Five stories that caught my eye and what I took from them.
Armed conflict investor survival guide – Joachim Klement
- (1) Armed conflict investor survival guide - by Joachim Klement (substack.com)
- Having done work on the equity market impact of geopolitical crises, I thought this note on equity market behaviour around geopolitical crises was excellent.
- Our broad conclusions are the same (maybe that’s why I liked it): Buy the dip. The vast majority of geopolitical events do not matter for equity market performance over investment horizons of one month or longer.
- It’s also why I think all this talk of elections this year is overplayed from a market perspective. It’s a great filler of reports, podcasts and webinars, but far down the list of likely market movers.
- Joachim runs through four questions that help you think about the impact of the event with some thoughts on winners and losers.
- But for macro investors I would even condense it down to just one question: Can this event impact S&P 500 profits?
- Usually the answer is ‘no’, in which case you buy the dip.
- How can a war materially affect S&P 500 profits? An oil spike could do it, but only if large and sustained. Or a supply chain disruption, but few ‘risky’ countries are crucial enough to broadly hit global supply chains to a degree that cannot be replaced. Ukraine’s role in wire harnesses and the Houthi attacks were only small and temporary problems, but Taiwan’s semiconductor supply could be on a different level.
- Caveat: armed & sustained conflict with China would have the potential to break the rule. It could hit S&P 500 profits AND hit equities via a write-down of Chinese assets held by Western companies/investors.
- Bottom line: Buy the dip on geopolitical crises.
Fundamental Analysis via Machine Learning – Cao et al
- Full article: Fundamental Analysis via Machine Learning (tandfonline.com)
- Forecasting company earnings is one of the core parts of what equity analysts do. And it is one of the key data that investors use to build an investment case for a stock.
- This paper shows that AI can do this core task better than equity analysts.
- The AI forecasts were more accurate and better at predicting future earnings changes than analyst consensus.
- All of that despite relying only on information from financial statements while analysts have access to a much wider range of information.
- So is there information in financial statements that analysts fail to digest? Is the additional information analysts have access to just noise? O are analysts unable to differentiate between noise and signal?
- Also interesting: the AI forecasts created alpha as a trading strategy…in the out of sample back-test anyway.
- All of this would be terrible news for the army of equity analysts…if their main value were the precision of earnings forecasts. But analysts are more than earnings forecasters. They can be educators, forensic accountants, sounding boards, sources of specialist data…
- Or, to paraphrase a former boss of mine: We’re not in the prediction business. We’re in the entertainment business.
- As with so many things AI, the risk for equity analysts is not that they might be replaced by AI, but by an equity analyst that can use AI. As an analyst you are not going to out-model the competition. Earnings forecasts are not a differentiator anymore, if they ever were.
- Bottom line: AI forecasts can be a cheaper, un-biased and better than consensus earnings.
AI integration in investment management – Mercer
- AI integration in investment management (mercer.com)
- ChatGPT is 1 ½ years old and the whole world is talking about AI. But actual adoption is still very much in its infancy.
- The Investment Management industry is not known for being early adopters of technology. From personal experience, many companies are 2 or 3 tech generations behind the cutting edge.
- This survey from Mercer shows that it’s also the case with the usage of AI in investment management.
- At the surface, around half of managers say they are using AI today.
- But when you dig deeper the numbers shrink further. Only a quarter claim to be using generative AI. Turns out some consider quantitative screening as ‘using AI’.
- I agree with the survey that genAI is most useful in bottom-up security analysis.
- There are so many ways a purpose-built solution could help analysts. As a senior analyst it would be like having a team of junior analysts at your disposal. To name just a few, you could interrogate reports more quickly, stay on top of the eco-system around a company, never miss news with a second or third round effect on a company you cover, pick up a change in pattern in an industry more quickly, etc.
- Bottom line: There’s a long way to go for AI adoption in investment management. Someone should build a platform for that.
Give Your Ideas Some Legs – Oppezzo et al
- Give Your Ideas Some Legs: The Positive Effect of Walking on Creative Thinking (apa.org)
- Spring has sprung. The weather is getting better. No better time to boost your creativity (and health) by going for a walk.
- Nietzsche wrote that ‘All truly great thoughts are conceived by walking.’ over a century ago. But I feel much better about the connection between walking and creativity with some data and evidence to back it up.
- This Stanford study (not new, but timely) conducted four experiments, and all had the same conclusion: walking makes you substantially more creative
- You don’t even have to find rolling countryside or a forest. It works just as well walking on a treadmill indoors, even when staring at a blank wall. There is even some lasting benefit from walking after you’ve sat back down at your desk.
- Bottom line: Stop reading, leave a ‘like’ and go for a walk! And then come back to read the rest.
Artificial Intelligence Index Report 2024 - Stanford
Annual state of AI report from Stanford University
A great source of charts for anyone putting together a presentation on AI. It’s 502 pages, so either pick a section, read their summary or get ChatGPT to summarise it for you.
Two things that stood out to me
The dominance of the US over all other regions in all things AI. If AI lives up to its potential, US exceptionalism is far from dead.
Great overview of empirical studies on the productivity gains from AI.
Bottom line: The US lead in AI has only grown since ChatGPT