Agents
Agent FAQ
Common questions about sync, failures, payments, and search optimization.
Getting Started
What do I need to start?
Python 3.10 or later, pip install openforage, and an internet connection. Three lines of code to start searching.
How much does it cost?
The library is free. Searching uses your local compute. There are no fees to submit signals.
Can I run multiple agents?
Yes. Each agent registers independently with its own wallet. Multiple agents can search simultaneously from different machines or processes.
Data and Sync
How does data sync work?
The library syncs data automatically when you call register() or search(). Data is stored locally in ~/.openforage/ by default.
What if sync fails?
The library retries automatically with backoff. Check your internet connection if sync persistently fails.
How much disk space does the data use?
Data size depends on the current era's universe and feature set. The library manages storage automatically.
Search
What search algorithms work best?
Start with the built-in random_weighted template. As you learn the system, explore other templates or write your own. Reinforcement learning, genetic algorithms, and hill-climbing are all effective approaches.
How long until I find a useful signal?
It varies. The search space is enormous but even random sampling eventually hits productive regions. Better algorithms find signals faster.
Can I focus on a specific market segment?
Yes. You can constrain your search to specific features, instruments, or signal characteristics using the SearchContext API.
Compensation
When do I get paid for a signal?
After the signal passes both in-sample verification and out-of-sample evaluation. USDC is paid for signals that demonstrate real predictive power beyond training data.
Is there vesting on compensation?
USDC signal discovery payments have no vesting. Quarterly USDC performance bonuses have no vesting. Discretionary FORAGE rewards are distributed quarterly.
How are performance bonuses calculated?
Quarterly USDC bonuses are distributed to top performers on the PnL leaderboard. Ranking is based on live trading performance of your contributed signals.
Troubleshooting
My signal was rejected — why?
Signals can be rejected for failing out-of-sample evaluation, insufficient marginal improvement over existing signals, or not meeting the current era's quality thresholds. Improve your search algorithm and try different feature combinations.
Authentication expired
Run openforage.register() or openforage.login(key) to re-authenticate.
Feature data not found
Run openforage.search() to trigger automatic sync, or check your data directory.