AAAccell developed an algorithm to calculate value and risk profiles of illiquid assets such as real estate funds, by estimating a replication portfolio.
AAAccell can determine a replication portfolio of illiquid assets to perform exact risk calculations.
Our ML algorithm is built on derivatives using thousands of risk factors of equivalent products for price determination, thus similar products with at least monthly available market data or portfolios of several products (replication portfolios). The price determination from the replication portfolio is embedded in a comprehensive reporting that proves that the selected replication portfolio reflects the market price of the underlying asset. Our clients can use the reports for internal purposes to understand exact risk profiles. However, the results also help in the regulatory compliant distribution of funds.
- Illiquid Assets
- Machine Learning
- Risk Reporting
- Fund Issuers
- Fund Managers
- Institutional Investors
- PRIIPs Reporting
- Risk Analytics Fund
- Risk Reporting
AI-Based Real Estate Fund: We offer customised solutions based our research on alternative investments to clients like Edmond de Rothschild
Our algorithm, developed exclusively for Smart Estate, was designed to analyse systematically complex market conditions and evolutions and is based on more than 8,000 economic, urban, social and financial variables, which are then merged into dynamic algorithms.
Background: Problem lies in illiquid assets
Real estate, airplanes, or private equity investments have something in common: They are priced by heuristic methods typically on a yearly frequency. Hence, the value of such assets, their risks and prices are unknown during the year. However, Traditional methods are not able to calculate assets with not at least monthly market prices.
Change – Regulatory demand
Over the last years, there have been several regulatory changes, which affect investment products of illiquid nature. Examples include:
- MaRisk: The Minimum Requirements for Risk Management regulation (effective December 2005), states that positions in a banking book subject to market price risks must be valued at least quarterly. Depending on the type, scope, complexity and risk content of the positions in the banking book, daily, weekly or monthly valuation, determination of results and communication of risks may also be required.
- PRIIPs: The PRIIP Regulation (effective January 2018) requires a key information document for packaged retail and insurance-based investment products. This document demands the calculation of a market risk measure for each product within its scope, as well as performance analysis.
- MiFID II: According to the new regime of the MiFID II Directive (effective January 2018), sellers of financial instruments (issuers) must determine for which customers their financial instrument (definition of the target market). This classification is also based on the exact calculation of the individual financial instrument’s risk.
- FIDLEG: The Financial Services Act (planned effectiveness January 2019) needs a Basic Information Sheet (BIB) is to be issued for all financial instruments offered to private customers. The BIB should enable a well-founded investment decision and a real comparison of different financial instruments in a simple and understandable way.
AAAccell has developed a system based on modern, cutting-edge quantification and AI to solve the low-frequency or illiquid asset valuations in a regulatory compliant way. AAAccell is pooling worldwide data on asset classes, market instruments and a wide variety of economic data. Our methodology was built on derivatives and combinations of tree-based machine learning as well as differential evolution optimization, using hundreds of risk factors. The reporting is generated for price volatility, value-at-risk or expected shortfall.
The RegTech solution consists of replicating a substitute model. The model provides a time series with at least monthly data points and supports the user in the target value as well as the entry of specific modeling parameters. We use advanced Machine Learning algorithms to create the reproduction of a vast data bank we created. Depending on the availability of empirical data, a target value estimation is carried out in advance, on which we simulate additional stress scenarios. The time series obtained can be used in equal measure to meet the regulatory requirements, for example, to comply with PRIIPs.
With our RegTech solution, we have successfully helped dozens of clients across the DACH region. This includes pension funds and asset managers.