How many questions, on an average basis, shall I solve daily, to become a top-level competitive programmer?

As many as you’ll manage to solve while practicing intensively instead of running for numbers.

This question in fact make not too much sense. You don’t even give definition of “top-level competitive programmer”. I know people who say a coder with yellow rating is basically a stupid newbie. I know people saying that making into CF div1 means being a top-level contestant, an inspiration for everyone around. Which one fits better for your scale?

You didn’t put any timetable - do you want to reach that level in a year, or two, or ten years?..

Number of problems that you solve isn’t the exact indicator of how productive you are, or how intense is your preparation, or how much new stuff you learned. There is some probability that you’ll use fixed number to motivate yourself: “I haven’t solved enough problems today, I should keep practicing instead of going to do something else”. But, on the other hand, there is decent probability that you’ll intentionally or unintentionally switch to easier problems in order to successfully get that number each day, or you’ll not spend enough time on problems - not reading editorials or codes by other contestants after you solved a problem, or not even figuring out all the details of solution after getting AC (that especially matters for people who are often getting AC using intuition or guess as key step of solving a task).

I’m currently having a training camp, mostly preparing to World Finals and doing something related to competitive programming for around 10–12 hours per day, and I will try to recall my daily routine over last several days so you may try to figure out how many question I’m solving daily.

On Thursday we had team training. The team got 8 problems AC’ed there, I don’t really remember how many of them did I solve. After that in the evening I participated in Codeforces round, in which I solved 5 problems.

On Friday we had a team training. The team got 8 problems solved; once again I can’t say how many of them did I solve - in order to do that I’ll need to open problemset and go through it problem by problem, and I’m not even sure it will help. In the evening I spend most of the time on past contests - reading editorials and upsolving problems, and also I was busy for several hours with some things not exactly related to competitive programming training.

On Saturday we had a training at which we solved 9 problems. In the evening I started working on approximate problem from Bubble Cup, rewriting parts of it from scratch, making a few dozen submissions and spending a few hours thinking on what to do there and how to solve given problem in at least a little bit smart way.

On Sunday we had day off, so I participated in Yandex.Algorithm, solving 5 problems there, and then sort of did Russian Code Cup (leaving it in the middle of the contests while having 3 problems solved). For remaining part of the day, I was working on approximate task from Bubble Cup, making a total of 250+ submissions and spending a few hours on coding, several hours on thinking and several hours on simply trying different stuff to see if it works. Late in the evening I decided to stop because I was pretty sure that I’m not able to improve my results much without coming up with something different/more optimal. Therefore I switched to other task from Bubble Cup. I coded a solution which turned out to be too slow, then coded another solution and yet another one, without being successful.

Finally, on Monday, we had a training in which we solved 8 problems - this one was today so I remember it rather clearly; I sort of solved 1 task (which took me around 1 minute and which could have been solved by any member of the team - I was just lucky to read it first) and tried to solve one more, ending up wasting almost 2 hours of computer time before teammate helped me to get the actual solution. After the training I got back to Bubble Cup problem, coded another solution which was still too slow, and then spend a few hours modifying and tweaking it to finally get AC.

Now I got a few questions for you :) What is my average number of problems per day during timespan I described? Does solving ~2 problems on Monday mean that it was much less productive than ~8 problems on Sunday? Also, since I’m solving around 2 problems per team training on average (probably even less), does it mean I’m wasting a lot of time there, spending whole 5 hours and not getting a decent number of “problems solved by me”? These questions should give you an idea about my point of view on the question that you asked :)
courtesy-Bohdan Pryshchenko(Quora)