YWR: The Death (and Rebirth) of HF Capital
YWR: The Death (and Rebirth) of HF Capital
Erik: Dennis!! Thanks for the birthday call! Appreciate it! It’s been a while. How’s the new CIO job going? Oh. I’m sorry. Can you talk freely?
Dennis: It’s fine. I’m in one of those little phone call rooms. How’s the job? Oooof… Good I guess. The firm is massive and we’re growing, but the role isn’t exactly what I thought.
Erik: Why what’s up? It’s the dream job.
Dennis: Yes, it’s a great firm and everyone is nice, but the CEO says I have to figure out how to make AI part of the investment process. I thought I was going to be working with the analysts on investment ideas, but this is turning into more of a tech job. Now I’m talking to cloud data warehousing firms like Snowflake and learning AWS Bedrock. I’m just not sure it’s a good fit for me.
Erik: OK, well look at the positives. It pays well and everyone is nice. And can you look at this AI project in a positive light and see how maybe it fits into a bigger picture pathway for your career?
Dennis: That’s what I was thinking about. I was thinking maybe in 5 years it will be a valuable skill set to have implemented AI into an investment process for a big asset management firm..
Director. Stop clip there.
Did you catch that?
“I was thinking maybe in 5 years it will be a valuable skill set to have implemented AI into an investment process for a big firm..
Wow. Biggest understatement ever.
Up until now there’s been a dividing line in investment strategies for fund managers.
You are either ‘quant’ or ‘fundamental’.
If you’re a quant and you have an investor meeting you talk about:
your models,
computing power,
the smart geniuses on your team,
risk management processes,
backtests and how statistically your model works over time.
And you didn’t talk too much about what you own at any one point in time, because that isn’t really what matters. It’s always changing. What matters is the underlying model.
If you are ‘fundamental’ you talk about:
Your values as a fundamental investor,
How you invest for the long-term,
The great companies you own in your portfolio,
The positive culture of the firm,
How everyone on your team has an MBA from Wharton,
How you don’t sell on the lows,
How many company meetings you have done over the last 12 months.
And you don’t talk about too much about compute, data, network effects and your quantitative skills.
The investment process slide in your deck probably looks something like this:
Or this:
Folks…. I have some bad news.
If your investment process slide looks like the ones above, I think it works for another 18 months. Tops. Maybe less.
Actually, it’s probably already dead.
AI changes everything, especially asset management.
In another two years we will be at the point where you won’t be able to meet an investor and with a straight face and tell them you pick investments by waking up in the morning, reading the Financial Times, running a few screens and trying to pick great companies. Even if that is really the way you should do it, and that’s how Buffet did it.
Investors and trustees are not going to buy it.
They are going to want to see more tech/AI/bells and whistles in the investment process and if you don’t have it the firm in the next meeting will.
Maybe that firm is Bridgewater.
The Connecticut-based hedge fund is developing a sophisticated machine-learning engine. This fund, a blend of various AI models, will execute investment strategies for a select group of clients.
Spearheading these efforts is Bridgewater's newly formed Artificial Investment Associate (AIA) Lab. Comprising 20 seasoned investors and machine-learning experts, the AIA Lab's mission is to transition Bridgewater's entire investment methodology to machine-learning techniques. The lab aims to replicate every facet of the investment process using AI, from analyzing global financial trends to formulating and testing investment theories through machine-learning models. Human oversight will maintain control over risk management, with a provision for a manual 'kill switch' if necessary.https://www.thewealthadvisor.com/article/bridgewater-assc-looks-future-new-automated-ai-fund
It’s OK. You don’t have to be Bridgewater overnight.
But in the near future fundamental analysis as an investment process is going to have to be ‘fundamental investing augmented by AI’.
You’re story is you’re still fundamental, but your analysts are assisted by AI. This is Phase 1 for your firm.
Then longer term, Phase 2, ‘fundamental analysis’ might mean training AI models on how to recognise good investment ideas for the long-term rather than actually picking the companies.
In other words, the Buffet of the future won’t read annual reports. He will train AI’s to think and invest like him.
Or, we don't need Buffet and we can just feed the AI all his annual letters so it can learn how he did it and figure it out. Maybe the AI will figure out a Buffet 2.0 that is even better.
In Phase 2 your investment process looks more like this.
Imagine a pension fund investor meeting where the Trustees say to the CIO:
‘Frank, the board wants to know why we are 40% invested in Treasuries? That seems strange. What is the rationale for this?’
And the CIO says:
‘Sure Dave. I’m not really sure why the AI does what it does, but it’s usually right. Our team has checked all the hardware and software and everything looks fine. We are using the most up to date software version of the AI model, checked all the inputs, and this is what the AI wants to do. It just loves Treasuries. In fact it wants to go 60% Treasuries, but we put up guardrails to limit it to 40% in any one asset class, and it went to the max.’.
That seems farfetched. Right.
But listen to Youtube commentators describing a chess match between Deepmind and Stockfish (two AI’s playing each other). Even though these commentators are chess masters themselves at times they don’t know why the AI’s are doing what they are doing. The AI’s are playing so many moves ahead and using strategies and attacks nobody has ever seen before that sometimes they just say “I don’t know why it did that. This is God Level Chess’.
Lee Sedol was the same way after losing to Deep Mind. Yes, it really hurt, but to his credit he wanted to learn the new moves and strategies the AI’s were innovating in this ancient game.
5 Steps to start the AI Transition.
Erik.. this is great. Thanks. Love the big picture stuff. But I know my firm. There is no way we are ever going to be able to transition to an AI firm even if we wanted to. It just seems too intimidating.
I understand. But I encourage you to remember the Montana Bear wisdom. You don’t need to outrun the bear, you just need to outrun the guy next to you.
90% of asset management business will not make the AI switch. Lots of asset management firms will have had a good run over the last 20-30 years, the CEO’s are rich, and so these firms or hedge funds will just shut down and become ‘family offices’. They will not try to make the transition. They will be eaten by the bear.
So you don’t need to become Bridgewater. You just need to be better than the 90% who are going to do nothing. Which means anything will be good. Just make a small step. Then another and take a long-term view. In 2 years you will be surprised at what you have achieved and be glad at the building blocks you have put in place.
Here are 5 steps to start the process.