My New Friend, Claude

A reader left a link to a short post from Felix Salmon provocatively titled How AI could end the ETF boom. Ok, you've got my attention but, is this actually something new? A few years ago Schwab starting offering thematic baskets of stocks and if you search on Google you can find several similar products from other companies. 

It's safe to say that at least some of these products use some form of technology to assemble and then maintain these baskets. As AI evolves, some may already be using AI and maybe others soon will. Salmon's post seems more like a report on an evolutionary step not a revolutionary step. 

I played around a little bit with Claude AI from Anthropic. This link may have the output, it is the link I see but I don't know if it will show the results of my inquiry. First I asked it to suggest an all-weather portfolio that included managed futures. It pretty much spat out Ray Dalio's all-weather with a 15% allocation to managed futures. 

Then I asked it what it thought about including global macro and if it thought it was a good idea, how much should go to global macro. It suggested the following.

Claude AI suggested QAI for global macro and I used EBSIX in my version. The rest of my version uses names we regularly use for blogging purposes. BKLN allows us to avoid interest rate risk for this backtest. 

As we often see, a decent portfolio strategy this time from AI that can be improved upon. 

Then I asked it to Build a basket of global publicly traded stock exchange companies which may not be the first thing people would think to ask but is a niche I have been very interested in for probably 20 years. Again, I don't know if that link will work. Claude spat out a list of publicly traded exchange stocks from the around the world but it also included one or two private companies. It left out the New Zealand Stock Exchange. 

"What about NZX Limited? Did you leave any others out besides NZX?" It told me I was right and found a few others in addition to NZX. This was as specific as it got with construction;

Portfolio Allocation Strategy

• Core holdings (40-50%): ICE, LSEG, Deutsche Börse, HKEX
• Growth component (25-30%): Nasdaq, SGX, B3, NSE
• Value positions (20-25%): CME, Euronext, ASX, TMX
• Consider equal-weighting to avoid overconcentration in largest players

Most of these are in Testfol.io.

There are some interesting numbers in the Claude basket related to volatility and Sharpe Ratio. Both exchange portfolios held up much better in 2022.

I'm not saying AI's role is nothing. It will evolve to add some amount of value, maybe a lot of value but I think it is too early to know. I have mentioned before that Matthew Tuttle includes AI queries in his daily email so I am guessing he believes it is farther along than I do and he may be right. Whatever the current state of AI in the portfolio construction process, I would encourage advisors to engage with AI at least a little. 

That sentiment is similar to Bitcoin and the cryptos. I think an advisor should have something to say about Bitcoin when a client asks and I think an advisor should have something to say about AI when a client asks. 

Portfolio managers have been using computer based screening forever so AI doesn't seem that different conceptually, all the better if AI screening is a big improvement over more common screening methods. 

Circling back, I took a couple of names out of the Portfolio 1 backtest to have it go back further and got results that were a little more interesting. 

The information, analysis and opinions expressed herein reflect our judgment and opinions as of the date of writing and are subject to change at any time without notice. They are not intended to constitute legal, tax, securities or investment advice or a recommended course of action in any given situation.