How to build a portfolio in a fragile world

The world is becoming an increasingly volatile place, adding to uncertainty for investors. Wars are raging in Europe and the Middle East. Global trade tensions are rising, to the extent that the US has announced a 100 per cent tariff on Chinese electric vehicles. But it is not all doom and gloom. Inflation has subsided, which, coupled with the surge in demand for artificial intelligence (AI) technologies, has driven the S&P 500 to new highs. Even the much-maligned FTSE 100 hit record highs last month as UK inflation fell back towards 2 per cent.

Whether the stock market will continue to rise or take a turn for the worse is impossible to know. In the UK, a snap election has been called, and one of the most contentious elections of the last century is on the horizon in the US. Rising debt in both countries is putting a cap on how much they can subsidise the green transition, potentially leaving the world vulnerable to disruption from climate change. Yet, despite all these fears, Goldman Sachs has forecast that AI has the potential to increase annual global GDP by 7 per cent over 10 years – or almost $7tn (£5.5tn).

In a world that has been rocked by several major events, it’s easy to believe that what the investment world now needs is a greater ability to predict the future – through bigger computers that can process ever more data and better analysis. But there is an argument that instead we should focus on strategies that can thrive in a volatile environment, by minimising the downside risk from both domestic politics and geopolitics, while leaving us exposed to what could be the unbounded upside from technological innovation.

In his book Antifragile: Things That Gain From Disorder, statistician and former options trader Nassim Nicholas Taleb urges people to create more “positive optionality” in their life and their investments. This means designing opportunities where there is a chance for huge returns from positive Black Swan events, while minimising losses from negative Black Swan events. This can apply to people’s personal lives, but also political systems, businesses and investment portfolios.

To understand the idea of optionality, consider financial options as an example. An option gives the holder the right to buy or sell an asset at a given price. An investor can buy a ‘call option’ to secure the right to buy a stock at a stated price in the future. Meanwhile, a ‘put option’ gives them the right to sell. The premium is the price paid for this right.

A contract will only last for a designated period. Once it expires, the holder loses the right to buy or sell the asset. The premium is dependent on a few factors, including how far away from the strike price the share price is, how volatile the stock is and how long the contract lasts. If it is a volatile stock with a long contract time, then the premium paid for a call option would be higher, because there is a greater likelihood that the price might exceed the strike price.

Call options reduce the downside without limiting the upside. If an investor buys a stock, the most they can lose on the investment is 100 per cent, whereas with an option the most they can lose is the premium, which will be less than the stock price. However, there is no cap on the amount the option can increase in price. In other words, there are positive asymmetric returns.

The other upside of the option is that increased volatility can be a benefit. If there is more volatility it increases the likelihood that the price of the stock will rise above the strike price. Of course, it is also more likely to fall below, but because the downside is capped while the upside is unlimited, volatility increases the expected return. In other words, it is ‘antifragile’.

Taleb acknowledges that financial derivative options tend to be expensive because the sellers are aware of the positive asymmetry. The key is to find areas of underappreciated, implicit optionality. “Because of the domain dependence of our minds, we don’t recognise [optionality] in other places, where these options tend to remain underpriced or not priced at all,” he writes.

Trial and error

An underappreciated type of optionality is experimenting. It provides information with potential unbounded upside, and the cost is the time and effort put into the initial experimentation. “Any trial and error can be seen as the expression of an option, so long as one is capable of identifying a favourable result and exploiting it,” states Taleb.

In the case of a company, the price of a trial is the amount spent on research and development (R&D). It doesn’t always pay off, but the downside is limited to the initial investment, in the same way that a loss on an option is limited to the premium paid for it. 

Apple (US:APPL) has produced many products that failed. In the 1990s, it invented the Newton digital notepad, which flopped, much like the iPod hi-fi speaker system it released in 2006.

However, in 2007, the iPhone was released, and it went on to become one of the world’s most profitable consumer items. Last year, the iPhone contributed $43.8bn in revenue to the company. This is not to mention the fact the iPhone created a whole new computing platform which Apple has come to dominate. Apple doesn’t split out its App Store revenue, but ‘Services’ contributed $22.3bn last year and is now the fastest growing part of the business.

