Introduction

Now that you have Launched the SDK, how do you get down to work?

The Basic Workflow

The Design Logic

Because the point of Judge Research is to protect your IP while working with others, there necessarily is a separation between your dashboard on the web portal, the AI & what you use to interact with it. This allows your tools to sit on your server and gives you complete discretion regarding what information is visible to the system.

The Workflow

The SDK includes convenience classes to simplify the process of fetching historical market data, perform statistical and technical analysis, and submit features into the Judge GA.

The basic steps are:

You’ll use the SDK to do preliminary research, send features into the system and write easy-to-share reports for you and your colleagues.
When you are ready to submit features to Judge Research, you’ll first submit the historical data for that series,
And then schedule a script to submit those features live every discrete time period (45 minutes and/or four hours for the alpha test).
You’ll use your dashboard in the Judge Research web portal to receive findings from the SBGA. Once the alpha test begins, you can embed those findings in the reports you create in the SDK.

Toggle here to read about the functionality coming after the alpha test: Spinning up your own private SBGA.

Benefits

Plugging your live findings into your notebooks takes minutes. Likewise, the process of feature engineering is expedited by the convenience functions and clean relationship to outside data provided by the SDK. In very little time, therefore, you can:

Create live, interactive documents hooked up to massively parallel processes that analyze your features and algorithms across millions of modeling contexts.
This creates a level of efficiency & inferential rigour that is difficult to compare.
It also disciplines the design of research to be so fast and well-organized that findings are more likely to cumulate over time.