F.-K. Weber continuously evaluates market, financial and company data and provides recommendations that have been tested in advance on historical periods. Every issue is clearly documented.
Schematic representation of the expense ratio, not a measurement of a specific customer.
The platform combines real-time processing with models that are validated before productive use.
Market, financial and company data is processed continuously. Forecasts update automatically, without manual intervention.
Models are trained on historical time series and output probability bands rather than individual point values.
Portfolios and data sets are grouped by volatility and correlation to reveal concentration risks.
Processing scales from individual portfolios to enterprise-wide data sets without changing the analysis method.
Existing data sources can be connected via defined interfaces without replacing existing systems.
Each recommendation is logged with the underlying data points and the calculation time.
F.-K. Weber develops analysis models that are based on documented data sources and openly visible calculation rules. Each recommendation can be traced back to the underlying data point.
The focus is on reproducibility: the same input data leads to the same results. Deviations between the forecast and the actual course are recorded and incorporated into the next model version.
No model goes live without first being tested against historical market phases.
Data collection: Market, financial and company data is recorded and cleaned in a structured manner.
Model training: Algorithms are trained on historical time periods and different market phases.
Backtesting: Strategies are tested against past market conditions before being used productively.
Ongoing control: Deviations between forecast and result are documented and incorporated into the next version.
All models are tested on data sets from several market cycles before productive use. Deviations between the forecast and the actual course are documented and can be viewed upon request. Past test results are no guarantee of future developments.
The weighting of the analysis adapts to available funds, time horizon and risk appetite.
Monthly surpluses are distributed across different asset classes depending on the risk profile. The analysis takes into account existing reserves and the desired time horizon before suggesting an allocation.
Allocation proposal with ranges for conservative, balanced and offensive profiles, including historical fluctuation range of the underlying scenarios.
Example representation of the model output, no investment advice and no guarantee of results.
Fluctuating income is balanced against current obligations. The models show what proportion of income can be invested without liquidity risk.
Recommendation on the amount of reserves based on income fluctuations in recent periods as well as a suggestion for diversifying the remaining amount that can be invested.
Exemplary representation of the model output, no investment advice and no guarantee of results.
Existing positions are checked for correlation and concentration risk. The analysis shows which asset classes reduce the fluctuation range of the overall portfolio.
Correlation matrix of current positions as well as suggestions for rebalancing with expected changes in portfolio volatility.
Exemplary representation of the model output, no investment advice and no guarantee of results.
The platform processes structured market data, key company figures and third-party sources connected via interfaces. Every source is checked for completeness and topicality before use.
Data processing runs at an interval of 15 minutes. Model versions themselves are checked at regular cycles and retrained if necessary.
No. The interface is designed for use without prior technical knowledge. A documented interface is available for connecting your own data sources.
Yes. The platform is designed for browsers on desktop and mobile devices. A native app is currently not planned.
Data is transmitted encrypted and processed on servers located in Germany. Accesses are logged and can be traced.
Deviations between the forecast and the actual course are recorded, documented and incorporated into the revision of the affected model version.
Access includes an initial analysis of your data and an overview of the underlying backtest results. No automatic contractual commitment.