Portfolio Management Infrastructure for an Open Banking World
OpenBankingProject.ch was founded in February 2019 as a cross-organizational consortium to promote the development of open banking in Switzerland. OBP stands for opening up banking in the interest of the customer and today comprises 5 partners and 15 members. Within the context of open banking, the initiative focuses on enabling the use of standards (e.g. APIs, tokens), increasing the visibility of relevant stakeholders, preparing and sharing knowledge, and connecting companies.
At its quarterly member events, OBP regularly offers startup founders the opportunity to present their innovations in the financial industry. The community is particularly interested in the role open banking plays in the business model or core idea of a FinTech. At the member event on 12 November 2025 at Ergon Informatik in Zurich, FinTech founder Rayan Ayari from RAIIS was a guest speaker. RAIIS is a plug-in portfolio management system that helps financial institutions modernize and automate processes, reduce operational redundancies, and improve the accuracy of analytics. Below, Stefan Knaus (OpenBankingProject.ch) and Rayan Ayari (RAIIS) summarize the key messages from the impulse presentation.
The Infrastructure Challenge in Swiss Wealth Management
The Swiss wealth management industry manages CHF 3.4 trillion in assets under management, positioning the country as the third-largest hub in Europe, with 31 percent of managed assets belonging to foreign clients. The chart above shows the development of AuM over the past eight years. In this context, CIS stands for “Collective Investment Schemes.” Despite this impressive scale, Swiss banks face an efficiency problem. Each year, a gap of CHF 2 to 3 billion emerges. The reason is that the cost-income ratio of Swiss banks, at 68.8 percent, is significantly higher than the 60 percent achieved by the most efficient international market participants. This inefficiency is not a technological problem in the traditional sense. It is an infrastructure problem, one that open banking is uniquely positioned to address, albeit only partially.
Internationally, open banking initiatives have made significant progress in standardizing data. In the future, end customers will be able to view, in a standardized way within a third-party application, which assets they hold, which transactions have taken place, which custody relationships exist, and how balances are structured across institutions. In Europe, PSD2 has demonstrated the transformative potential of mandatory API standardization in the context of payments. European banks have implemented APIs that allow secure and standardized access for third-party providers, enabling them to offer account information and payment initiation services to end customers within their applications. In Switzerland, the implementation of open banking has so far followed a market-driven approach. With the Open Wealth Association, a promising initiative has already emerged that goes beyond payments and provides APIs for custody services and order placement in the investment domain.
However, data standardization represents only the first layer of a two-layer problem.
Layer 1 (Data Standardization):
Data standardization refers to the standardization of assets, transaction data, balances, and custody relationships, an area increasingly addressed by open banking APIs and regulatory frameworks.
Layer 2 (Calculation Standardization):
Calculation standardization refers to the harmonization of calculation logic for performance, costs, constraints, and compliance-related impact factors, an area for which no standardized industry-wide approach currently exists.
The consequences of this missing second layer of calculation standardization are clearly visible in practice. Different systems produce different results based on the same underlying data, depending on the methodology applied and the assumptions embedded within it.
Bridging the Gap Between Theory and Practice in Finance
Portfolio models, ranging from classical Markowitz optimization to factor-based approaches and modern machine learning methods, share a common structural limitation: the practical implementability of their results is systematically neglected.
This fragmentation reflects a deeper structural gap between financial theory and practical implementation that has existed for decades. Modern portfolio theory, beginning with Markowitz (1952) and extended by the Capital Asset Pricing Model (Sharpe, 1964) and multifactor approaches, provides mathematically elegant frameworks for portfolio construction. However, these models assume frictionless markets, costless rebalancing, and instantaneous execution, assumptions that fundamentally decouple theoretical optima from strategies that can actually be implemented.
DeMiguel et al. (2009) showed that in practice it remains difficult to translate theoretical advances into consistent future excess returns. Naive equal-weighting often achieves similar or even better performance than optimized portfolios. Estimation errors in expected returns and covariance matrices, combined with the cumulative impact of transaction costs, create a persistent gap between theoretical prescriptions and practically achievable results.
More recent computational approaches, such as machine learning for signal generation or optimization algorithms for allocation, have increased in complexity but often perpetuate the same structural limitation. Implementation constraints like transaction costs or capital availability are treated as downstream adjustments rather than being integrated into the modeling process from the outset. Reinforcement learning approaches offer a potential way forward, as they model such constraints as an integral part of the decision-making process rather than as ex post corrections. However, their practical application in portfolio management remains limited to date.
