Documentation

We are a team of battle-tested quantitative analysists and who have deployed many successful strategies that currently operate or have operated in cryptocurrency markets on behalf of private funds. Apart from that we also build AI agentic systems, but massive funds are still cautious in terms of employing agents for adapting defined strategies in real-time. That is the core inspiration for Castle.os - to build automated strategies that change when market conditions change without the need for human oversight and strategy analysis which drastically decreases efficiency. We develop our own strategies, backtest them, and then build agentic systems that can alter strategy parameters every time new data comes in. All development is done in Python with the use of LLama Index for agentic systems, and all agents run on DeepSeek R1.

Incorporating AI agentic systems into portfolio management unlocks a new dimension of operational efficiency and strategic potential. These systems not only democratize access to complex trading strategies but also provide a scalable and customizable framework that evolves with the market. As financial markets become increasingly data-driven and interconnected, AI agents represent the future of strategy management—blending human insight with machine precision to achieve outcomes that were previously unattainable.

One of the most impactful applications of AI agentic systems is in advanced trading strategies, such as pair trading. These systems excel in identifying and capitalizing on market inefficiencies by analyzing the correlation between asset pairs, ensuring that trades are executed at optimal moments. By employing real-time data feeds and predictive modeling, AI agents maintain the balance between assets, minimize risk exposure, and optimize returns. This dynamic capability not only enhances profitability but also reduces the manual overhead traditionally associated with such strategies.

That is why we are launching AI powered vaults and an agentic system for assisting users with all of their trading activities.

Castlenomics & Staking

Targeting the launch of HyperEVM, we want to decentralize strategy making by allowing our users to stake CASTLE tokens. CASTLE stakers will be able to create proposal for new vaults, each with a user-defined strategy. Other CASTLE stakers can then vote on which vault they want to see created or which vaults they dislike. The most positively rated strategy at the end of a repeating three week period will be developed by us along with a unique agentic system built to fit that strategy.

Proposing a vault comes with a fixed cost of 500 CASTLE tokens which are burned upon proposal submission, this creates a deflationary mechanism for CASTLE tokens and also serves as an antimeasure for proposal spamming. To further incentivize stakers, half of the commission we receive from deployed vaults on Hyperliquid will be redistributed in USDC to CASTLE stakers daily.

After the launch of the Agent app, stakers will be able to interact with an agentic system, allowing them to consult strategies or analyze perpetual futures and spot tokens, including supported on-chain tokens.