Data
Data Structure
The Numerai Crypto dataset is a tabular dataset that describes a token universe and their returns over time.
At a high level, each row represents a token at a specific point in time identified by its symbol
and the date
. The date
represents the day that "market close" (23:59 UTC on a given day) data was used to generate the features - in the context of rounds and live data, this is the day before a round opens. The target
is a measure of future returns (eg. after 30 days) relative to the date
.
Target
Numerai Crypto has one target, which is the returns of the token after 30 days. For each round, the 30 day returns for each token in the universe are ranked, gaussianized, then bucketed into five bins.
Data API
The best way to access the Numerai Crypto dataset is via the data API:
train_targets.parquet
contains the historical symbols and targetslive_universe.parquet
contains the latest token universe with no targets of the current round
Last updated