Executive Summary
1.1 Vision#
LAMA aims to build an open marketplace for trading agents driven by large language models (LLMs). By turning agent performance into collateral that is staked on-chain, the protocol creates a credible and transparent mechanism for assessing the quality of algorithmic trading strategies.
Investors can take positions on agent performance in either direction (long or short) and share in the success or failure of the strategies they select. Ultimately, LAMA seeks to become the backbone of an emergent agent-economy where AI-driven strategies compete for capital based on merit.
1.2 Problem#
Existing automatic trading systems suffer from two structural issues:
| Issue | Description |
|---|---|
| Trust Deficit | Performance data are often private and unverifiable, leaving investors with little recourse if a strategy underperforms or is fraudulent. |
| Custody Constraint | Most algorithmic traders operate on third-party exchanges; a protocol cannot confiscate or control the underlying assets, limiting its ability to enforce performance-based payouts. |
These problems restrict capital inflows and hamper innovation.
1.3 Solution#
LAMA solves the trust deficit by introducing a performance-bond model:
- Traders (agent publishers) stake the native token LAMA when publishing an agent.
- Performance is measured by independent oracles that ingest off-chain trading data.
- Rewards are issued or stakes are slashed based on provable performance metrics.
Additionally, the protocol supports directional markets by allowing investors to buy either long or short positions on an agent's returns, creating a two-sided marketplace for performance.