Leveraging digital technology and innovation in Wealth planning

Abstract:

Personalization is the key to success in Wealth planning. Companies are investing billions of dollars in digital innovation to equip their platforms, giving UHNW clients the best-of-the-breed personal touch. AI/ML is the bedrock of innovation. Starting from product suggestion to portfolio monitoring to macro-economic event prediction (systematic risk), AI/ML plays an unparalleled role. This article illustrates how the use of AI/ML creates a holistic personalization experience

Introduction:

Over the last decade, Wealth management has become an essential need of our lives. Although initially, it started as a primary choice of ultra-high net-worth (UNHW) investors, today, affluent mass-market investors have also gravitated towards it. At a macro level, it is a holistic approach to manage, preserve, & grow wealth to meet an individual’s life goals. Also, passing down to heirs. Owing to tremendous popularity and massive potential, all the Wall Street majors jumped into this trillion USD bandwagon to mint billions of dollars in revenue. A full spectrum wealth management comprises-

• Investment management

§ Conventional asset classes

§ Alternative asset classes

§ Cash (USD & foreign)

• Retirement planning
• Tax planning
• Estate planning

§ Wills

§ Charitable contributions

• Banking

§ Liquidity, e.g., cash-flow management

§ Borrowing, e.g. — credit products

• Hedging/protection
• Insurance
• Annuities

As mentioned earlier, this lucrative growth potential entices wealth managers of different sizes to jump into the wealth management business. In today’s market, we see players of different shapes and sizes — ranging from big banks (e.g., JP Morgan, Wells Fargo, etc.) to different-sized brokerage houses (e.g., Charles Schwab, Vanguard, Raymond James, etc.) to independent RIAs. With the fierce competition in the market, top players create competitive advantages, emphasizing ‘personalization’ along with a sharp focus on the low total cost of ownership (TCO). Digital technology with innovation is the key ingredient in creating personalized client experience with low TCO. This article will review step-by-step how this can be achieved by leveraging technology as a differentiator. There might be an initial investment in building an integrated technology ecosystem, but it has a shorter payoff period and is lucrative.

Wealth management lifecycle:

Whether it’s UHNW client or mass-affluent the primary lifecycle phases are identical. See below-

Wealth management lifecycle

Earlier we referred ‘Personalization’ as a differentiating factor among wealth management players. Let’s examine how sophisticated ‘Personalization’ plays critical role in building competitive advantages. Disclaimer, in this article we’ll illustrate how investment portfolio ‘Personalization’ meets life’s goal. Followed by we’ll share what Wall Street giants are doing using digital & innovation.

The word ‘Personalized’ implies — specific to a client, not one size fits all. It’s a complex multi-dimensional algorithm comprises of — client financial objectives, investment horizon, macro-economic situation, and restriction (personal preference). See examples below –

• Mr. Smith, an unmarried 27 yrs old entrepreneur, owns $250 million IT consulting firm globally, wants to get into the agricultural dairy business over next 10 yrs. Estimated $22MM needed to operate a large scale profitable dairy business. He has cumulative $1.2 MM in retirement account. Own his primary resident $3.4MM outright. He has 2 vacation properties on mortgages in Miami, FL and San Diago, CA, cumulative preset market value $8MM, and mortgage $5.2 MM. He has $9.7MM term life insurance too. It’s worth mentioning, he believes casinos or marijuana business impacts youth in our society

For Mr. Smith, the financial advisor, upon input the customer information (asset, liability, preferences, and restrictions) to the planning system (e.g. eMoney, AssetMark, etc.), it fetches optimal “model portfolio” fit his goal. The asset, liability and goals are summarized by the platform as below:

• Short term planning horizon:10 yrs, long term planning horizon: 45 yrs
• · Target: fund $12 MM end of 10th yr to start dairy business (rest financing)
• Last year 3 yrs annual earnings in the range of $2.2MM from his IT consulting firm. Company has revenue growth rate of 10.2% over last 3 yr, inflation adjusted growth rate 7.8%. Projected 11.7% growth rate for next 3 yrs (ML model Random Forest predicted)
• · $1.2MM current retirement, and annual contribution of $90K to SEP IRA puts him very comfortable position
• Won $3.4MM primary home outright
• Liability includes $5.2MM mortgage
• Restriction: no casinos and marijuana

