ab-be / numerox

Numerai tournament toolbox written in Python

numerox

Numerox is a Numerai tournament toolbox written in Python.

All you have to do is create a model. Take a look at model.py for examples.

Once you have a model numerox will do the rest. First download the Numerai dataset and then load it:

>>> import numerox as nx
>>> nx.download('numerai_dataset.zip')
>>> data = nx.load_zip('numerai_dataset.zip')
>>> data
rows      637184
era       133, [era1, eraX]
x         50, min 0.0000, mean 0.5025, max 1.0000
y         mean 0.499924, fraction missing 0.3095

Let's use the logistic regression model in numerox to run 5-fold cross validation on the training data:

>>> model = nx.logistic()
>>> prediction = nx.backtest(model, data, verbosity=1)
logistic(inverse_l2=0.0001)
      logloss   auc     acc     ystd   stats
mean  0.692885  0.5165  0.5116  0.0056  region     train
std   0.000536  0.0281  0.0215  0.0003    eras       120
min   0.691360  0.4478  0.4540  0.0050  sharpe  0.488866
max   0.694202  0.5944  0.5636  0.0061  consis  0.691667

OK, results are good enough for a demo so let's make a submission file for the tournament. We will fit the model on the train data and make our predictions for the tournament data:

>>> prediction = nx.production(model, data, verbosity=1)
logistic(inverse_l2=0.0001)
      logloss   auc     acc     ystd   stats
mean  0.692808  0.5194  0.5142  0.0063  region  validation
std   0.000375  0.0168  0.0137  0.0001    eras          12
min   0.691961  0.4903  0.4925  0.0062  sharpe    0.903277
max   0.693460  0.5553  0.5342  0.0064  consis    0.916667

Let's upload our predictions to enter the tournament:

>>> prediction.to_csv('logistic.csv')  # 6 decimal places by default
>>> upload_id, status = nx.upload('logistic.csv', public_id, secret_key)
metric                  value   minutes
concordance              True   0.0898
consistency           91.6667   0.0898
originality             False   0.1783
validation_logloss     0.6928   0.1783
controlling capital     False   0.1783

Examples

Have a look at the examples.

Install

Install with pip:

$ pip install numerox

After you have installed numerox, run the unit tests (please report any failures):

>>> import numerox as nx
>>> nx.test()

Requirements: python, setuptools, numpy, pandas, pytables, sklearn, numerapi, requests, nose.

Resources

• Let's chat
• See examples
• Check what's new
• Report bugs

Sponsor

Thank you Numerai for funding the development of Numerox.

License

Numerox is distributed under the the GPL v3+. See LICENSE file for details. Where indicated by code comments parts of NumPy and SciPy are included in numerox. Their licenses appear in the licenses directory.