Investing With A.I.

 

It's been 6 months since I asked ChatGBT to pick 10 equities for a hypothetical investment. It disavowed any expertise in the field, picked 10*, 5 of which were tech stocks, and my theoretical investment has since made me a theoretical 28%. Thinking that a good investment advisor would watch the markets and move the monies around from time to time, I asked it the same question again. It picked 8 of the original 10 picks, and dropped GOOGL and META in favor of JPMorgan and Home Depot. I adjusted my portfolio to remove the 2 and added some declared dividends and the newly favored equities. Now I guess I wait another 6 months to see what this gets me.

Since I'm already paying a boffin to manage my finances for me, it would probably be silly of me to fire him and follow the AIs advice on no record at all. Still, if I had, I'd have recovered all my losses from the Bidenomics "boom".

Managing an imaginary account like this is a bit more difficult than doing it for real. Since it's all imaginary, I have to make all the adjustments to the account myself, and Yahoo finance gets confused when I do a trade. In real life, I'd just call my broker, execute the buy and sell, and get a monthly statement showing the results. O.K. it's not THAT hard, but I'm lazy.

On 8/1/2023 8:31 PM, David Aitken wrote:

Managing an imaginary account is easy. It's when you have real money on the line that it gets emotional, and hard.

Well, yes. Any project involves work and money and when you eliminate the money part, it's easy to not get too attached to the project. Using AI like this is definitely an experiment with no precedence and so I'm not going to take this for a full ride until I'm convinced it won't blow up on the pad. Next year, maybe I'll try it on a small scale with real money. So far, though, it's working well "in the lab".

The other problem here is that I may not be able to "cross the same river twice" as GPT is itself an ongoing thing undergoing continuous changes. The 1/24/2023 version I started with is probably not the 7/23/23 version I just used to generate an update. GPT-4 is now the current thing, but GPT-0 is still available. Or is it now GPT-0.6?

An aerospace engineers approach to the problem. Excess caution -> Paralysis by analysis. At what point do I take the plunge? I'm thinking after 2 consecutive winning 6 month runs.

* Original picks: AAPL, MSFT, AMZN, GOOGL, FB, JNJ, PG, V, IVV, VTI. Don't recognize some of them? Go to Yahoo finance and type in those ticker Ids.