If I remember correctly, the estimate of "every 500 years" is invalid. Averaging the estimated periods between the events doesn't mean anything when there is such a huge disparity between the longest known period between events and the shortest.
The value of 500 years was calculated by a moving temporal average of a poisson distribution written on a slip of paper and pulled out of a hat.
And because they have information that dates only one prior event. That information is based upon core samples of cedar trees that were submerged after the last quake. The link between that and the great tsunami is largely anecdotal (journals of a Japanese emperor at the time that described the tsunami though it could have been caused by another undersea quake around the same time).
Once again, science is shaming itself by trying to establish trends based upon little precedence, as they are doing with global warming.