If you view a byte sequence as a time series then I suppose this could be a good file compression algorithm.