Of course it's not just you

I've been practicing for about 2 months now, I have almost 400 problems solved:

800 x128 900 x53 1000 x76 1100 x56 1200 x12 1300 x7 1400 x3 1600 x3 1700 x2 1800 x1

On average, it takes me about 4-6 minutes to solve 800-900 level tasks, but about once every 30-40 tasks there will be a task with this rating where I will be stuck for 30-40 minutes, if not more. (And this is assuming that I have already solved a lot of tasks. If you have solved less, then you should not worry)

I study with a mentor at my university, and having received information from him I can pass it to you:

1. Tasks of the level 800-1400 in 95% of cases do NOT require any specific knowledge: no graph theory, no trees, no dfs, bfs. Basically these are tasks with the tag “constructive” where you need to use 49% your logic and 51% your experience from past tasks.

EXCEPT(topics worth knowing for a beginner): - Binary search - Basic math (what is gcd, lcm, properties of remainders, what you can do with operations, what is xor) - Prefix sum\Suffix sum. Conclusion1: It is not worth studying complex topics at the start and trying to dive deep into them, at least not until you reach the 1400-1600 rating.

2. Don't get frustrated that an idea takes a long time to come if you haven't solved enough tasks. Most of the top people on this platform solve problems quickly because-because every problem uses something that's already been there once. They see constructs that have already been written in code once before, that have already been solved once before. It's just that all these constructs are “hidden” under a clever formulation of the task, which makes it seem to us that such a task has never happened before. Conclusion2: Solve tasks in large numbers and in the future you will see familiar patterns

3. Technique for solving a problem outside of a contests When you solve a problem. Allocate yourself a time of 30 minutes. If during this time you haven't come up with any idea — don't be afraid, don't insult yourself, just go to Editorial and try to get into the problem solving as much as possible. This will be 10 times more useful than if you spend 5 hours trying to come up with an idea for task 800, and in the end it turns out that you've overcomplicated everything and the task is actually simpler than it seems. Conclusion3: Don't be afraid to read Editorial.

*Personal advice from me: When the explanation from Editorial is not clear to me and I still can't understand the solution, I open DeepSeek and ask it to explain the problem to me\explanation from Editorial.

4. After you have solved a task and received Accepted — Go to the tab with other people's solutions and read the code of top users for this task. This will help you develop the skill of concise code writing, teach you some tricks (for example, you spent 100 lines of code to fill a matrix with diagonals, and a professional did it in 10 lines. Study his code and you will do the same next time) Conclusion4: Read other people's code after each parcel

5. Don't fixate on simple tasks, even if you think that higher-level tasks are impossible for you. Sometimes you need to step out of your comfort zone. And I'll say more — you really won't succeed BUT ONLY AT THE BEGINNING OF YOUR PATH. If you solve for example 1400 rated problems (this is not a call to solve them right now), it will automatically improve your skills in solving 800-1300 problems. Conclusion5: Don't be afraid of difficult tasks

6. Create virtual contests. Before every contest, most people have stress and excitement that can ruin everything and cloud your mind, causing you to perform worse at the contest than you could have. Virtual contests will give you the same emotions, but you won't lose your rating for it. Conclusion6: Create virtual contests at least once every 1-2 weeks to get rid of the excitement

7. Resolve your old tasks after some time. This method is used in chess and is called “The Woodpecker method” where you solve 100-200 problems in a month. Then you give yourself a break for 3-5 days. Then you try to solve the same 100-200 problems in 2 weeks. Then in a week.

Personally, I do this: I solve those tasks that I consider necessary. And after 2-3 months, when I already forget the conditions and solutions, I just open some random tasks from the solved ones and try to break my record: - By code brevity - By solution time: (Thinking time + realization time).

Of course, these are not all the tips, but at least those that I remembered at the moment and that I use personally.

Some days I think my iq drops and I can’t solve anything higher than 900. These past two days have been one of those days…

Bruh I practise 1700-1900, mostly I solve without editorial or hint. and still get stuck in easy problems during contest, if it was "only you" then i would have been 1800+ till now. But I am living in Specialist region for so much time, it feels like I will die as a specialist.

i cant solve 800s sometimes too

Oh yes, i brick C in every d2 i take

Of course it's not just you. I myself don't practice at all, as you can see by my heatmap that I haven't practiced any problems after I got to expert. I still got to CM anyways, but that's exception, not the norm.

Judging by your CF handle, I suspect that the year you were born could have significant implications to your current problem solving performance. 1971 was a turbulent year for Bangladesh. That means you probably had a traumatic childbirth which has significant impact on your performance. Also, you are 50+ at this point, so it might also have a role to play on your performance.

But you must remember that you are not alone, and rating is just a number. The most important part is that you are happy with what you have achieved and tried.