Introduction
Centering investment on the investor through goal-oriented strategies
Many investors struggle to align their investment strategies with unique life goals. Traditional approaches often prioritize strong returns or market benchmarks, without addressing specific objectives like retirement, homeownership, or education.
Goal-oriented strategies shift the focus to personal financial goals, considering both market risk and the risk of falling short of these goals. Portfolios are dynamically adjusted to reflect changing priorities. To implement these strategies effectively, dynamic programming and AI algorithms are used to create tailored investment pathways that adapt to evolving conditions.
Challenge
Reshaping portfolio generation with reinforcement learning
Our client, an international global asset management company, faced difficulties in helping their investors build portfolios specifically tailored to individual life goals. Their existing wealth management strategies were inadequate in addressing the diverse and evolving needs of their investors.
Advice standardization for financial advisors
Financial advisors (FAs) may often recommend self-crafted portfolios which may not follow the best practices. This creates the need for a robo-advisor assisted advice to address such issues and bring standardization across the organization.
Clarifying financial goals and identifying the right investment path
The company struggled to help clients articulate and prioritize clear financial objectives, making it challenging to guide them in selecting the optimal investment path. Balancing complex factors such as risk, returns, and personal goals required a more dynamic, personalized approach.
Ensuring robust evaluation and measurement
A significant challenge was developing a reliable evaluation framework to ensure the effectiveness and trustworthiness of proposed strategies. This required deep-diving analyses, back-testing, and performance comparisons to meet investor expectations for accuracy and robustness.
Monitoring progress toward goals
Traditional wealth management approaches focused on portfolio performance rather than tracking progress against specific life goals (e.g., retirement or homeownership). This left investors uncertain about whether they were on track, highlighting the need for a solution that could track progress toward these personal goals.
Delivering actionable insights to multiple stakeholders
The solution needed to serve both external clients (via an intuitive interface) and internal stakeholders, each with distinct needs. Balancing usability with technical complexity presented an additional challenge.
By using dynamic programming and machine learning, we’ve created a system that not only optimizes wealth but also aligns investments with what truly matters to clients.
Solution
Building a goals optimization engine (GOE) that delivered
The Goals Optimization Engine (GOE) is an AI-powered framework designed to optimize wealth creation by dynamically adjusting portfolios based on investor goals, risk tolerance, and market conditions. Beyond fulfilling the client’s core need for a GOE modeling framework, we also addressed key objectives, including enhancing portfolio flexibility, improving goal tracking, and providing actionable insights for both external clients and internal stakeholders.
GOE modeling
We developed a reinforcement learning model that generates optimal investment decisions by factoring in portfolio options, goal achievement probabilities, and the likelihood of exceeding loss thresholds. This process followed a three-pronged approach:
Recommendation engine
We extended the algorithm to offer personalized recommendations, including:
● Optimize initial wealth: Provide starting investment recommendations to optimize wealth and detect over/underfunded goals
● Adjust cashflow: Advise changes in planned investment cash flow as needed
● Suggest tenure adjustments: Recommend changes to the tenure and maturation horizon to ensure goals are achievable
● Model “what-if” scenarios: Simulate market events like crashes and recovery paths to aid decision-making.
Expanding opportunities for innovation
Beyond addressing the client’s immediate needs, we identified further opportunities to enhance investor confidence and expand the offering through:
● Scenario modeling for robust what-if analyses
● Enhanced recommendation engine features for further personalization
● Real-time adjustment capabilities, allowing investors to adapt seamlessly to changing circumstances
Adjustment features
We also incorporated responsible adjustment features, such as:
● Swing constraints: Prevent excessive shifts in portfolio weights and improve tax handling
● Fees inclusion: Account for the annual impact of entry/exit load fees and brokerage fees
● Quarterly rebalancing: Ensure agile responses to market uncertainties and maintain focused portfolios
Outcome
Driving a new investor approach
By developing the client’s GOE model, we delivered a stable, efficient product that not only addressed the core functionality but also supported other investor-centric goals.
Immediate outcomes
The algorithm demonstrated that GOE significantly increases the probability of achieving targeted goals across various scenarios. We also found that the GOE model outperformed benchmark portfolios. The decision support system showcased two clear use cases: portfolio return and risk computation, as well as portfolio generation. Additionally, the client received considerable positive press and saw a significant increase in client uptake of the GOE model.
Looking forward
The client is eager to expand the functionality of the recommendation system to account for tax implications and support multiple timeframes for diverse goals, further enhancing the product offering for investors.