Peter Cundill on statistical overvaluation

From There's Always Something to Do:

I almost stopped selling Japan short in the last quarter of 1989 because I couldn’t stand it anymore. But intellectually I was convinced that I was right and so I carried on and then in the first quarter of 1990 the Japanese market fell by 25% in eight weeks and I made back everything I’d given away since 1987 plus a good deal more. But I tell you statistical overvaluation is a funny thing – it can go on for a very long time, far beyond the limits of rationality, and it is a problem for the value investor in two ways: it can tempt one to compromise standards on the buy side and it may lure one into selling things far too early. I have less of a problem with the selling temptation because I have always loved cash – if you’ve got lots of it you will never have to pass up a great opportunity.