-For the NZProgComp, depending on what level you are depends on which questions you should target. Look at last years problem set if you can to get a feel for it. For 3rd years and above, you shouldn't do any of the 3 pointers. Chances are you're not going to get more than one 100 pointer out, so select which one you want to tackle early and work on it. Not all questions are created equal, there will be some 30 pointers that are easier (for you at least) than most of the 10 pointers. A strategy that worked well for us is when we got our questions, was to put the 3ptrs back in the paper slip and put it away, we weren't going to look at them until after the competition, give the 10 pointers to one predetermined member who was going to start coding them up immediately while the other two members would read the rest of the problem set and look for questions that are easier than their point values suggest (but be careful of hidden problems). We'd choose at least one 100 pointer to attempt and we'd start writing out high level pseudo code for the problems. 

-100 point questions look easy to do by brute force. However you have a time limit of about 180 seconds. Most of the problems you'll see when brute forced will be O(n!) or at least O(n** 2). If n is large (which it inevitably is) you're program may not complete for a million years or more. You will need some critical insight to reduce the search space massively. Correct solutions to these problems usually complete in under 5 seconds.