Skip to content

Getting Started

To successfully participate in the Algothon, you will need to set up a local development environment capable of handling data analysis and backtesting.

1. The Starter Code

All necessary data and the evaluation engine have been pre-packaged into our official GitHub repository. We highly recommend cloning this repository to use as the baseline for your algorithm.

  1. Navigate to the Algothon 2026 GitHub Repository.
  2. Clone the repository to your local machine.
  3. The repository will contain:
    • prices.txt: The training data.
    • eval.py: The official evaluation script used by our judges.
    • teamName.py: A boilerplate file for your algorithm.
    • requirements-dev.txt: The exact package set available at grading time - for setting up your local environment only, do not include this file in your submission.

2. Local Environment Setup

We recommend Python 3.12. For local development you're free to use whatever tools you like (Anaconda, a plain venv, Poetry, etc.) - but the packages actually available when your algorithm is graded are a fixed, explicit set, not the full Anaconda distribution. See Accepted Packages for the exact list.

To minimise "works on my machine, fails on submission" surprises, we recommend matching the grading environment locally using the requirements-dev.txt from the starter repo: 

python -m venv .venv
.venv\Scripts\activate      # Windows
source .venv/bin/activate   # macOS/Linux
pip install -r requirements-dev.txt

If your own algorithm needs a package beyond this list, do not add it to requirements-dev.txt - create a separate requirements.txt listing only the extra package(s) and include that in your submission .zip instead. See the Submission Guide for details.

3. Testing Your Algorithm

Before submitting your model to the live leaderboard, you must ensure it executes correctly against the provided evaluation script.

Run your code locally using: 

python eval.py

If your code does not compile or throws errors when run against eval.py, it will automatically fail on the competition servers.