During a recent research trial, we set out to explore the possibilities of artificial intelligence when paired with Unity data. By supplying Claude with genuine stock identifiers from Unity alongside a single prompt, the system successfully engineered 1,400 novel features from the available information. Furthermore, this process yielded Neutral Benchmark Contribution (NBMC).
Seeing these results naturally leads to an exciting question regarding what your own agent could accomplish with this same data. We warmly invite you to read our comprehensive guide to see exactly how this experiment was conducted and to discover how you can construct your own unique features utilizing Unity.
You can access the complete walkthrough here: https://blog.numer.ai/one-prompt-generates-1-400-features/
Additional resources are now available to support your experience with Unity. To ensure you are fully equipped for the inaugural Unity round taking place on November 6, we have introduced enhancements to our onboarding process, Numerai Predict, Numerai CLI, and Model Uploads.
For further details regarding the Unity dataset, please visit the following link: https://forum.numer.ai/t/the-unity-data-release/8367
The evolution of agentic research on Numerai has reached an exciting new milestone with the introduction of Unity. This development empowers agents to go beyond simply building models from pre-existing features, granting them the ability to engineer their own.
Ahead of the formal submission period starting on November 6, participants currently have the opportunity to send in their Unity model predictions to verify their results early.
Feel free to dive into the details and discover the Unity dataset by visiting https://forum.numer.ai/t/the-unity-data-release/8367