I Used AI to Model My 2 Kid's Portfolio : Here's What I Learned

A father's attempt to turn baby bonus cash and ang bao money into a real investing plan, with a little help from AI.

When my sons were born, like most Singaporean parents, I got a pile of well-meaning cash: the Baby Bonus, and stacks of ang bao from grandparents and relatives. The usual advice is "put it in a savings account" or "buy some insurance plan." I wanted to do something different, I wanted to actually invest it, and watch it grow alongside them.

The problem was, I had no idea what the numbers would actually look like 10, even 20 years down the road. So I decided to try something: I asked Claude (an AI assistant) to help me model it out, year by year, using real STI ETF data.

Here's what happened.

Setting the Rules

Before any modelling, I needed to lay out my actual plan in plain terms:

• Baby Bonus: I estimated I'd be able to invest about $1,600 a year until the elder son turns 6.5, then $800 a year until younger son turns 6.5 — roughly mirroring how the cash gift typically gets used up over time.
• Ang bao money: Kids typically collect around $2,000 a year in ang bao from relatives. I planned to invest all of it, but only until he's around 10. After that, I expect they will want a say in their own money, so I'll shift from "investing for them" to "manage your money."

Laying it out year-by-year, the fresh capital I'd be contributing looked like this:

Year	Baby Bonus	Ang Bao	Total
2026	$1,600	$2,000	$3,600
2027	$1,600	$2,000	$3,600
2028	$1,600	$2,000	$3,600
2029	$1,600	$2,000	$3,600
2030	$800	$2,000	$2,800
2031	$800	$2,000	$2,800
2032	$0	$2,000	$2,000
2033	$0	$0	$0

That's $22,000 of fresh capital over 8 years, averaging roughly $2,750 to $3,150 a year, depending on how you slice it.

Bringing In the STI ETF

They already held 7,101 units of the STI ETF (ES3) as a starting point. To model growth, I asked Claude to pull real historical data: the STI ETF's annualized total return since its 2002 inception sits around 8.12% (dividends reinvested), with a current dividend yield of about 3.3% a year. Backing out the yield gives an implied price-growth rate of roughly 4.8% a year.

Using those figures, we simulated the portfolio twice a year, every February and August, mimicking when dividends typically land and I'd top up with fresh cash, from 2027 through 2033.

The result: by the end of 2033, when my ang bao contributions stop, the plan showed the portfolio growing from 7,101 units (~$37,800) to roughly 12,371 units, worth about $92,400.

Then We Let It Ride

Here's the part I found most interesting. After 2033, I don't plan to add any more fresh capital, that's when I want my son to start learning to manage it himself. But I still wanted to see: what if we just let the dividends keep reinvesting, untouched, all the way to when they turn 23 and 21 in 2046?

Running the same assumptions forward, no new cash, dividends automatically reinvested twice a year, the model projected:

Year	Portfolio Value
2033	~$92,400
2038	~$137,700
2043	~$205,200
2046	~$260,700

From $22,000 of actual money put in, plus the ETF units I already held, compounding alone could nearly triple the portfolio's value between 2033 and 2046, purely from dividends buying more units, and those units appreciating over time.

A Healthy Dose of Skepticism

I want to be upfront: this is not a forecast, it's a "what if." A few things I keep reminding myself:

• The 8.12% return and 3.3% yield are historical, over a 20+ year horizon, small changes to these assumptions swing the final number by tens of thousands of dollars.
• Markets don't move in a straight line. Real returns will come in bursts and crashes, not a smooth compounding curve.
• No fees, no taxes on dividends (Singapore doesn't tax personal dividend income, thankfully), and no real-world share-lot rounding were factored in.

What this exercise did give me wasn't a guaranteed number — it was a framework. Having AI walk through the actual math, year by year, with real historical data, turned a vague "I should invest for my kid" intention into something concrete I can track, adjust, and eventually hand over.

The Real Point

Honestly, the dollar figure at the end matters less than what I want this to teach my son. By the time he's 10 and asking about his ang bao money, I want to hand him not just a portfolio, but the reasoning behind it, why we chose an index fund, why we let dividends compound, why patience matters more than picking hot stocks.

If AI can help me get that story straight for him, that's a bigger win than any specific number on a spreadsheet.