Cryptocurrency portfolios are highly dynamic. A portfolio that looks relatively safe today can become significantly riskier tomorrow as volatility changes, assets become more correlated, or market conditions shift.
We built Wallerina to address this problem through a combination of quantitative finance, AI, blockchain data, and cloud infrastructure.
Wallerina is a quantitative crypto portfolio risk advisory platform that analyzes a user's wallet, current market conditions, and investment objective to calculate an adaptive stablecoin allocation. Rather than asking an AI model to make a financial decision, we designed the system around a different principle:
The mathematics decides. AI explains.
From Wallet Data to Risk Analysis
The workflow begins with a public blockchain wallet address.
Wallerina retrieves the user's portfolio and identifies the assets held in the wallet. Asset classification is performed using contract addresses and chain information rather than simply trusting token names or symbols. Unknown assets are treated conservatively.
The portfolio is then combined with market information and external signals to build a complete snapshot of the user's current exposure.
From this snapshot, our quantitative engine calculates several risk metrics, including:
- Conditional Value at Risk (CVaR)
- Individual asset risk contribution
This gives us a mathematical representation of the portfolio's current risk rather than relying on a subjective assessment.
Monte Carlo Simulation
Historical statistics alone are not enough to understand how a portfolio could behave in the future.
Wallerina therefore uses Monte Carlo simulation to generate thousands of possible future portfolio scenarios.
Instead of predicting a single future price, we model a distribution of possible outcomes. The simulation uses correlated shocks between assets so that changes in relationships between assets are reflected in the portfolio-level analysis.
We also avoid blindly extrapolating historical price appreciation by using a zero-drift assumption by default.
The goal isn't to claim that we can predict the future.
The goal is to understand the range of possible outcomes and the uncertainty surrounding the portfolio.
Turning Risk Into an Allocation
The next stage is the allocation engine.
This is a deliberately deterministic component of the system. It is the module responsible for calculating the target stablecoin allocation.
The allocation starts with the user's investment objective and then applies bounded adjustments based on factors such as:
Each adjustment is recorded as an explicit allocation driver.
This makes the final recommendation auditable. Instead of simply saying:
“Your portfolio should hold 40% stablecoins.”
Wallerina can explain which factors caused the allocation to move toward or away from that level.
A tolerance band is also used so that small changes in market conditions do not cause unnecessary recommendation changes.
Using Market-Implied Information
Wallerina also incorporates prediction-market information.
Binary prediction-market prices can be transformed into market-implied probability distributions around certain events or price levels. These signals provide an additional perspective on uncertainty beyond historical market data.
Importantly, these signals are treated as inputs to the quantitative system rather than as guaranteed predictions.
Where AI Fits In
AI is used after the quantitative
Loading article