That Warren Buhai, formerly a resources analyst and now a portfolio manager in Stanlib’s multi-asset portfolio team, is keen to use this interview with Currency to feed into their AI models shouldn’t come as a surprise. His team is hypersensitive to biases in their ranks, he explains, so they have been doggedly using AI to sniff out their own investing blinkers over the past couple of years.
“I did this exercise a few weeks ago using [Microsoft’s AI tool Copilot] to give me the potential blindspots of the team – some of them were obvious and some of them were left-field,” says Buhai. “It would say something like: ‘Your negative scenarios are less than 20% of your forward-looking scenarios, and those scenarios are critical because the downside is potentially huge but you guys hardly discuss them.’ A complete blindspot.”
For the luddites among us, it may be depressing to know that AI – whether it’s Copilot or Claude or ChatGPT – is vital to sift through hours of meetings in seconds and spit out evidence of your own shortcomings. But it’s far from the only tool Buhai’s team has spent the past seven years refining.
Having cut his teeth initially at Standard Bank corporate finance, and then as a mining analyst at Stanlib from 2005, Buhai started out conventionally enough: building models of mining conglomerates and, in some cases, individual mines. It was a great way to learn a business, he says.
“The guy I worked for at the time had a five-line model of each mining company because he knew each of the key drivers after 20 years in the business. But he said he could only get to that point having gone through these thousand-line models for each division, and then he could figure out how to summarise it,” he tells Currency.
Which all sounds totally pre-2022 – the landmark year that ChatGPT and generative AI burst onto the scene, upending just about everything, not least in asset management.
Catching the bias
Part of the problem with traditional valuation models, says Buhai, is that it’s quite easy to engineer the answer you’d like to obtain. Bias, in other words, finds a way through under the guise of empirical evidence.
This was very evident when it came to his fellow mining analysts. “What amazed me was how different [their] models were in the individual lines – in currency price forecasts and commodity forecasts – and then when it came to the overall valuation they would fall within 10% of each other. It proved to me that the more complex the model, the more you could get any answer you want.”
It was just one of the main frailties he spotted in the way analysts operate. Another is that a bizarrely small number spent enough time thinking about the discount rate – which would imply a company’s cost of capital based on a country’s long-bond rates. In his corporate finance team they’d spend endless hours debating the cost of capital for multinational firms and how to calculate it correctly to get to the right value for a deal.
“Then to come into this industry on the buy side, where it’s almost a trivial input, was so weird,” he says.
In too many cases, he says, vulnerability in investment conclusions came down to bias.
“These guys had been in the industry for 15 to 20 years and their view on Anglo or Billiton or Glencore was a bias – they’d had a fight with management or they loved the company or were influenced by internal house views,” he says.
Sell-side analysts, Buhai says, hardly ever deem a company a “sell”, fearing they’ll lose access to management. It means buy-side analysts, like Buhai, interpret “hold” calls “quite cautiously”.
The ADD generation
But if you’re aware there is perpetual bias in share recommendations (and sometimes perverse motives even), how do you perform an honest assessment of an asset’s prospects?
This is where Stanlib’s theory of market “lenses” comes into play, which is compiled using reams of data. There are six of them: macro-economics and valuations; momentum and liquidity; sentiment and volatility.
Take the first: macro-economics and valuations. These are key fundamentals for any asset class but especially equities, and they play a more important role the longer the time period. As Buhai explains, “valuations, when you back-test them, are only useful in some markets over more than seven years”.
Still, imagine telling clients a thesis will pan out in almost a decade; a tough sell for any asset manager at the best of times, let alone when markets are turning on a dime and people can flip in and out of positions through trading apps – underscoring the collision between long-term data and instant gratification.
“I don’t think people have the ability anymore to think very long term. It’s what they should do, but I don’t know if they can,” says Buhai. “For us as professional money managers we can only say: we know economics and valuation are important, but our tactical asset allocation time horizon has moved to six months because access to information is so much quicker and the participants in the market behave so much quicker.”
Which is why momentum and liquidity matter. Flows are critical – where’s the money going to and where’s it coming from? In other words, is there more buying than selling; are analysts revising their earnings forecasts lower; and are interest rates rising or falling, which affects liquidity. If the answer to all these questions shows that a situation is deteriorating, that’s an important signal to consider derisking positions, he says.
Follow the evidence
Taken together, “the other lenses have got to give you a good reason to keep on holding an asset class. You can’t just have blind faith in the future even though the momentum is dying and the asset is plummeting,” says Buhai.
It’s all about following the evidence, and the war in Iran is a case in point. “History tells us that when you look forward three, six or 12 months, most of the time markets are higher [after a war]. You’ve got to have a good reason to go against what history tells us happens,” he says.
Which takes us neatly to the last two short-term lenses: sentiment and volatility. These, says Buhai, “are usually useful at extremes”. In other words, when the market gets extremely bullish or extremely bearish, they can act as good contra-indicators. That’s when it may be useful to consider going against the tide.
Of course, picking the bottom in any market, even armed with historical data, is fraught. But Buhai says this is just one indicator. “The process is telling you to think about buying. It’s not telling you that you should buy, but think about it. Do you pull the trigger or not is a much harder question, especially in the heat of the moment.”
So, to the only questions investors care about, do these lenses produce better returns?
Unequivocally yes, says Buhai.
“If I can say: ‘Here’s the data and we’ve tested it where you can get probabilities of 70% or more,’ you lean into them. If you don’t have [a process] you can rely on, you are flying blind, and the risk with that is that you make emotional decisions to the detriment of your long-term objectives. And the results, so far at least, have been good.”
Embrace dissenters
Embracing dissenting views is also critical skill when it comes to conquering bias.