Even before the iPhone became a success, former chief executive Steve Jobs understood the importance of trial and error. In 2005, he told the Stanford graduating class to, “stay hungry, stay foolish”. This attitude might be costly in the short term, but in Jobs’ eyes the cost of failure was never greater than the potential for success.

Despite the example of Apple, the stock market has still not managed to grasp the asymmetric returns of trial and error. In October 2021, Facebook chief executive Mark Zuckerberg rebranded his company Meta (US:META). At the same time, the company announced it would be breaking out the performance of its virtual reality business, known as Reality Labs, in its next results.

When the results were published, it turned out that in the three months to December 2021, Reality Labs had made an operating loss of $3.3bn on revenue of just $877mn. Then in the three months to March 2022, Reality Labs revenue dropped to $695mn while the operating loss stayed at $2.96bn. By this point, the market had turned on Meta, the share price having fallen 40 per cent from the start of the year.

Losses from Reality Labs continued to mount, and eventually investors lost patience. In October 2022, Altimeter Capital chief executive Brad Gerstner published an open letter titled Time to Get Fit. In it, he urged Meta to keep its Reality Labs spend to no more than $5bn a year. “An estimated $100B+ investment in an unknown future is super-sized and terrifying, even by Silicon Valley standards,” Gerstner wrote.

Meta’s share price had collapsed due to the perceived profligacy of its spending. In 2022, Meta had ramped up its capital spending to $31.4bn, from $18.6bn the year before. This, coupled with rising interest rates, caused investors to completely lose faith in the business. The narrative was that because Mark Zuckerberg had a controlling stake, he was pursuing passion projects and failing to focus on shareholder returns.

The belief was that Meta’s increased capex and R&D spending was solely going into the metaverse project. The reality was that data centres and server spend could also be used to improve AI capability, with the aim of using big data to boost the returns from its advertising business.

In December 2022, ChatGPT came out and within months the world was obsessed with generative AI. The huge investment Meta had made in graphics processing units (GPUs) from Nvidia (US:NVDA) put it in a prime position to profit. Meta had been using AI to improve its targeted advertising to great effect, but now it could redeploy these computers to train a large language model. Since then, it has added AI chatbots to Facebook, Instagram and WhatsApp, as well as launching a platform for its business customers to use AI to generate advertising content.

In 2023, Meta was one of the best performing companies in the world, with its share price rising 193 per cent. It is now seen as one of the companies likely to profit from generative AI technology as it integrates it across the business. “We are confident in Meta’s proven ability to successfully scale and monetise the new products and expect Meta will be able to do so with AI,” wrote Jefferies analyst Brent Thill last month.

Survivorship bias makes it seem as though AI was always going to be a more profitable investment than virtual reality. However, back in 2021 it was impossible to know this. At the time, Meta was pursuing a diverse strategy by investing in virtual reality while also building out its data centres. No one, not even OpenAI, expected the release of ChatGPT to be the phenomenon it was. But for Meta investors, because of the billions it has invested in AI data centres, it proved to be a positive Black Swan.

The risks of standing still

In the short term, investors worry about companies wasting money. However, the real threat of decline comes when they stop investing in research. This won’t necessarily cause a dramatic collapse in the share price, but over time as competitors are allowed to catch up, there will be a slow and steady decline.

In 2012, BT (BT.) had a near monopoly on British telecommunications networks. The newest technology coming through at the time was ultra-fast fibre-optic broadband, but BT failed to invest fast enough in updating its networks in the early 2010s. This left a gap in the market for alternative providers to enter.

In 2011, CityFibre was formed with the aim of providing ultra-fast internet. This attracted a host of other alternative network providers wanting to fill the gap left by BT. In 2018, BT realised its failure and announced its ‘Fibre First’ scheme. It ramped up capex from £3.1bn in 2017 to £5.3bn in 2023, but by this point it was too late. Today, competition is driving down prices, which has lowered BT’s return on its investment, with its return on equity hovering below 10 per cent. Consequently, the share price has dropped 75 per cent since 2015.

