I am frequently astonished that it so often results in correct predictions of experimental results.
or simply: