Рус

Eng

Рус

Eng

What is allowed and what is not?

Qualification
round guide

Recommended hardware specifications

Key dates

Tasks and scoring system

Key qualification round dates

The qualification round page on All Cups: cups.online/en/rounds/1804

Qualification round start date: September 21, 10:00
Qualification round end date: October 6, 23:59
Online class on preparing for the qualification round: September 19, 15:00
Online class on analyzing the qualification round tasks: October 12, 15:00
Deadline for publishing the cutoff score: October 12, 23:59
Deadline for publishing the list of main round participants: October 12, 23:59

* Moscow time (GMT +3)

Qualification round tasks

During the qualifying round, participants will solve 6 tasks and submit their solutions to the platform:

2 algorithm tasks that require writing code;
2 mathematics tasks on topics directly related to data analysis;
1 machine learning task;
1 data analysis task — a series of 10 analytical questions based on tabular data.

In machine learning and data analysis tasks, each submission must include a Jupyter notebook (.ipynb) with executed cells and saved outputs that reproduces the submitted answer file. A submission without such a notebook is not counted and the task is considered unsolved.

What you can code in

PROGRAMMING LANGUAGES

C++

20

Node.js

14

Rust

1.85

Kotlin

2.1

PHP

7.4

C#

Mono 6.12

Go

1.20

Java

11

Python

3.11

For algorithmic tasks, we recommend C++ and Python.

For machine learning tasks — Python.

Scoring system

The maximum score for the Qualifying Round is 1,000 points. Based on the scores received, an overall ranking of participants is formed in descending order of results.

Algorithm tasks (Station 32 and Relay Communication Windows)

Your code is checked automatically on a set of dedicated tests.
Points are awarded for each passed test (see Table 1 for details).
If your output does not match the reference answer, the test is worth 0 points.
Each task is worth from 0 to 100 points.

Mathematics tasks (Tunnels and Dispatcher's Log)

Solutions are checked automatically against a reference answer — the correct answers are hidden.
Points are awarded only for a fully correct solution.
Each task is worth either 0 or 100 points.

Machine learning task (Silence of AZ-26)

What is assessed is the prediction accuracy of your model, measured by the ROC-AUC metric.
The test set has public and private parts: the public part is used for intermediate feedback, while the final ranking is based on the private part.
Points are awarded based on the participant's percentile rank among all participants.
The task is worth from 0 to 300 points.

Data analysis task (Predictive Satellite Diagnostics from Telemetry)

10 analytical questions on tabular data, arranged in order of increasing difficulty.
Each question is scored separately; the points per question are specified in the task statement.
The task is worth from 0 to 300 points (the sum of points for all questions).

Hardware specifications

To be perfectly positioned to solve the qualification round tasks, make sure that your workspace matches the following specifications:

Processor: at least 5 cores, core clock speed of at least 2.1 GHz;
RAM: at least 16 GB;
SSD: at least 128 GB;
Graphics card: Nvidia, at least 8 GB RAM;
Monitor: at least Full HD (1920 x 1080), 60Hz refresh rate;
Keyboard and mouse.

💡 You do not need your own GPU: model training is available in free cloud environments (for example, Kaggle with a free GPU).

⚠️ Please note: Your computer does not necessarily have to meet the above-mentioned requirements (such as the storage device type and capacity, processor clock rate, etc.). However, having lower specifications may result in longer execution or debugging times. Take possible delays into account when planning your time.

Uploading the solutions

You can upload your solutions to the All Cups platform from any device. However, we recommend using a computer.

Each task has a daily submission limit:

algorithm tasks — at most 100 submissions per day for each task, and all submissions count towards the limit, including those that ended with an error;
the data analysis task — at most 5 submissions per day;
the machine learning task — at most 20 submissions per day.

For the data analysis and machine learning tasks, submissions rejected because of a format error do not count towards the limit.

All Cups has an automatic solution validation system. On average, coding tasks take 1–2 minutes to process, and ML tasks — up to 10 minutes. If many solutions are uploaded at the same time, a queue will be set up, and the validation process may take longer.

Here is a detailed guide on using the algorithm task validation system.

Checking result codes

Error FAQ

After each upload, your solution receives a short code.
Here is what it means and what to do.

OK	Correct answer	The task is passed — congratulations!
WA	Wrong answer	The answer is incorrect. Check your logic and the output format against the task statement.
PE	Presentation error	The number of lines in the output does not match what was expected. Fix the answer format.
RE	Runtime error	The program crashed and did not produce an answer. Check for exceptions and crashes.
ML	Memory limit	The memory limit (128 MB) was exceeded. Optimize your RAM usage.
TL	Time limit	The execution time limit (2 s) was exceeded. Speed up your algorithm.
IL	Idleness limit	The idle time limit was exceeded — the program "hung" or is waiting for input that isn’t there.
SV	Security violation	The program performed unsafe operations. Remove system calls.
CR	Crash / non-zero exit code	The program returned a non-zero value. Check your return code and program termination.
SE	Server error	An error in the checking system — not your fault. Report it to an organizer.

Rules and grounds for disqualification

Each participant shall comply with the Olympiad’s Regulations and Rules.

Grounds for disqualification:

using several accounts or trying to log in using another person’s account;
attempting to hack the solution validation system;
publishing or discussing the solutions before the end of the qualification round;
using another person’s solution, even with modifications;
disclosing one’s solution to other participants;
using AI tools (ChatGPT, Gemini and others), including the AI answers shown by search engines;
Internet search is allowed: you may use documentation, reference materials and search engines. Only AI tools and the AI answers that search engines show in their results are prohibited.

Disqualification procedure

After the solutions are submitted, they are checked for plagiarism and attempted violations.
If violations are discovered, they are reported to the jury.
The jury considers each incident and makes a decision.
If a violation is confirmed, the results are annulled, and the participant is eliminated from the ranking.
An email is sent to the participant, describing the violation.
The participant can appeal within 24 hours.
The jury considers the participant’s arguments and may restore their score.

An appeal may be filed only against the jury’s decision on disqualification, within 24 hours of receiving the notification email. No appeals are accepted regarding the scores or the results of the solution checks of the qualification round.

10a Radio Street, bldg. 1, Moscow

Federal State Autonomous Educational Institution of Higher Education “State University of Education”, 2026

Qualification round guide

Qualification
round guide