AI for Treasury/ALM. Finally.

We deliver AI-powered optimization models and software for Treasury and ALM, supporting financial institutions in meeting their strategic goals amid complex markets and regulatory requirements.

Contact UsLearn more

Our Solution

DeepBalance

Dynamic asset/liability allocation optimising performance while respecting institutional, market, and regulatory limits.

Environment Simulation

Macro Scenarios
Market Scenarios

Balance Sheet Simulation

Cash flows Forecast
Analytics/Metrics
Customer Behaviour

Balance Sheet Optimization

Deep Learning AI Engine
Targets and Constraints
Optimal multi-step Rebalancing Strategies

Balance Sheet Management

Financial Planning
Best Products
Lending Rates
Smart Treasury/ALM:
Empower your Treasury and ALM teams with our AI engine, built for real-world balance sheets. Set objectives, define constraints, and get actionable multi-step recommendations that respond to market shifts, customer behaviour, and regulatory rules.
Always-on, Data-Driven Support:
DeepBalance works as an artificial Treasury/ALM assistant, trained on thousands of simulation scenarios, able to generalise and respond effectively to unexpected events and evolving market conditions.
Leverage Complexity:
Financial institutions face inherently complex balance sheets, which often require simplified models and decisions. Our solution uses deep learning to uncover hidden dependencies across the balance sheet and its market drivers, supporting decisions that reflect real-world complexity.

How It Works

DeepBalance in Action

To our knowledge, DeepBalance is the first practical application of deep learning for balance sheet optimisation in Treasury and ALM. Our models, trained on thousands of scenarios, navigate complex regulatory and institutional constraints while bypassing limitations of traditional dynamic programming methods.

Financial professionals can rapidly analyse, test, and compute detailed optimisation strategies, while our self-learning algorithms automatically adapt to your institution’s requirements, determining optimal asset, liability, and capital allocations for multiple future scenarios while maintaining regulatory compliance.

Step 1

Scenario Simulation

Generates thousands of financial scenarios covering interest rates, customer behaviour, and economic indicators.

Step 2

Optimization Model Training

Our deep neural networks train rebalancing of your balance sheet on generated scenarios using backpropagation, optimizing for performance targets across possible futures while enforcing provided regulatory and strategic constraints.

Step 3

Balance Sheet Optimization

Enter the economic forecasts of your institution into our trained neural network model to instantly receive respective optimal rebalancing action plans across multiple time steps

Case Study

Balance Sheet in a negative rates environment.

To illustrate the power of DeepBalance and its operating principle, we use a simple, stylised EUR balance sheet assuming today is 29th of March 2016. At that time, European markets were in a negative interest rate regime, making asset/liability management particularly interesting.


Contact us to get the full case study and test your balance sheet!

Balance Sheet Rebalancing


The DeepBalance model has actively managed all portfolios:

Assets:

  • The liquidity buffer was increased immediately to address the LCR constraint violation (51% < 100%).
  • DeepBalance preferred fixed-rate loans, generating higher income than floaters in the prevailing negative-rates environment.
  • The DeepBalance model’s cumulative 6-month interest income (€73.69 mln) was consistently greater than that of the benchmark strategy (€72.68 mln) by €1.01 mln.

Liabilities:

  • DeepBalance significantly rebalanced liabilities towards current accounts, the cheapest source of funding.
  • The DeepBalance model maintained the cumulative funding cost (€15.90 mln) consistently lower than the benchmark strategy (€38.05 mln), resulting in cumulative funding cost savings of €22.15 mln over the 6-month period.

Annualized Return on Equity (RoE)


Over the six-month period, the benchmark run-off strategy delivered a cumulative RoE of -0.73%, while DeepBalance achieved 6.21%.

Resources

News, references, materials ...

Industry Recognition

The Deep ALM research by Prof. Josef Teichmann and our Scientific Advisor Prof. Thomas Krabichler earned the Swiss Risk Award (2020), undescoring the work’s relevance for financial industry.

Deep Dive into DeepBalance

See the thorough discussion of the methodology behind DeepBalance in this research article co-authored by our scientific advisor Prof. Thomas Krabichler.

Swiss Innovation Grant

We received a financial grant (Innocheque) from the Swiss Innovation Agency (Innosuisse)!

Team

Experts in Finance and AI

Based in 🇨🇭 Switzerland, our Team unites deep expertise in quantitative finance, machine learning, and financial enterprise software to help financial institutions and their Treasury and ALM teams succeed in increasingly complex economic and regulatory environments.

Contact

ALM is exciting!

Connect with our team to get the full case study and explore how DeepBalance can elevate your Treasury/ALM.