2 February 2024
Our latest edition of Investment Notes starts with a reminder on the circle of competence and cautions about making big decisions on the most recent news. With this disclaimer, we still share our views on what the latest changes can mean for investors. We also share some important points on the latest tariffs. In the end, we share a chart that may soothe the anxiety from all the worrying headlines.
Let’s dive in.
Let’s dive in.
A word of caution
Being a value investor does not mean buying low P/E stocks. It also means thinking about downside risks and not losing money. Benjamin Graham discussed this idea in the last Chapter 20 (Margin of Safety) of his best-selling book, The Intelligent Investor.
One essential step in maximising the Margin of Safety is knowing your Circle of Competence.
One essential step in maximising the Margin of Safety is knowing your Circle of Competence.
"I don't look to jump over 7-foot bars: I look around for 1-foot bars that I can step over."
- Warren Buffett
A local driver in Lagos or Mumbai can easily beat the professional.
Over time, I have learnt to be very serious about admitting the limits of my knowledge and not stepping out of those boundaries. It is great, of course, if you can expand them over time.
For me, AI is one such area. I think many investors will do much better in the long term by avoiding areas they know little about. In the short term, this may be hard as the general flow of money into the sector will lift all the boats, making Tech holdings a sure win. But historically, this has not worked unless someone could call when the market will change its direction.
Even Stanley Druckenmiller, probably the world’s best macro investor, struggled in 1999-2000. First, he shorted Internet stocks in 1999 and was quickly down $600mn. He then reversed, hired young tech analysts, and returned 35% for that year by going long the tech sector. He was out of the sector by the start of 2000 but could not resist seeing stocks skyrocketing and his fellow PMs making a lot of money. So, he bought tech stocks again in March 2000, at the market's peak. Since then, the NASDAQ went downhill, and Druckenmiller lost $3bn. For reference, he recouped the losses and ended the year in the green.
Even though the recent market action reminds me a lot of 1999-2000, I do not want to make any suggestions about today’s market and what it will do in the future.
The main point is that most people lose money in the stock market because they lack a clear strategy and focus on the latest articles. This week’s 20-30% daily moves in some of the largest companies may be a sign that too many market participants are just speculating rather than investing.
Question to consider: What is your circle of competence, what sectors you know particularly well?
Over time, I have learnt to be very serious about admitting the limits of my knowledge and not stepping out of those boundaries. It is great, of course, if you can expand them over time.
For me, AI is one such area. I think many investors will do much better in the long term by avoiding areas they know little about. In the short term, this may be hard as the general flow of money into the sector will lift all the boats, making Tech holdings a sure win. But historically, this has not worked unless someone could call when the market will change its direction.
Even Stanley Druckenmiller, probably the world’s best macro investor, struggled in 1999-2000. First, he shorted Internet stocks in 1999 and was quickly down $600mn. He then reversed, hired young tech analysts, and returned 35% for that year by going long the tech sector. He was out of the sector by the start of 2000 but could not resist seeing stocks skyrocketing and his fellow PMs making a lot of money. So, he bought tech stocks again in March 2000, at the market's peak. Since then, the NASDAQ went downhill, and Druckenmiller lost $3bn. For reference, he recouped the losses and ended the year in the green.
Even though the recent market action reminds me a lot of 1999-2000, I do not want to make any suggestions about today’s market and what it will do in the future.
The main point is that most people lose money in the stock market because they lack a clear strategy and focus on the latest articles. This week’s 20-30% daily moves in some of the largest companies may be a sign that too many market participants are just speculating rather than investing.
Question to consider: What is your circle of competence, what sectors you know particularly well?
More words of caution
There is a long list of inventions and predictions that have not lived up to expectations. Remember high expectations around Google Glass, 3D Printing, Space Tourism? Sometimes, predictions were too pessimistic like the 1876 Western Union internal memo: “The telephone has too many shortcomings to be considered as a serious means of communication.” Or, perhaps, the most famous prediciton by Thomas Watson, the founder of IBM: “I think there is a world market for maybe five computers.”
The most recent example is the correction in Alphabet’s share price after the release of ChatGPT.
Behavioural science has many examples proving that we tend to overreact to news in the short term and underappreciate its long-term importance.
Before sharing my other thoughts on DeepSeek, here are the results of the polls I ran on X...
The most recent example is the correction in Alphabet’s share price after the release of ChatGPT.
Behavioural science has many examples proving that we tend to overreact to news in the short term and underappreciate its long-term importance.
Before sharing my other thoughts on DeepSeek, here are the results of the polls I ran on X...

