Trusted AI Framework for Wealth & Investments

Widespread adoption of any technology is a complex outcome, driven by many variables.

Our experience with customers and feedback from regulators (Bridgeweave is one of the 40 firms that have been accepted by for feedback from the FCA Advice Unit) has led us to design a 6 point framework for Wealth and Investments industry.

We road test every product using this framework before releasing it for widespread use by our customers.

TAIF in Action

We have developed a robust framework for Trusted AI using carefully chosen set of components to create coherent, automated templates that are suited for Wealth and investments.


We go to great lengths to make sure that all our products comply with our own high standards.

An illustrative example for our Portfolio Insights Product that delivers signals to an investment advisor, is shown below.

Accuracy

+

Question: How accurate are the results?

Explanation: Machine learning algorithms to create the most optimal portfolio for the user.
Performance of the optimised portfolio is tested against performance of a dynamic benchmark index.

Output: No of times the algo has beaten the benchmark

Reasoning

+

Question: Why does the algo think that there might be a problem?

Explanation: The entire portfolio is compare with a dynamic benchmark on different criteria such as performance, concentration, volatility, Sharpe ratio etc. and concerns are highlighted.

Output: Two funds in the portfolio (23% by value) have not performed well. They are dragging down the overall portfolio score for the portfolio.

Methodology

+

Question: What is the methodology followed by the algo?

Explanation:

1.  InvestAI selects two lowest ranked funds for optimisation.

2.  InvestAI reviews hundreds of possible combinations and runs optimisation models for shortlisted fund that match the portfolio

Output: The best fit fund is suggested to the user.

Transparency

+

Question: How does the suggested action compare with current position?

Explanation: Impact of the recommendation on the portfolio is analysed and presented to the user in an easy to understand way.


Output:

Explainability

+

Question: Can the suggestion be explained clearly?

Explanation: Multifactor scores are calculated for each portfolio details of each factor are provided with each suggestion.

Output:

Audit Trails

+

Question: Are audit trails available for the action suggested?

Explanation: Each fund is screened against the knowledge fabric so as to avoid fund concentration, AMC concentration and adherence to allocation guidelines that match individual investor risk profiles.

Output: This record is saved and available for future audits.

An illustrative example for our Portfolio Insights Product that delivers signals to an investment advisor, is shown below.

Accuracy

+

Question: How accurate are the results?

Explanation: Machine learning algorithms to create the most optimal portfolio for the user.
Performance of the optimised portfolio is tested against performance of a dynamic benchmark index.

Output: No of times the algo has beaten the benchmark

Reasoning

+

Question: Why does the algo think that there might be a problem?

Explanation: The entire portfolio is compare with a dynamic benchmark on different criteria such as performance, concentration, volatility, Sharpe ratio etc. and concerns are highlighted.

Output: Two funds in the portfolio (23% by value) have not performed well. They are dragging down the overall portfolio score for the portfolio.

Methodology

+

Question: What is the methodology followed by the algo?

Explanation:

1.  InvestAI selects two lowest ranked funds for optimisation.

2.  InvestAI reviews hundreds of possible combinations and runs optimisation models for shortlisted fund that match the portfolio

Output: The best fit fund is suggested to the user.

Transparency

+

Question: How does the suggested action compare with current position?

Explanation: Impact of the recommendation on the portfolio is analysed and presented to the user in an easy to understand way.


Output:

Explainability

+

Question: Can the suggestion be explained clearly?

Explanation: Multifactor scores are calculated for each portfolio details of each factor are provided with each suggestion.

Output:

Audit Trails

+

Question: Are audit trails available for the action suggested?

Explanation: Each fund is screened against the knowledge fabric so as to avoid fund concentration, AMC concentration and adherence to allocation guidelines that match individual investor risk profiles.

Output: This record is saved and available for future audits.