Auto comment: topic has been updated by phsads (previous revision, new revision, compare).

You might consider taking a 20–30 minute walk in the fresh air before the contest. It really helps.

Generally I believe it is more productive to try things than to talk about which things to try. Improvement is very complex, so specific advice may not address your weaknesses. Maybe you can attempt problems above your rating, practice speed-solving problems below your rating, practice by topic to find which ones you are weak in, practice proving your solutions, etc. It's anyone's guess what will truly be effective, and I think it's unlikely that the precise answer you need will be given. But as long as you keep trying new things, it is impossible to stay stuck forever.

I think sleep before contests is helpful

Or when you feel bad or angry by some problem, open the windows, and watch out. It can make calm you down

I think overall you're consistent enough because your highest minus is -71 (which is around 2100 perf, 200-300 worse than your current rating).

Your rating in the end of 2025 also higher than the start of 2025 (2398 vs 2334).

(Rant) In the other hand, I think I even in worse situation because my perf in my worst day is 800-1000. My best day is 2000. Despite my avg rating is around 1500 (the difference is so huge that I think I'd also seeking for advice the same as you're looking for)

"at least not until i hit a specific rating"

And what is your goal?

wow we made the exact same mistakes for the last two contests (I also lost time implementing D2 and I also finished implementing D like 5 minutes after the contest)

Maybe you could try for a more simpler approach when training. Get rid of those "virtual contest 3 times a day", or "read editorial after 3 hours" bullshit, cause they are the easy way out.

The science about improvement is complicated, but first and foremost, you can't really force progress. You have to enjoy what you are doing. Take as much time as you need, do things that you like, and learn things that you want to learn. Even if there is an upcoming competition (I take that you are still in high school, so there are the OI contests), trying to shove things down your throat isn't the way to go.

That worked for me for pushing past my plateau, so maybe you could try.

...how is this inconsistent? Sometimes we do better, sometimes worse. You have mostly low deltas here and none above 100 in absolute value.

You misread statements from time to time? Everyone does. Pay more attention if you want it to happen less.

You have bugs from time to time? Everyone does. Same. There are well-known ways to simplify debugging but doesn't seem like it's an area in which it'd give you tremendous gains.

You overthink from time to time? Everyone does. I specifically overthink a ton of things but it's partly because I'm not just aiming to get ACs ASAP but to figure things out.

Sounds like your main bottleneck is speedforces which is inevitable with the number of problems and contestants. For every tiny speedup you get, you'll have a noticeable jump in rating at your rating. The only way to get faster is practice practice practice, but when you say you don't have time for virtual contests...

If you were very stressed or got little sleep, are you saying you wouldn't perform noticeably worse? If so, that's a very impressive skill you got, one might even call you superhuman... I assume you're just expecting extremely good results when the actual effect is avoiding extremely bad results. There's also a contradiction between "I'm more relaxed" and "I can't stop obsessing over rating" which makes me doubt your ability to assess not just the outcome but also the condition.

This reads like "help I'm not perfect how can I become perfect". You're making normal mistakes. You'll be making them if you're nutella, just on harder problems since you'll breeze through easier ones. Tons of practice can get you higher, you just have to invest that time and effort.

Nice pfp