By the end of the experiment, I'd seen the majority of algorithms and data structures problem classes that I'd likely ever see in an actual interview. Just doing the daily practice was enough to keep my muscle memory sharp, and in an interview setting, being quick to bang out an algorithm almost always gives you good marks.
To test it out further, after I did my first interview (and felt I aced it), I did not do any daily practice for about 3 weeks between that and my next one. The difference was shocking. I was noticeably slower at answering very similar questions, and occasionally got stuck on things I knew I should know.