I wish more programming competitions followed the format of the ICFP. 72 hours, teams of a half dozen or so, solving an open-ended problem, often an optimization problem where an exact solution is NP-hard. You're judged by how well you do, not by whether you get the correct answer (which often isn't computationally feasible anyway).
That seems to most closely approximate real-world programming problems.
I completely agree the ICFP contest matches real-world problems better. I can't recommend it enough. The longer timeframe requires discipline and good planning, while also providing a bit of a buffer for early (minor) mistakes.
Having a strong and well organized team is huge. You'll learn a lot about your teammates in the process, both behaviorally and technically. However, I've had to go solo the last two years and still enjoyed it immensely - it just changes how you prioritize some of the development.
That seems to most closely approximate real-world programming problems.