Before trying to give a proper answer to your question, I'd make some remarks.

Both h264 and h265 are highly customizable: as Markus Schumann remarked, there is a "vast set of parameters" by which you can control quality and size. You don't say which encoder you used and which parameters you set. Also, you don't say anything about the quality of the video file you are transcoding. I think that the most sensible thing would be to start from the original file and transcode it to h264 and to h265 with similar parameters. For example, if you are using ffmpeg and are using the defaults, you should go with the defaults for both. Then you could make a comparison and try to find out which parameters should be set in order to allow the h265 encoder to accomplish a file size reduction of about 50% compared to the h264 encoding. I am really at a loss to understand your purpose in re-encoding from h265 to h264. It's not to reduce the file size: you say you expected to get more or less the same file size. Is it quality? But then, change some parameters and go with the h265 encoding again, from the original file.

I know that some people try to reduce the file size by re-encoding to h265 video files that were encoded in h264, which sort of makes sense — not so much sense actually, unless the h264 encoding was of extremely high quality with a very high bitrate, so that the loss of visual quality due to an additional encoding pass would be negligible. But: h265 to h264? No.

As to your question. Quite a lot depends on the quality of the file that is transcoded. The h264 pass would try to reproduce all that it finds in the h265 file, including artifacts, blurred areas, etc. In a bad quality video, artifacts are often crisp and well-defined, and your h264 encoder would try to reproduce them as best it can. You seem to think that the quality loss from the h265 pass would be "understood" as quality loss by the h264 encoder, but this isn't true; it would probably add a good deal of complexity to the sequence of video frames you are trying to re-encode. This is my best guess about the increase in size, since I know nothing about the encoders you are using, the parameters, the video quality etc.

As an advice, I'd say: stick to a single (one or two pass) encoding and try to set your parameters right for reaching a good compromise between file size and video quality. But remember that no combination of parameters is so versatile to cover all cases. If you use ffmpeg libh265 with default parameters on a blurred 640x480 video, the file size you get may be higher than the original; but if you set crf to 31 or 32 you might accomplish a substantial reduction in size with no visual loss.