Understandable Artificial Intelligence

Overview

EDS was given 2 sample portfolios by a UK pension fund advisor. Since only the asset weights were known, EDS tool deducted investors' preferences and the goals of each fund

Portfolios were optimized using combined holdings and Passive Indexes
The goal in the second portfolio was to increase exposure to U.K. inflation
- We show how EDS software increases that exposure while attempting to keep other metrics stable

Balance Every Risk and Opportunity

The most powerful portion of our tool is the ability to balance every risk and opportunity included at the same time

When optimizing a fund, EDS tool focuses on the following:
- Maximizing units of return for every unit of risk
- Maintaining likely investment preferences
- Using this approach ensures to get the most from each trade-off

Enhanced Insights With A.I.

Many of the details in these portfolios were hidden. Using our tool we back out some initial estimates

Our A.I. was given relatively little about these portfolios
In a real world scenario these insights should be reviewed and validated by human team members
- The purpose of our approach is always to be working in tandem
Duration Decomposition
- Using given data we can back-out the asset duration (rate/inflation)
- Using our A.I. we can extend that to the LDI portfolio and return seeking portfolio
- We estimate ~40-42 year interest rate duration for LDI

Extended Insights With A.I.

Our tool compares the relationships between factors and trade-offs

In these examples, this provides some powerful benefits
In many cases duration targets are set by mandate
- We look beyond that to illustrate other possible insights
Groups and types
- Interest rate and inflation exposure are important to managing these types of portfolios
- Our A.I. can look at factors in isolation but more importantly together
- This allows us to look at groups like global interest rates or inflation and almost any other

Client 1 Solution Funds

Funds were selected between the 2 client funds and passive indexes

*Data from EDS, Redington, & Bloomberg, courtesy of London Business School

Portfolio Metrics

Risk Exposures

EDS solution maintains similar exposure to risks
Any of these can be adjusted depending on actual client preferences

Current Portfolio EDS Solution

Standard Deviation Decomposition

Current Portfolio EDS Solution

*Data from EDS, Redington, & Bloomberg, courtesy of London Business School

Factor Risk Decomposition

Current Portfolio EDS Solution

Portfolio Optimization

Client 2

Opportunistic Range Seeking

Given the approach of our tool, it is possible to extract existing implicit preferences from a portfolio. We can then build on these and opportunistically seek other objectives or avoid risks

In this example we focus on increasing exposure to U.K. inflation
Using our estimate of implied preferences we then scale up this objective
- We can see the A.I. opportunisically finding the best available trade-offs at each point.
These spectrums are often not smooth
- This highlights the power of our tool as it can find these 'non-convex' or 'hidden' solutions

Portfolio Metrics

Conclusion

We seek to add value to your process and team with an innovative and powerful suite of customized tools

Let your human team focus on more valueable work than data cleaning and search
- Machine Learning can point them to new insights more efficiently
Combining models allows for a more complete view of risk
- A deeper understanding of the connection between models
Keeping better balance
- No human can keep track of everything, we believe A.I. can supplement the process and provide more informed decision making an

Disclaimers

Please remember that past performance may not be indicative of future results. Different types of investments involve varying degrees of risk, and there can be no assurance that the future performance of any specific investment, investment strategy, or product made reference to directly or indirectly in this presentation, will be profitable, equal any corresponding indicated historical performance level(s), or be suitable for an individual's portfolio.

Our projections are based on current market conditions which can vary over the coming months and weeks. Additionally, our projections are based on historical market behavior which may vary unexpectedly. Using Machine Learning, our tool should adjust to new market fluctuations but we might not be able to avoid short term volatility.

Get In Touch

info@EconomicDataSciences.com

www.EconomicDataSciences.com

info@EconomicDataSciences.com

EconomicDataSciences.ai