It’s using the screen space reflections which can only reflect what is currently visible, you can adjust the SSR quality in the post process volume, but you should also add some reflection capture actors which will render out full reflections to cover areas where the SSR doesn’t.