numerapi

library2.23.2pypypi✓ verified 84d ago

Python client for the Numerai machine learning competition. Automates data download, prediction upload, staking, and account management. Current version 2.23.2, released monthly.

ai-ml

pip install numerapi

github ↗homepage ↗

INSTALL

IMPORT

SIG · NUMERAPI

numerapi

ai-mlpythonv2.23.2

harness data pending

Install & Compatibility

Where this runs

No compatibility data collected yet for this library.

Code

Verified usage

Verified import paths — ran on the pinned version, not inferred.

NumerAPI

✓ from numerapi import NumerAPI

✗ import numerapi

NumerAPI is the main class, not a module.

Initialize NumerAPI and download the latest training data.

from numerapi import NumerAPI napi = NumerAPI() # download current dataset napi.download_current_dataset('training_data.parquet', dest_path='./')

Debug

Known issues

breakingThe `sign` method from v1 was removed. Use `napi.upload_predictions()` with `model_id` parameter.

fix

Use `napi.upload_predictions('path/to/predictions.csv', model_id='your_model_id')`.

affects: >=2.0.0

gotchaThe `download_dataset` method returns a path string, not a file object. Many users expect a file handle.

fix

Store the returned path and use it for reading: path = napi.download_dataset('filename'); data = pd.read_parquet(path).

affects: all

gotchaAPI keys must be set via environment variables `NUMERAI_PUBLIC_ID` and `NUMERAI_SECRET_KEY` or passed directly to constructor. Not using environment variables leads to authentication errors in production.

fix

Set environment variables before creating NumerAPI: os.environ['NUMERAI_PUBLIC_ID'] = '...'.

affects: all

Errors

Common errors & fixes

NumerAPIError: 401 - Unauthorized

Missing or invalid API credentials (public_id/secret_key).

fix

Set NUMERAI_PUBLIC_ID and NUMERAI_SECRET_KEY environment variables or pass them to NumerAPI() constructor.

ValueError: No model_id provided and no default model set

Calling upload_predictions without specifying model_id when no default is configured.

fix

Provide model_id parameter: napi.upload_predictions('preds.csv', model_id='your_model_id').

FileNotFoundError: [Errno 2] No such file or directory: 'numerai_training_data.parquet'

Download path not correctly specified; napi.download_dataset returns the path, but user tries to use a hardcoded name.

fix

Use the returned path: path = napi.download_dataset('numerai_training_data.parquet', dest_path='./'); data = pd.read_parquet(path).

Upgrade

Version history

2.23.2latest on PyPI · released Jun 2, 2026

Audit

Dependencies

requestsrequiredHTTP client for API calls

python-dateutilrequiredDate parsing

Agent activity

9 hits · last 30 days

node

Resources

githubhttps://github.com/uuazed/numerapi ↗

homepagehttps://github.com/uuazed/numerapi ↗