Models are ranked by their weight and correlation within Numerai's meta-model. The meta-model is an ensemble of all our data scientist's predictions. Generally, a lower logloss score results in a better rank.
You retain all intellectual property rights to your model. You never have to tell anyone how you built it and you never have to tell us who you are. You only upload your predictions.
Use this dataset to train your machine learning algorithm. The
target column is the binary class you’re trying to predict. All other columns are features.
Once you’ve built your model, you can use it to generate probability estimates on this new dataset.
To upload your predictions, sign in and click the Upload Predictions button on the leaderboard. The format of your prediction upload should be a CSV file with two columns:
probability. The probability column is the probability estimated by your model of the observation being of class 1.
The public leaderboard displayed on the homepage is calculated on a subset of the total tournament data. Earnings are calculated on the remainder of the tournament data: the private leaderboard. This discourages overfitting to the public leaderboard.
Earnings accrue based on the meta-model rank on the private leaderboard. They are presented based on the public leaderboard rank, but accrue based on the private leaderboard rank.
Earnings are denominated in USD but paid out in bitcoin.
You may not have more than three accounts on Numerai. Duplicate accounts may be banned or penalized.