 93~95, that of BT-SVM is 95~97, and that of DBT-SVM is 97~98, with the highest accuracy of 98.33. The test time of this method is also the shortest, with an average of 6~7ms; The training effect of volleyball is different in different fields. The DBT-SVM algorithm proposed in this paper is feasible to evaluate the training effect of volleyball teaching