Can AI propose a best practise portfolio? I tested ChatGPT and Google Bard and compared their results with expert opinions.
We’ve all heard the suggestions that AI could take away the need for financial advice and funds management – but are we really there yet? I asked ChatGPT and Google Bard to create me a perfect $10,000 portfolio and compared it to research on portfolios and what our fund managers have said in the past.
It’s worth noting that ChatGPT’s data is only up-to-date to 2020 so there’s probably a lot it is missing – including shares that have delisted – but either way, here is what the AI tool recommended.
Asset allocation:
The investments it suggested specific to each category were as follows – you’ll notice it also included ETFs in its suggestions.
We’ve all heard the suggestions that AI could take away the need for financial advice and funds management – but are we really there yet? I asked ChatGPT and Google Bard to create me a perfect $10,000 portfolio and compared it to research on portfolios and what our fund managers have said in the past.
It’s worth noting that ChatGPT’s data is only up-to-date to 2020 so there’s probably a lot it is missing – including shares that have delisted – but either way, here is what the AI tool recommended.
Asset allocation:
The investments it suggested specific to each category were as follows – you’ll notice it also included ETFs in its suggestions.
Australian equities
International equities
Australian bonds
Cash or cash equivalents
Google Bard is updated daily – it’s worth noting that unlike its counterpart, it is more cautious in its approach. It talks about a ‘sample portfolio’ rather than making a specific suggestion and talks about options to consider without naming particular companies.
Asset allocation:
I pushed it to name specific investment options and this was the outcome.
Australian equities
International equities
Index ETFs for equity exposure
For bonds, it suggests looking at those issued by the Australian government and did not suggest any corporate bonds or bond ETFs.
Cash
Asset allocation
It’s worth noting that both suggest fairly similar asset allocations of around 70% equities, 30% fixed interest and cash – though ChatGPT makes the effort to break it into Australian and international equities.
Without specifying the kind of investor we were after, both platforms gave us asset allocation options akin to a traditional growth portfolio. But that also means it is missing a lot in the form of diversification and modern tools.
For example, international bonds, property and real assets, alternatives like private markets and even currencies.
The use of alternatives in portfolios has seen significant growth in recent years as options have expanded for investors and allowed them to incorporate different return drivers and styles in their investments.
Suggested Australian equities
ChatGPT suggests a broad-based index ETF, along with a few different large-caps (Commonwealth Bank, CSL, BHP, Woolworths). It implies reasonable diversification given the direct shares come from different sectors and are all high-quality companies.
There are mixed views from experts on these stocks.
Janus Henderson's Daniel Sullivan points to BHP as a top 10 holding for their portfolio, while Woolworths was recently nominated by Plato Investment Management's Peter Gardner as a high conviction dividend pick for coming markets.
Commonwealth Bank has traditionally been a strong performer for shareholders, though Reece Birtles from Martin Currie argues that it is looking overvalued due to risks to earnings from the constrained supply of finance.
CSL recently disappointed the market in an update, but Anthony Golowenko from MLC Asset Management, nominates it as one of the quality growth names he has been tilting his portfolio towards.
Market Index’s broker consensus tool ranks these stocks as follows:
Google Bard also suggested a broad-based index ETF and specified a number of direct shares including Commonwealth Bank, Westpac, Woolworths, BHP, Rio Tinto. There’s some crossover with ChatGPT there but otherwise it’s worth pointing out that this isn’t a highly diverse list given double-ups for big banks and big miners.
Morningstar recently identified Westpac as its most undervalued pick in the financials sector for the coming quarter. Rob Crookston from WILSONS and Andrew McKie from Elston Asset Management also selected it over Commonwealth Bank in a recent episode of Buy Hold Sell due to its better valuation.
Both were mixed in their view of Rio Tinto compared to BHP, with Crookston leaning towards BHP due to its copper exposure, while McKie veered towards Rio Tinto on a relative valuation basis. Both noted that they viewed BHP and Rio Tinto as high quality businesses.
Market Index’s broker consensus tool ranked as follows:
Suggested international equities
ChatGPT suggested a broad-based ETF excluding the US market while Google Bard identified an ETF tracking the S&P 500, and the MSCI World Index (even if it struggles with ticker codes). Both Google Bard and ChatGPT made the same suggestions in terms of direct shares with Apple, Microsoft, Amazon.com and Alphabet featuring. Google Bard also suggested Tesla.
Some key concerns on the direct shares front would be that these stocks are all in the technology sector and thus don’t offer much diversification. While tech has been on a run this year, investors were hit hard last year.
That said, investors may feel torn.
These companies, along with Nvidia (NYSE: NVDA) and Meta Holdings (NASDAQ: META) were responsible for more than 95% of market gains in the S&P 500 since January. The Swell Asset Management flagship fund holds Amazon, Meta, Alphabet and Microsoft viewing them as offering growth opportunities to access artificial intelligence and machine learning.
Munro Partner’s Qiao Ma also highlighted Microsoft’s new AI-enabled software Microsoft 365 Co-Pilot as having enormous high margin revenue potential.
On the flip side, Hugh Selby-Smith from Talaria Asset Management recently pointed to Apple as a potential value trap given challenges in evaluating Apple’s future business activity and CAPEX.
Suggested fixed interest investments
Google Bard steered clear of naming specific investments when it came to fixed interest, speaking broadly about corporate bonds, government bonds and bond ETFs.
ChatGPT suggested one broad-based index ETF along with Australian government bonds – but no reference to corporate bonds.
Neither suggested international exposure in this front which would be a missed opportunity for investors.
Suggested cash investments
In a surprise twist, Google Bard finally got more specific with suggestions while ChatGPT went more general.
ChatGPT’s suggestion was a high interest savings account with a reputable bank. By contrast, Google Bard suggested high interest savings accounts, term deposits, money market funds and certificates of deposit. It also referenced examples of providers for each, such as Westpac and ING, or ETF issuers like iShares and Vanguard. It was arguably a better set of options for an investor to think about.
It’s worth noting that alongside a more detailed list of suggestions, Google Bard actually proposed a smaller allocation of 5% rather than ChatGPT’s 10%.
At this stage, there’s a lot to be said for professional management of asset allocation and investments.
The portfolios suggested by ChatGPT and Google Bard had some positives in the form of reasonable traditional diversification of asset allocation, but were lacking asset classes like alternatives, property and international fixed interest. Diversification was also an issue when it came to direct share suggestions.
Noting all this, some of the positives of using AI for portfolios is as a starting point for research – and using them in combination was also worthwhile if you consider that ChatGPT was more solid in terms of equities suggestions, while Google Bard offered a more comprehensive take on cash and cash equivalents.
Google Bard struggled to identify ticker codes and ETF names, which is a distinct downside too.
AI is only as effective as the search data available to it – it’s not necessarily going to look at indirect influences on the performance of a company or be able to extrapolate the future from recent announcements. Of course, that could change eventually – it’s always improving and learning but for the time being, a human designed portfolio still has an edge over the generic AI design.