Firme Guardanza combines AI-assisted risk management with a dynamic stop-loss system that adjusts its thresholds based on market liquidity and volatility in real-time, rather than relying on fixed rules.
| 24h implied volatility | Continue |
| Order Book Depth | Active |
| Stop-loss threshold | Dynamic adjustment |
| Correlation between assets | Recalculated |
The Firme Guardanza engine receives streams of high-frequency market data and normalizes them before applying trained predictive models to recognize patterns of incipient volatility. The goal is not to predict price movements with certainty, but to identify risk conditions before they peak.
Processing is designed to operate in low latency windows, so that changes in liquidity or order book structure are reflected in recommendations without significant delays.
| Dimension | Focus |
|---|---|
| Sampling Rate | Continue |
| Analysis horizon | Short and medium term |
| Model validation | Periodic backtesting |
Unlike a static stop-loss, set as a constant percentage, the Firme Guardanza system recalculates its thresholds based on available liquidity and a real-time volatility index. This allows you to react differently to a calm market than to one with low book depth, where a forced sale would have a greater impact.
| Criterion | Static | Dynamic (Firme Guardanza) |
|---|---|---|
| Market liquidity adjustment | No | Yes |
| Response to volatility changes | Limited | Continue |
| Temporary threshold review | Fixed | Recalculated |
| Exposure in shallow events | Undifferentiated | Tight |
The following situations describe how the platform is used in practice, based on continuous processing of market data.
The system analyzes the correlation between the assets in a portfolio and indicates deviations from the target weights defined by the investor. Rebalancing proposals are presented as recommendations, not as automatic executions, so the final decision remains in the hands of the manager.
When multiple assets in a portfolio show increasing correlation, the system raises the aggregate risk score and adjusts protection thresholds in a coordinated manner, rather than treating each position in isolation.
Market depth metrics are combined with sentiment indicators derived from volume and activity in order books, in order to contextualize price movements that are not accompanied by sufficient liquidity to sustain them.
The process is divided into three clearly differentiated phases, each one documented and verifiable.
Market data, order books and liquidity metrics are captured continuously from connected sources, without prior transformation, to preserve their traceability.
Each model fit is back-tested over historical periods before going into production to ensure consistency across different volatility regimes.
The system presents signals and threshold settings as proposals. Final application of any recommendation requires human confirmation.
The system is designed to support decision making, not as a substitute for professional judgment. The person responsible for the portfolio retains authority over final execution at all times.
No. Integration with exchanges is done through API keys configured with read-only permissions and, when applicable, operational permissions, but never withdrawal permissions. The funds remain in the user's account at the corresponding exchange at all times.
The system requests the minimum level necessary to read positions, transaction history and market data. The keys are stored encrypted and can be revoked at any time from the dashboard of the exchange used.
Predictive models are validated against historical periods of high volatility, but no system can guarantee identical behavior in the face of an unprecedented event. Under conditions of extreme volatility, the system prioritizes exposure reduction and requires explicit confirmation before making large adjustments.
Access is organized by tiers depending on the volume of data processed and the number of connected wallets. Specific details are provided during the initial consultation as they depend on each client's operational scope.
Aggregated market data can be used to improve predictive models on an anonymous basis. A client's specific portfolio data is not shared with third parties or used outside the context of their own account.
Request an initial consultation to review how the active protection system would be configured on your current portfolio. A member of the team will analyze your case before proposing any integration.