Vicuna Patrimorio processes high-volume market data through predictive models and publishes the resulting signals alongside their execution logs. Every output is timestamped and open for review, not just promised.
Dashboard preview: rolling accuracy charts, latency distribution, and a public audit trail of model outputs versus realized market outcomes — updated continuously.
Every signal generated by Vicuna Patrimorio is recorded with a timestamp, the input conditions, and the resulting outcome. Nothing in the log is edited after publication.
Illustrative preview of the log format. Full feed with symbol-level detail is available inside the dashboard.
Vicuna Patrimorio ingests high-frequency price, volume, and order-book data, then applies statistical and machine-learning models trained to isolate alpha from noise. The system weighs signal-to-noise ratio continuously, discarding inputs that do not improve predictive accuracy across the validation set.
Volatility clusters are detected before they fully form, using rolling variance and correlation shifts across correlated instruments. Position sizing and exposure recommendations adjust automatically as cluster risk changes, with the stated goal of reducing drawdown relative to an unmanaged baseline.
The processing layer is built to handle increasing data volume without added latency per request. Model inference runs on a stateless architecture, which allows throughput to scale horizontally as additional instruments or data feeds are connected.
Designed to sit inside an existing trading stack, not replace it.
Connect exchange feeds, broker APIs, or custom data sources through a single ingestion endpoint. Formats are normalized automatically before entering the model pipeline.
Incoming data is filtered for volatility clusters and signal relevance, then scored against the current model version before any output is generated.
Scored signals are returned as structured decision support — ready to feed into an execution system or a manual review queue.
A methodology document describing model architecture, feature selection, and validation procedure is available on request for teams evaluating Vicuna Patrimorio at an institutional level.
Data processing occurs within infrastructure compliant with EU data protection requirements. No personal trading data is shared with third parties, and processing records are maintained in line with GDPR obligations relevant to the German market.
Models are validated using walk-forward testing with strict out-of-sample separation. Each model version is benchmarked against its predecessor before being promoted into the live signal pipeline.
Review the live log, the model documentation, and current signal accuracy before committing to a workflow change.