DeepSeek
There will be winners
It is possible that the DeepSeek phenomenon just means that AI will become more commoditised and widely available sooner than we thought. Some beneficiaries could be companies relying on call centres for customer services. Airbnb, for example, receives about 100 million calls a year to their customer service and even more queries via chat and emails. This requires the company to use the help of about 30 thousand employees to handle questions and help fix problems. Apart from the general salaries and fees to outsourcing companies (who may help handle call centres during peak seasons), Airbnb also faces costs when it provides an alternative accommodation to a guest or at least has to cover the payment fees (10-15% of revenue) when a booking gets cancelled because of poor communication.
I would not be surprised if Airbnb saves a few hundred million dollars on customer service if it properly integrates AI into its operations. For reference, its 2023 Operations & Support expenses amounted to $1.7bn (17% of revenue), and its FCF was $3.8bn.
On top of this, Airbnb could benefit from providing more personalised recommendations and creating itineraries for customers, which should improve the service and lead to possibly more booking value per customer or a higher number of bookings.
There are many other potential winners. One of the points from reading business biographies and case studies on companies like Netflix, is that corporate culture can make a big difference. It is worth asking whether the companies in your portfolio have the right culture and mindset to use the latest AI breakthroughs to their advantage.
I would not be surprised if Airbnb saves a few hundred million dollars on customer service if it properly integrates AI into its operations. For reference, its 2023 Operations & Support expenses amounted to $1.7bn (17% of revenue), and its FCF was $3.8bn.
On top of this, Airbnb could benefit from providing more personalised recommendations and creating itineraries for customers, which should improve the service and lead to possibly more booking value per customer or a higher number of bookings.
There are many other potential winners. One of the points from reading business biographies and case studies on companies like Netflix, is that corporate culture can make a big difference. It is worth asking whether the companies in your portfolio have the right culture and mindset to use the latest AI breakthroughs to their advantage.
Narrative and valuation multiples could change
While earnings drive stock returns in the long term, stock prices can experience significant moves simply because of changes in the narrative. I am thinking of Chinese stocks here, which have gone through a challenging period. In addition to regulatory and macro issues, the stocks suffered from the perception that the US sanctions would hinder Chinese growth in the IT space, including companies like Alibaba.
However, just this week, Alibaba claimed that its AI model (Qwen 2.5) outperformed DeepSeek in various benchmarks.
I can see how markets move beyond the idea of ‘US exceptionalism’ and become less sceptical about some of the international stocks, including Alibaba.
However, just this week, Alibaba claimed that its AI model (Qwen 2.5) outperformed DeepSeek in various benchmarks.
I can see how markets move beyond the idea of ‘US exceptionalism’ and become less sceptical about some of the international stocks, including Alibaba.
Less demand for chips and power?
I have no clue what DeepSeek news means for demand for chips and related IT infrastructure, including the power sector.
At some point, we may end up with ‘too many roads built to nowhere’ with too much capacity, failing to earn adequate returns.
On the other hand, I also see the point of the bulls who refer to Jevons paradox. Here is the quote from Wikipedia:
At some point, we may end up with ‘too many roads built to nowhere’ with too much capacity, failing to earn adequate returns.
On the other hand, I also see the point of the bulls who refer to Jevons paradox. Here is the quote from Wikipedia:
“The Jevons paradox occurs when technological advancements make a resource more efficient to use (thereby reducing the amount needed for a single application); however, as the cost of using the resource drops, overall demand increases, causing total resource consumption to rise.”
These days, for some reason, it has become popular to refer to this English economist when discussing the outlook for chips or oil.
AI Superinvestor?
Personally, I would love to use AI more in my stock research. If you have suggestions or tips or just want to brainstorm together, please don't hesitate to get in touch.
A few researchers seem to have achieved good results using LLM models and claim their strategies beat the market.
Here are some of them:
My immediate understanding is that AI can boost productivity (going through more companies or identifying critical drivers) but cannot replace human judgement in the investment approach that I follow, which is finding businesses that can do well over the next 5-10 years. Most of the models try to forecast expected changes in the short term.
As a side note, I should note one crucial lesson from reading Daniel Kahneman's Noise: Machines can often beat humans not because they are ‘smarter’ but because they are just more consistent. Their consistency comes from following rules, while humans can make different decisions on the same topic depending on the time of day or other random factors.
A few researchers seem to have achieved good results using LLM models and claim their strategies beat the market.
