The Sfromiledra engine processes market data in real time and adjusts risk levels before volatility translates into avoidable losses. Operations are based on trained models, not on decisions made under pressure.
Traditional stop-loss systems are activated when the price crosses a fixed level, without distinguishing between a temporary liquidity correction and a real trend change. This causes premature closing of correct positions and unnecessary drawdowns.
The Sfromiledra engine identifies clusters of volatility before they translate into a sharp price jump. Analyzes order dispersion, market depth, and execution speed to distinguish signal noise, and shifts the protection threshold accordingly.
Each component processes a high volume of market data and returns recommendations limited to the risk profile defined by the user, without replacing their final criteria.
Models trained on historical price, volume and liquidity that estimate the probability of short-term volatility scenarios. The outputs are expressed as confidence ranges, not certainties, so that the final decision remains the operator's.
Continuous processing of market flows to signal relevant changes in microstructure before they are reflected in the closing price of the candle. The latency between data and signal is kept minimal by pipeline design.
Exposure and protection recommendations that are recalculated as market conditions change, scalable from an individual account to multi-asset institutional portfolios.
Sfromiledra was born from the observation that a large part of avoidable losses in trading do not come from poor analysis, but from decisions made under emotional pressure at the wrong time. The team's goal is to replace that reaction with a measurable and repeatable process.
The platform is built on a simple principle: AI should inform the decision and protect capital, but should not operate opaquely. Each recommendation is accompanied by the variable that originated it, so that the user can audit it.
The transparency of the process is what sustains trust in the system. The pipeline is described below without marketing simplifications.
Market feeds are normalized and synchronized with minimal latency, including price, volume and order book depth when permitted by the broker.
Neural networks trained on time series identify recurring volatility structures and compare them to the current state of the market.
The resulting signal is translated into a stop-loss level adjustment or an exposure recommendation, prioritizing drawdown reduction over aggressive return maximization.
The same engine is configured differently depending on the objective: medium-term strategic coverage or active intraday position management.
Risk management teams use the system's recommendations to adjust coverages when the model detects a sustained increase in correlation between assets, a common indicator of market stress. Decisions are documented with measurable results against the internal benchmark.
In high volatility sessions, the system recalculates the protection threshold several times per trade, avoiding closures due to market noise while maintaining a defined exit in the event of a confirmed trend change. The strategic decision is supported by the history of recalibrations.
The platform connects via API to the market feeds and, when the broker allows it, to the order execution layer. Integration is performed in signal read-only mode or in automatic level adjustment mode, depending on the configuration chosen by the user.
Access credentials are stored encrypted and are used exclusively for the authorized connection to the broker or data provider. The processed market data is not shared with third parties unrelated to the provision of the service.
The system incorporates action limits: when price dispersion exceeds the ranges observed in model training, it reduces the confidence of the signal and prioritizes capital protection over entry or exit optimization. In no case is execution guaranteed in extreme market conditions, since this also depends on the liquidity available at the broker.
Request access to connect your market data and evaluate the behavior of the smart stop-loss on your own trading, before deciding on a full integration.
Sfromiledra does not offer profitability guarantees. The system is designed to limit the risk of emotional decisions, not to eliminate market risk.