Amazon (US:AMZN) founder Jeff Bezos was acutely aware of the risks of standing still. He was obsessed with instilling ‘Day One’ thinking into his employees. This meant starting every day thinking fresh about the business and how it could be improved. In a 2016 letter to shareholders, he expanded on this theory by explaining the main risks to Day One thinking, and how businesses can fall into ‘Day Two’ stasis and then decline slowly into irrelevance.

One of the risks he identified was a failure to embrace external trends. “The outside world can push you into Day Two if you won’t or can’t embrace powerful trends quickly,” he wrote. “If you fight them, you’re probably fighting the future.” In this letter eight years ago he identified cloud computing, machine learning and AI as the main external trends that Amazon needed to invest in.

It was only the year before that Amazon had started breaking out results from its cloud computing division, Amazon Web Services (AWS). At the time, some investors were sceptical about its investment in AWS, but since 2015 the division’s operating profit has risen 1,266 per cent from $1.8bn to $24.6bn. Last year, AWS made up two-thirds of the company’s profit. In 2022, Amazon would have made an operating loss of $10bn without AWS’s contribution.

Not all of Bezos’ 2016 prophecies came true. In the same letter, he talked about the potential of Amazon Alexa and how it was a struggle to keep the voice-activated hardware Echo in stock. However, underperformance meant at the end of last year Amazon laid off hundreds of employees in its Alexa business. For Amazon investors, this wasn’t a problem because AWS’s exponential profit growth was driving the business forward. Since 2016, Amazon’s share price has increased 400 per cent.

Balance sheet strength

Constant innovation is essential to keep a business alive, but there are only a few companies that have the cash to make these asymmetrical bets. The thing Meta and Amazon had in common is that they were both cash generative and had relatively little debt on their balance sheets.

In short, Meta could risk losing money on AI and virtual reality because of its huge cash-generative advertising business and robust balance sheet. “Meta’s robust free cash generation allows it to continue funding innovation while also returning cash to shareholders,” explained Thill last month. This comment came after Meta had increased its 2024 capex guidance from $30bn-$37bn to $35bn-$40bn, to support its “artificial intelligence roadmap”.

For a business to be antifragile it needs to combine a strong balance sheet and free cash flow generation with an innovative mindset. The strength of the balance sheet gives businesses the opportunity to capitalise on opportunities during periods of volatility. “Redundancy is ambiguous because it seems like a waste if nothing unusual happens,” however, “redundancy is not necessarily ‘wussy’; it can be extremely aggressive”, comments Taleb.

In 2022, Nvidia made most of its money from selling its GPUs for gaming. In the three months to May that year, it made $4.6bn from gaming and $3.7bn from computing. At that point, investors knew that Nvidia had potential as an AI stock, but investors were worried demand from gamers would fall as the world emerged from lockdowns. The share price was down 40 per cent over the previous five months.

However, the positive Black Swan of the release of ChatGPT at the end of the year turbo-charged demand for its GPUs. In the three months to April this year, Nvidia’s revenue rose 262 per cent year on year to $26bn, ahead of the $24.6bn expected by FactSet broker consensus. Most of this growth came from the AI data centre business, which saw revenue grow 427 per cent year on year to $22.6bn. This was split between the compute division, which makes up 86 per cent of the data centre business, and networking.

Nvidia was only able to pivot quickly and invest heavily in R&D when ChatGPT proved to be a hit because of its strong balance sheet. In its 2022 financial year, it spent $2.8bn on R&D, while last year it had ramped this up to $8.7bn.

All this spending has allowed the company to release its newest chip, Blackwell, which specialises in AI. Chief executive Jensen Huang said demand for Blackwell is already outstripping supply and he expects “a lot” of revenue from Blackwell this year. “Nvidia could grow faster than we previously thought as Blackwell, networking, software and sovereign areas become bigger contributors,” said Melius Research analyst Ben Reitzes.

In 2022, Nvidia had the potential to make exponential returns, but there was a risk that the downturn in gaming would drag on earnings. However, this was a bet Taleb would likely have taken. His main problem is with companies that offer medium risk and medium returns. In other words, the upside is capped but there is still exposure to negative Black Swan events.