Here are some of them:
- Fine-Tuning Large Language Models for Stock Return Prediction Using Newsflow
- AI in Investment Analysis: LLMs for Equity Stock Ratings
- Can Large Language Models Beat Wall Street? Unveiling the Potential of AI in Stock Selection
- Financial Statement Analysis with Large Language Models
- What AI Sees in the Market (That You Might Not): Ten ways investors are, or should be, using large language models
My immediate understanding is that AI can boost productivity (going through more companies or identifying critical drivers) but cannot replace human judgement in the investment approach that I follow, which is finding businesses that can do well over the next 5-10 years. Most of the models try to forecast expected changes in the short term.
As a side note, I should note one crucial lesson from reading Daniel Kahneman's Noise: Machines can often beat humans not because they are ‘smarter’ but because they are just more consistent. Their consistency comes from following rules, while humans can make different decisions on the same topic depending on the time of day or other random factors.
Human Analysts - a dying breed?
Equity analysts are in rapid decline. According to Bloomberg, “compared with their post-financial crisis peak…the biggest banks globally have slashed the ranks of equity analysts by over 30% to lows not seen in at least a decade. Those who remain often cover twice, or even three times, as many companies.”
“At the world’s 15 biggest banks, the number of equity analysts has fallen to about 3,000 from almost 4,600 a decade ago, according to Vali Analytics.”
“At the world’s 15 biggest banks, the number of equity analysts has fallen to about 3,000 from almost 4,600 a decade ago, according to Vali Analytics.”

Enter Garry Kasparov
I remember reading in one of Garry Kasparov’s books (Deep Thinking: Where Machine Intelligence Ends and Human Creativity Begins) that a computer playing with a human can beat all other computers. His idea was that while machines were good at brute calculations of millions of variants, humans (professional players) could still understand the strategy better (positional weaknesses, game plan, relative strength of pieces, etc.). I am not sure this thesis would hold today as AlphaZero (trained by playing millions of games against itself) has taken chess to the next level.
Still, at least for now, using AI for certain parts of the investment research process should help investors.
Still, at least for now, using AI for certain parts of the investment research process should help investors.
Tariffs
I really dislike commenting on the news, but this Investment Note seems to be an exception.
So, here are a couple of WSJ articles that are worth paying attention to.
Tariff Threat Prompts Automakers to Find New Suppliers, Consider Higher Prices
Trump’s New Import Tariffs Will Jolt the Economy
So, here are a couple of WSJ articles that are worth paying attention to.
Tariff Threat Prompts Automakers to Find New Suppliers, Consider Higher Prices
- “The supply chain for building cars and all their parts is intertwined across North America; it’s common for components to cross borders multiple times before going into a vehicle. Because of that, analysts say costs would jump across the industry, but would vary across brands and models.”
- “U.S. car buyers on average would face price increases of roughly $3,000, according to estimates from analysts at Wolfe Research.”
- “The added costs and other fallout from the tariffs could cut earnings per share for GM by about half, Evercore analysts have estimated.”
Trump’s New Import Tariffs Will Jolt the Economy
- “A 25% tariff on goods from Canada and Mexico would bring the inflation rate up to about 3.2%, keeping it well above the Fed’s 2% target, according to Capital Economics.”
- “Grocery-price increases could be the inflation consumers notice first as tariffs force importers to pay 25% more for food, economists say. Mexico provides about half of U.S. fresh produce imports and is a particularly important supplier in the winter.”
- “New tariffs could raise U.S. prices for gasoline, jet fuel and home heating oil, because Canada supplies about 60% of U.S. crude-oil imports and Mexico another 10%, Gresser said. Together, those imports make up about 30% of the crude oil used in the U.S. Many domestic refineries are set up to process Canadian oil, and adjusting away from it isn’t a simple task, he added.”
- “The new across-the-board 10% tariff on Chinese goods will hit many consumer items for the first time, sparking possible retail price increases and retaliation from China. Consumer electronics, including smartphones and laptops, are one category that could face price increases.”
The Chart
If the news of the past week is too overwhelming, consider this chart.

This chart is based on various studies showing that stock prices change more overnight than during the day. So, for example, if you always buy a stock in the morning and sell at the end of the trading day you will significantly underperform the broader market.
There are various theories that try to explain this phenomenon, but one conclusion looks quite simple to me. Day trading may not be as profitable as a boring buy-and-hold strategy. Consider this chart when you read the next headline on a new "game-changing" product.
There are various theories that try to explain this phenomenon, but one conclusion looks quite simple to me. Day trading may not be as profitable as a boring buy-and-hold strategy. Consider this chart when you read the next headline on a new "game-changing" product.