Using this as input data using ML’s RAG model, system responded back with following model portfolio. Assumed $1.5MM annual investment over next 3 yrs -

• Domestic large cap: 27% of portfolio. Benchmark against S&P, growth rate 18% YoY. Sleeves by 3 sectors
• International large cap: 13% of portfolio. Benchmark against MSCI
• 11% portfolio US Treasury with projected avg return 3.2–4% over next 3 yrs
• 10% of portfolio composition from FX. INR & JPY, as his IT consulting firms operates from India and Japan. It also works as hedging
• 25% of portfolio from private credit. Over next 3 yrs potential growth of 11–12.5%
• 12% Money market with 4.85% return. Over next 3 yrs potential growth of 2.5–3%
• Remaining amount kept as Cash for liquidity and emergencies
• Restriction: No investment in the casino related entertainment or marijuana related securities

Obvious question why this composition?

Reason being the LLM in portfolio building engine taken into consideration — age, targets (e.g. starting dairy operation), and risk profile to suggest ‘aggressive growth with hedging’ portfolio. See how this model between high watermark and low watermark meets his target to open dairy operation

3 yrs portfolio growth created using Planning with AI/ML

Next aspect is selection of securities under the ‘large cap’ in domestic as well as international sleeves. To make it optimize, model suggest different SMA strategies under UMA portfolio structure. This will give him maximum tax benefits.

SMA sleeves-1: Domestic large cap Tech focused. Benchmark against NASDAQ100 composite

SMA sleeves-2: International Tech sector focused. Benchmark against MSCI

SMA sleeves-3: Domestic Tax harvesting. Benchmark against DJI

SMA sleeves-4: Domestic Fixed Income. Benchmark against NASDAQ100

SMA sleeves-5: Domestic Large cap. Benchmark against S&P500

Traditional vs ML planning model

Now review how digital frameworks with innovation make it happen:

This ecosystem is heavily relied on API-first model to ingest the input and determine the intent. In the above example, intent comprises of what portfolio composition bring maximum statistical confidence and least possible standard deviation to meet Mr Smith’s objective to fund dairy business in 10 yrs.

LLM Orchestration layer ingests the intent from API call to create embeddings and meta-data. Meta-data including customer details would be used to query RDBMS to obtain additional customer details, risk score, goals, TOA info, CIP info, AML/KYC info. Also, model portfolio info from platforms like EnVesnet, Orion, etc. In the above example, planning platform suggested ‘Large Cap Equity’ model right fit. Primary focus of this model to ensure in 10 yrs landmark Mr. Smith portfolio reaches $12MM or more. Also, if he dies within 10 yrs, then term life-insurance payout of $12.5MM would fund his dairy vision. It does factor in market volatility impact (linear extrapolation of past 5 yrs volatility along with macro-economic models).

This is where Agentic AI’s curated content with similarity search (using RAG) from Vector DB plays critical role. It ingests the input and creates the intents (embeddings). Embeddings are far more personalized than static data in DB. They get updated from training data, as well as testing data of LLM in real-time. Hence accuracy in response is much higher and holistic. Typically, above 98.5% confidence level. On the contrary traditional static model portfolios operate in the range of 73–78% confidence intervals. RAG models using Vector DB prevent hallucination. See below RAG model high level platform interaction-

RAG based curated Wealth Planning

Here is the high-level comparison:

Conclusion:

This article illustrates how digital technology and innovation can be a game-changer. It accelerates end-to-end planning efficacy and ‘Personalization’. As mentioned earlier, ‘Personalization’ builds the winning strategy and competitive advantages. Big Wealth management warehouses, e.g., Morgan Stanley, J.P. Mogan, etc. spend north of a billion dollars in creating “Personalization” by investing in building digital integration, training their advisors, to name a few. The de-facto model Wall Street giants use is that the wealthier the client, the more “personalization” is inevitable. Robo-advisors don’t work for UHNW clients. On the other hand, it’s excellent for low-touch retail wealth customers.

It’s worth mentioning that AI/ML is still in its youth. As it matures, it’ll bring more sophistication and a personal touch to wealth planning. The best part ML models, on the job it trains itself. This enhances the efficacy of the model over time, and free-up advisors time. Companies that invest early in AL/ML will have more command over the market and clientele. In essence, a brighter future await for them