arcticorangutan on Nostr: In 2017 I turned 18 and bought $1000 worth of Apple stock. I remember thinking I was ...

In 2017 I turned 18 and bought $1000 worth of Apple stock. I remember thinking I was a genius when my “investment” tripled in a relatively short period of time.

In my naïveté about markets, I used the opportunity to sell.

ChatGPT did the math and those shares, if I had held on and after accounting for stock splits, would be worth $350,000 today!

Except, in fact Chat GPT was hallucinating. The correct value today of $1000 invested in Apple Stock in 2007 is somewhere closer to $37,000, depending on the exact purchase date and whether one accounts for dividends. What ChatGPT seems to have gotten wrong is that it did account for stock splits, but it did so with a 2007 price that was already historically adjusted.

So this is a post about the power of compounding in investments and one lesson is that if you have a high conviction bet, you should not let your profits lure you into selling or trying to trade.

However, this is equally a post about remaining cautious about the output from LLMs. They present their results confidently and articulately and they can be extremely useful, but they often still get simple things wrong… resulting in potential orders of magnitude of errors.