judge_logo.png


A protocol for quant funds and researchers to work together without revealing their IP.


Introduction

We completed our alpha test over Q2-Q3 2022. It was a data test, designed to see the accuracy of the forecasts that can be generated when quants coordinate in a decentralized manner without revealing their core IP to one another.

Why they Participated

Besides helping to build a new way to work in one of the systemically most important parts of the economy, there are massive first mover advantages built into the system. Funds that submit useful features claim their ‘land’ in the feature space. These early subspaces will be of extraordinary value going forward.

You can think of the early reward structure as analogous to a standard DeFi project, where the first participants do very well, except that here those outsized gains last far into the future, as long as the same features continue to be submitted.

For more details, see **How it Works** and our Tokenomics Overview.

Who Participated

A group of a dozen or so funds or individual quants & devs, the lion’s share of which have decades of experience at top buy & sell-side institutions.

How They Participated

The main workflow was to send features into Judge Research’s API. You specify which series the feature is meant to help explain - e.g. spot BTC-USD at the 4 hr periods - and a several minor details. The first time data was sent into the API, the user sent in all of the past observations, from the most recent time period to the first observation date, which is specified in our API doc. ****Subsequently, one observation was sent just before the live discrete time period ended.

Exact Specifications of the Dependent Variables

In the long run one will be able to choose their own depedent variables. For the alpha test, we focused on three variables over two discrete time periods, leaving participants with six options:

  1. The percentage change of ETH-USD: (close - open) / open
  2. BTC-USD’s percentage change: (close - open) / open
  3. The volatility of those two series, as defined by (high - low) / open

At 15 minute & 30 minute discrete time periods.

The historical time series you submit for should start at 2021-10-01 00:00:00.

For submitting your live data, you must submit it before the turn of beginning of a new discrete time period. For more details, see Scheduling Your Live Send Scripts.

Conclusion