To find businesses with positive optionality we’ve run a screen ranking the S&P 500 companies by balance sheet strength, free cash flow margin and R&D spending as a per cent of revenue. Meta came out top of this list, followed by other big tech companies including Cisco (CSCO), Alphabet (US:GOOG), Microsoft (US:MSFT) and Nvidia. Also in the list is semiconductor designer Qualcomm (US:QCOM), industrial software companies Ansys (US:ANSS) and Synopsys (US:SNPS), and Airbnb (US:ABNB).

A 90/10 portfolio

It would be risky, though, to fill your portfolio with high-spending tech companies. To mitigate the risk, Taleb advises using a ‘barbell strategy’. He describes this as “a combination of extremes kept separate, with avoidance of the middle”.

In other words, playing it safe in some areas of life to protect yourself from negative Black Swans, while taking lots of risk in others to retain exposure to the benefits of positive unforeseen events.

Using this theory, Taleb recommends portfolio construction that differs from the traditional 60/40 split. In the traditional construction, investors are expected to have 60 per cent of their portfolio in stocks with the remaining 40 per cent in bonds. The theory is that this should make them resilient to changes in the economy. Traditionally, when economies slow down, company earnings fall and stock prices drop. To boost the economy, central banks usually cut interest rates, which pushes up the price of bonds.

However, this portfolio is vulnerable to stagflation. In this scenario, when inflation is increasing but growth is slowing, it is not possible for central banks to cut rates to boost the economy. As was the case in 2022, this would cause both stocks and bonds to fall together.

Daniel Siluk, bond fund manger at Janus Henderson, tells Investors’ Chronicle that since the pandemic lockdowns there has been increased volatility in the relationship between bonds and stocks, meaning bonds haven’t been acting like a reliable hedge to stock prices. “In order for this to return you will have to see a slowdown in growth, inflation come down to target and company earnings to suffer, and then bond prices will start rallying,” he explained.

Siluk notes that investors didn’t complain about the positive correlation between bonds and stocks in the post-financial period because loose monetary policy meant the prices of both were moving upwards. It was only when they started moving down in tendem that investors started to be concerned.

To mitigate against these periods of correlation, investors could adopt Taleb’s barbell portfolio. “If you put 90 per cent of your funds in boring cash (assuming you are protected by inflation) and 10 per cent in very risky, maximally risky, securities, you cannot possibly lose more than 10 per cent, while you are exposed to massive upside,” he writes in Antifragile.

The fragility of banks 

Taleb is dismissive of companies promising average returns and average risk because of their potential exposure to negative Black Swans. In particular, this applies to banks, which carry a lot of balance sheet risk without much upside. “Businesses with negative optionality such as banking had a horrible performance through history: banks lose periodically every penny made in their history thanks to blowups,” writes Taleb.

An example from last March was the rapid decline of Silicon Valley Bank. At the time, it seemed as though the bank’s balance sheet was in a strong position, with most of its assets being long-term US Treasuries. However, while its assets were in long-term bonds, its customers’ deposits were short-term liabilities, which created a duration mismatch. When interest rates then started to rise rapidly, the bank had to write down the value of its assets.

Its customers, mostly venture capital firms and start-ups, then started to panic and withdraw their funds. On 8 March, Silicon Valley announced a $2.25bn capital raise to strengthen its balance sheet and reassured depositors their cash would be safe. However, this only led to more panic among its customers and by 10 March trading was halted. Within a couple of days Silicon Valley Bank’s shareholders lost all their money. That risk of total wipeout was always there, but there was never any chance of the 3,300 per cent share price increase Nvidia has experienced in the past five years.

In hindsight, it is easy to say that Silicon Valley should not have been so exposed to the risk of rapid interest rate rises. Yet the Fed itself was insisting that the likelihood was that the bout of inflation being experienced was transitory. Despite its access to vast amounts of macro-economic data, even it didn’t know what was going to happen.

The illusion of the post-Cold War era was that the world had become a permanently more stable place. However, history shows that volatility is inevitable and no matter how large our computers become we are never going to be able to forecast accurately. What Taleb’s book provides is a framework for how to live, and invest, when we don’t know what is going on.