Stanlib’s team also develops market scenarios, and then independently votes on their probabilities each week. “One of the things we are most worried about is herding behaviour, because we talk to each other every day. Diverse people’s views are important to getting the right outcome.”
Buhai had visceral first-hand experience of this when he finished his articles in chartered accountancy and joined Standard Bank’s corporate finance division. “For the first time, I was working with engineers and lawyers, and speaking to them was such a refreshing thing because they thought completely differently to me. It was a massive eye-opener.”
His own interview for the Standard Bank job, more than 20 years ago, was probably just as much of a culture clash. The team there were all from fancy private schools in KwaZulu-Natal, but Buhai had gone to a government school, Highlands North Boys High School in Joburg, and then studied part-time after matric.
“They asked me: ‘What do your mates do?’ expecting me to say they’re also getting into investment banking, and I said: ‘Well, they’re in the trades.’ And they said: ‘Oh, what are they trading? Derivatives, futures?’ I said: ‘They’re traders – you know: plumbers, electricians. And they liked that. It meant I thought differently to them,” he says.
Divergent views, he says, are integral to Stanlib’s multi-asset philosophy; if everyone is bullish on one stock, and one person isn’t, it’s the minority report you want to hear. “The voice that’s dissenting is the most important one, often by far,” he says – particularly for the junior members on a team, who struggle to speak up.
Forged in fire
Highlands North is a pretty hard-bitten Joburg hood to grow up in, though Buhai doesn’t fit the archetype of aggressive market punter. And the reason for that is twofold: Bernie Madoff and the global financial crisis.
Enticed to join a private wealth manager in Boca Raton, Florida, in 2007, he and his wife thought: “Why not?”
“From the day I arrived all I had was clients asking me: ‘Can you get me into Bernie?’” Yes – the Bernie Madoff, Ponzi schemer extraordinaire.
Buhai thought: “Who is this guy?”
Seeing that Madoff was chair of the Nasdaq, Buhai got hold of some of his marketing material – a one-page affair that had an extraordinarily linear chart showing Madoff’s fund versus the S&P 500, and some explanations on his finesse with fancy financial structures.
“You couldn’t draw a straighter line with a ruler. I thought: ‘This is a bit odd, but phew, this guy’s impressive.’ And our saving grace was the guy I worked for had been burnt by another Ponzi scheme a few years before, which taught him never to give money to outside people, so he managed all the money himself.”
When Madoff’s scam collapsed in December 2008 – eventually revealing a $65bn fraud – Buhai says the town of Boca Raton went from boom to bust overnight. “Retirees had to go and work at McDonald’s, the property market cratered – it was horrendous,” he recalls.
It happened at the same time as the global financial crisis — a cataclysm that profoundly impacted Buhai’s psyche.
“I can see from my own personal investments that I’m more conservative than I should have been in my approach to risk. And my training as an accountant was always risk first. What I learnt from doing articles was: look for the dark side of what’s going on in companies,” he says.
Having made the “mistake” of coming back home for a conference in the months after the crash, Buhai realised how homesick he was. He sold his house in Boca (at a 50% write off for the bank), and returned to South Africa and Stanlib in 2009.
Cycles always matter
Yet for all his conservatism, his choice of resource stocks as the place to start is hardly for the faint of heart. Not least because in many cases mining shares are all about the great, beguiling prospect.
Buhai says he soon learnt not to get sucked into a story that makes you believe the commodity market is not cyclical. “That’s always the biggest risk I’ve found with resources: that when it’s gone up multiple times – and I saw this in the last six months – everyone comes out from the sell-side with reasons as to why it can only go higher.”
While central banks have helped to elongate economic cycles – think China and its stimulus programmes that drove resources shares up for three years after the 2008 crash – this still doesn’t break the cyclicality of it all.
“Cycles remain driven by human nature, and supply and demand. And when prices are high, commodity companies tend to overinvest so that, seven years down the line, you’re going to have an oversupply. They’re cyclical for a reason.”
Ride your winners
That’s not to say he didn’t have some stonking winners – like Wesizwe Platinum back in the day. In Wesizwe’s case, Stanlib got in early and the stock went from being 2% of its portfolio to 10% as the price soared.
One call, in other words, can make all the difference. Just like Warren Buffett’s decision to buy Apple in 2016 – a contrarian bet at the time that has, thanks to compounding, generated almost half of his company’s return since it began life in 1955.
This is another formative lesson: to get your position sizes right (ideally, in the winners) and cut your losses quickly.
Stanlib’s decision not to go overweight Naspers when the tech share came to account for as much as 25% of the value of the entire JSE eight years ago is a hard learning that has shaped Buhai’s approach. While the portfolio did have Naspers exposure, it wasn’t enough – though it all came down to very sound reasons, like valuation fears and concentration risk. It was pension fund money the company was managing, after all.
“I think with the framework as it exists today, we may have interpreted those signals differently,” he says. Like placing greater weight on the momentum, liquidity and economics in China.
“We thought we were doing the right thing for our clients: we had a much more diversified position in companies that were winning, but they weren’t winning enough. It always is a relative game.”
A relative game, and as much art as science, sometimes. Which is why doing your utmost to identify your biases, as early as possible, is an underrated secret weapon.
ALSO READ:
- South African value legend on what he’s buying, and why fund management is broken
- Bond maven, author, investing evangelist: Cartesian’s Anthea Gardner
- Surviving Steinhoff, and listening to the team naysayer
Top image: Stanlib’s Warren Buhai. Picture: supplied.
Sign up to Currency’s weekly newsletters to receive your own bulletin of weekday news and weekend treats. Register here.
