Ample Croissélis processes your data streams continuously and returns actionable allocation recommendations. Complete configuration of a wallet can be done in less than 60 seconds, with no intermediate manual steps.
Interface overview: risk matrix, volatility vectors and asset allocation view, updated with each market cycle.
The Ample Croissélis engine is based on a three-layer processing structure. Each reduces a distinct type of noise present in raw market data, before passing the result to the next layer.
The model decomposes each asset into a set of vectors (momentum, liquidity, cross-correlation) and projects their short-term trajectory. This decomposition isolates directional signals from the statistical noise specific to high volatility markets.
Each recommendation passes through a control matrix which limits the exposure according to the observed volatility and the correlation between positions. The objective is not to eliminate risk, but to guarantee the integrity of the portfolio in the face of an isolated market shock.
Flow ingestion and recommendation recalculation run on a dedicated infrastructure, without a batch queue. Allocation adjustments reflect millisecond-level market conditions for microstructure-sensitive profiles.
The deployment of a wallet follows a fixed and transparent sequence. No step is hidden, each phase corresponds to an identifiable technical operation.
Encrypted connection to your broker or data source via API. No identifying data is stored in plain text.
The engine applies the predictive models to your asset universe and generates a target allocation adjusted to your risk tolerance.
Single validation of the proposed allocation. Deployment runs immediately, without additional manual configuration.
Ample Croissélis does not add a decorative layer to financial analysis. The interface exposes exactly the data necessary for a decision: exposure, volatility, correlation, without superfluous indicators.
Each module is versioned and documented. The model parameters are searchable, allowing a technical user to understand the logic behind each recommendation rather than being subjected to it.
Visualization does not replace analysis, it makes it readable in a few seconds.
Performance, drawdown and Sharpe ratio consolidated on the same time scale, without aggregation which would mask a peak in volatility.
Net exposure and cross-correlation between positions, updated with each engine recalculation.
Distribution by asset class and holding horizon, compared to the recommended target allocation.
The model adjusts to the strategy time horizon without changing the underlying infrastructure.
The engine recalculates momentum vectors over windows of a few minutes and flags exploitable microstructure gaps before the spread tightens. The emphasis is on latency of restitution rather than on the depth of the analysis.
Over a horizon of a few days to a few weeks, the model further weights sector correlation and volatility cycles to identify entry points where the risk/return ratio remains favorable beyond the daily noise.
For multi-asset portfolios, the engine produces a breakdown of risk by factor and a sensitivity matrix, which can be used directly in a reporting or allocation committee process.
Direct answers to the most common questions asked by technical users.
The ingestion pipeline processes streams continuously through persistent API connections. Recommendations are recalculated for each flow update received, without deferred batch processing. The observed latency depends mainly on the upstream data provider, not on the Ample Croissélis engine.
Connection is via limited-scope API key, encrypted in transit and at rest. No login credentials are requested or stored. The infrastructure is hosted on a cloud environment isolated by client, with logging of each access.
The model is trained on historical price, volume and implied volatility series, supplemented by aggregated order flow data. The parameters are re-evaluated periodically to limit the risk of overlearning on a past market regime.
Portfolio deployment in less than 60 seconds after data feed connection. No card required to view dashboard demo.