How Theseus Works

Theseus leverages the Archimedes AI Engine to continuously analyse and optimise asset allocation considering liquidity, risk, yield potential, price momentum, and market sentiment.

Archimedes AI Engine

Theseus V2 is a sophisticated trading model deployed on Mozaic’s Archimedes AI Engine. The Theseus model uses a pseudo event-driven framework to optimise asset allocation within GMX’s GM pools. The AI continuously calculates optimal weights for each asset and determines when to rebalance the portfolio.

Asset selection (Tradeable Universe)

Archimedes carefully chooses which GM pools to invest in and considers several factors to make the best choices for the vault:

• Liquidity: The vault only allocates to pools where liquidity exceeds agreed minimum risk management thresholds. This ensures sufficient market depth to enable efficient trading into and out of positions.

• Risk assessment: Archimedes evaluates each pool's risk profile based on historical volatility, correlation with broader market trends, and other proprietary metrics. This information informs the risk-adjusted weighting of each pool within the portfolio.

• Yield optimisation: The AI analyses the yield potential of each GM pool and considers both the current APY and the projected sustainability of returns. This includes evaluating trading fees, liquidity mining rewards, and other yield-generating mechanisms specific to each pool.

• Price momentum: Archimedes analyses price trends to identify GM pools with assets showing favourable momentum. This allows the vault to capture potential capital appreciation in addition to yield.

• Market sentiment: The AI also considers broader crypto market sentiment indicators to adjust the portfolio's overall risk exposure. This allows for dynamic changes to vault allocation in response to changing market conditions.

Archimedes synthesises these factors using advanced machine-learning algorithms, continuously optimising the vault’s allocation strategy. This approach enables the AI to dynamically weight the above factors, adapting to evolving market dynamics and maintaining an optimal risk-reward profile for the vault.

Rebalancing strategy

The Archimedes AI Engine continuously monitors market conditions and portfolio performance to determine optimal rebalancing times. Rebalancing may be triggered by various factors, including:

• Significant market movements;

• Changes in relative asset performance;

• Shifts in yield opportunities across GM pools;

• Risk management thresholds being met or exceeded.

The rebalancing process is iterative, and Archimedes regularly reassesses and adjusts the portfolio composition. This approach balances the need for strategic adjustments against the cost of excessive transactions, ensuring the vault’s strategy remains efficient in changing market conditions.