This leads to three recurring limitations in both traditional and computational portfolio approaches. First, forecasting and allocation are usually optimized separately, meaning statistically good predictions do not necessarily lead to better financial outcomes. Second, many models rely on rigid decision rules that inadequately account for temporal dynamics, interactions, and the broader market context. Third, key constraints such as transaction costs, liquidity, or capital requirements are often considered only after the fact, as corrections to model output rather than as integral components of the optimization. These weaknesses are not inherent laws of quantitative models but rather the result of missing infrastructure, particularly the lack of standardized mechanisms to consistently model costs, constraints, and capital flows from the start.
For the second layer, calculation standardization, an infrastructure is needed that acts as a flexible plug-in layer on top of data standardization, accepts any data source, and directly integrates real-world constraints. Such an infrastructure must function without data migration and be able to process holdings, prices, and transactions from arbitrary sources. It should reconstruct the entire trading history, including capital flows and cash positions, allocate costs transparently and causally, check regulatory requirements in real time, and track capital movements with sufficient precision to enable GIPS-compliant reporting. A key operational principle applies: the new layer complements existing systems; it does not replace them.
Why Artificial Intelligence Needs Both Layers First
he current enthusiasm for artificial intelligence in wealth management faces an uncomfortable truth: AI cannot optimize what is not measured consistently. The promise of machine learning in portfolio management, adaptive allocation, automated rebalancing, personalized strategy construction, depends on training data and evaluation metrics that are consistent, comparable, and precise. If calculation methodologies differ across systems, apparent alpha generation may reflect measurement differences rather than genuine outperformance. The prerequisites for effective AI in portfolio management are clear: clean and structured data (Layer 1), reproducible calculations (Layer 2), and domain expertise to interpret results. The order matters: infrastructure first, intelligence second.
Conclusion and Outlook for the Swiss Financial Center
Switzerland’s leadership in wealth management has traditionally been based on the quality of client relationships, regulatory stability, and geographic neutrality. These advantages remain, but they are increasingly complemented by operational efficiency as a competitive factor. The data point to a structural challenge: 71 percent of wealth managers prioritize asset growth, yet profit margins have remained unchanged despite an 11 percent increase in assets under management in 2024. Cost margins have risen from 30 to 34 basis points.¹ The industry is growing in terms of assets but simultaneously under pressure on profitability. Standardized infrastructure could open a path to transforming Switzerland’s competitive position from a relationship-based advantage to an infrastructure-based one, creating operational capabilities that compound over time and are difficult to replicate.
The efficiency opportunity in Swiss wealth management is substantial and structural. Open banking enables data standardization, making multi-custody operating models more economically viable. The standardization of calculations extends this foundation and delivers comparable, auditable, and constraint-sensitive portfolio analytics.
The infrastructure layer is currently being defined and built. For an industry managing CHF 3.4 trillion in assets, the development of standardized calculation infrastructure deserves particular attention from institutions seeking operational improvements and clear competitive differentiation.
Stefan Knaus is a consultant at Business Engineering Institute St. Gallen AG.
Rayan Ayari is Co-Founder & CEO of RAIIS GmbH.
Contact: info@raiis.ch | www.raiis.ch
References
¹ Swiss Asset Management Study 2025, zeb consulting and Asset Management Association Switzerland (AMAS).
² Markowitz, H. (1952). Portfolio Selection. The Journal of Finance, 7(1), 77–91.
³ Sharpe, W. F. (1964). Capital asset prices: A theory of market equilibrium under conditions of risk. The Journal of Finance, 19(3), 425–442.
⁴ DeMiguel, V., Garlappi, L., & Uppal, R. (2009). Optimal versus naive diversification: How inefficient is the 1/N portfolio strategy? The Review of Financial Studies, 22(5), 1915–1953.
⁵ Based on RAIIS pilot deployments including Aargauische Pensionskasse (shadow rebalancing), Swiss Bank (ESG integration), and Enzler Vermögensberatung (financial ratios platform).
⁶ Berg, F., Kölbel, J., & Rigobon, R. (2022). Aggregate confusion: The divergence of ESG ratings. Review of Finance, 26(6), 1315–1344.
⁷ Alkan, D., Paraschiv, F., & Ayari, R. (2025). Green Fees: Sustainability Impacts on Portfolio Management. International Review of Financial Analysis (accepted). DOI: 10.1016/j.irfa.2025.104812.
8 Open Wealth Association. (2025). Open Wealth APIs. https://openwealth.ch/
- Portfolio Management Infrastructure for an Open Banking World - 06.01.2026
- A current look at Open Finance in Switzerland - 09.09.2025
- Potentials and challenges of the Swiss E-ID (2/2) - 22.04.2025