After 12 monthly bouts between Man and Machine, the Battle of the Portfolios is finally over. “Man” is represented by Warren Buffett’s Berkshire Hathaway and various global indexes like the DJIA and the S&P 500 Index. We added our own AI-powered Aggregate Value Fund (AVF) just for benchmarking. The Machine is represented by Deep Deep, Aggregate Asset Management’s proprietary machine learning AI, developed by Eric Kong together with the team of analysts comprising Cheen Wee Kiang, Dr Chai Woon Huei and Lam Ji Ming. This contest was greenlighted by Aggregate’s founding director Kevin Tok and we are extremely grateful to The Edge Singapore for managing this contest, particularly the editorial team who helped us with the great storytelling.
The quick summary is this: Berkshire won the battle but Deep Deep beat the indexes after factoring all relevant transaction costs and dividends. It is a laudable achievement but what is incredible in this 12 months of digital blood and human toll, Deep Deep managed to beat Berkshire six times. Not too bad for a silicon brain.
Deep Deep has also beaten AVF (lest our own investors are perturbed) because Deep Deep is an extremist thought experimentation using 15 stocks in 15 different countries and we asked Deep Deep to pick the No. 1 stock in each country based on our AI.
In line with my use of science fiction references, we will make this and the coming two articles a trilogy paying homage to Star Wars (Episodes 4–6). In this article, I will focus on the performance metrics; in the November article I will talk about all the lessons we have learned in this journey and in the final article in December, what plans Aggregate have.
Just like in Star Wars, which we have introspective episodes where Rey and Luke Skywalker jumped off their respective X-wing fighters to meditate and reflect on what they have learned as Jedis, we hope to bring the saga to a climatic conclusion of what is in store for Aggregate. Since its inception on Oct 17, 2022, Deep Deep, after hitting a high of 26.03% in August, is now at 21.43% return after 12 months, with the decline in line with markets. Deep Deep has demonstrated in our rigorous back-testing that it can achieve an average of 6% outperformance above market but the AI is not powerful enough to reverse market trends. Deep Deep is impressive but not God-like, at least in its current iteration.
See also: Together in electric dreams
As mentioned previously, 46.66% of Berkshire’s listed holdings are in just one stock: Apple, which surged in recent months, and helped Berkshire return 26.12%. This bet on Apple cannot be the only reason Deep Deep was beaten but it is nonetheless a key. In other words, Deep Deep cannot beat concentrated bets that turn out right. You may be the smartest computational guy in the room but you can’t beat a lucky Stephen Chow or what Hollywood calls “having a plot armour”.
Now, once we factor the dividends and trading costs, Deep Deep outperformed all the indexes. Here is a gem of information worth exploring. Had we not rebalanced the portfolio and just simply held on to the first 15 stocks, Deep Deep would hit 22.98%, just a shade below Berkshire’s. As one datapoint is not a trend, we will not speculate further but it is an interesting factoid. Perhaps Lin Yutang, a philosopher who preaches the art of stillness (and Eric Kong is an avid reader of his books) is right, doing nothing sometimes has merits.
See also: Deep Deep's mission almost over, now to 'Xenomorph'
The Good, Bad and Ugly
Let’s look at the individual markets and stocks returns. Korea achieved a stunning 75.35% since inception. This is due to the NEOOTO’s 39.99% rise and till now, we are unsure what Deep Deep saw in the stock and that is perhaps worth exploring in the future — the case for explainable AI. US stocks came in second and Germany, third.
Next, the performances of the top 10 stocks are not the usual suspects widely recommended. They are neither FANGST stocks nor the equivalent in Asian markets. It is like a low-budget indie movie The Night of the Living Dead, costing just US$110,000 but made around US$30 million. Conversely, if you look at the 10 worst stocks, nothing really stands out.
Now, if you do a simple average return of the top 10 stocks (28.42%) vs the bottom 10 (–12.11%), Deep Deep has shown an impressive ability. Sure, it is not flawless and it will pick some duds but they are not so explosive as to cause irreparable damage to the portfolio. When we bought Home Bancorp just before the US small bank crisis in March, we suffered a 14.66% loss. As the portfolio consists of 15 equally weighted stocks, we managed to scrap through that month with a positive 1.09% return. Diversification is still relevant in the age of AI-powered investment. We must never forget that!
This brings us to the end of this well-fought battle between Man and Machine. Deep Deep has not said its goodbye to our readers yet. And we want to use this opportunity to thank our readers who followed us throughout this journey. Deep Deep, if it has a heart, would have said a big heartfelt thank you to all of you.
For more stories about where money flows, click here for Capital Section
Harry Huo is head, special projects, at Aggregate Asset Management
This article is a product advertisement. The Edge does not directly or indirectly make any endorsement and cannot verify the performances of the Deep Deep Machine Fund, which we understand is not an independent, dedicated, segregated fund. The Deep Deep Portfolio’s US$500,000 capital is a part of the AVF fund, which is, in turn, a part of Aggregate Asset Management’s AUM of approximately $550 million. While the AVF fund is audited yearly by EY and reports to MAS, this does not mean that the Deep Deep Portfolio is independently audited
