Do you know that thing in which coders start reviewing many problems and solving them consecutively in increasing level of difficulty and sometimes classifying by category. So, for example you just hear about meet in the middle and begin looking for problems that use that technique all over the world? And they call it practice?
I wouldn't recommend it. Specially if you are talking about a long term period.
I noticed it the other day. Although learning the theory and the implementation of problems is useful it is not the same as practicing for contests. During a contest, you don't know in advance what technique you will need to solve the problem, you have a time limit and you have the stress caused by time limits and the impending doom of losing everything you have dreamed of because of a one letter mistake.
I know, I know, someone suggesting to not "practice" is surprising. Let me tell you something, I used to do that, it seemed like a good idea at the time. There was a moment in which I actually solved 20 problems a day. (That moment was the month before the first and only on-site final I qualified to). But I have stopped doing it for a couple of years already and not intentionally. It turns out that you get tired of it. You start one year thinking "Ok, this is the year I'll get better and get plane tickets and stuff" and you set a practice schedule to solve problems everyday, for the first week it works then you eventually get bored and stop doing it. That by itself should tell you it is not very effective.
This year I had the most incredible set of consecutive matches in topcoder. To the point I almost became red (which is less relevant to me than you would expect).
You would think "Wow, vexorian must have gone to some sort of hardcore programming camp.". That's the cool thing, I didn't. All the period before and during this streak I didn't actually "practice" at all. Instead, that period was heavy on me because of having to write editorials and trying to write problems. That is when I noticed why those things were more effective than practice, they are closer to training for a programming contest because there is stress and time limit involved.
So, there I reached my conclusion. Don't ever solve problems outside of the context of a contest again. It is worthless and gets you bored. Instead, use all that free time you were going to use on a problem solving schedule on something that is far more useful: Participate in contests. Do you think there are not enough contests to do that? There are tons of them: take a look at this calendar if you want. Participating in contests is fun and a far more effective practice in the long term than just solving problems all day. Just try to make sure you do get better after each contest, if you fail at something keep that in mind for the next contests. If you didn't solve a problem because you didn't know some theory, make sure to read about that theory after the contest.
Are you not able to find good contests at a good schedule for you? Well, I heard that people simulate contests with other people when practicing for the team contest that shall not be named. I think you could do something like that, get a coach to assign some problems you didn't previously see before and compete against your own theme with a time limit in solving those problems.