Risk Engine

The AI agent acts as a constant risk validator for your portfolio. This is implemented from fundamentals via custom model finetuning and graph optimisation, resulting is a robust and accurate risk management agent.

The agents approach to risk follows the core principles of modern portfolio theory with the agent employing a multi-faceted approach to identify, measure, and mitigate risks in real-time. By leveraging mathematical rigor and data-driven insights, it enables precise risk forecasting and optimal allocation strategies, empowering financial institutions to maintain resilience in volatile markets.

The core functionality relies on traditional measures such as standard deviations and covaraince of assets to measure portfolio variance, looking at granular tick-level data from major CEX's and DEX's via custom connections. These metrics have been optimised via principal component analysis to uncover hidden patterns in multi-asset dependencies, ensuring a comprehensive understanding of systemic risks. The model constructs and measures your portfolios position on an efficient frontier in real time in order to manage the safety of loan repayment.

The models efficient frontier was constructed via the use of synthetics portfolios.

Risk Tips

Different assets have varying degrees of impact to the risk metric. A portfolio of long memecoins would be deemed risky, a split portfolio of memecoins, staking, investments in tokenised assets would be deemed less risky. Offsetting positions also reduce the risk metric, e.g. long vs short perpetuals, as this greatly reduces risk, especially exposure to broad market risk.

The agent is capable of suggesting portfolio changes to reduce your risk metric, opening the door to taking on more leverage via under-collateralised lending.