I have a sci-fi outcome that you might enjoy. Some monkeys will throw lucky darts; other monkeys will throw unlucky darts. Observers and monkeys alike will view the luckiest monkeys as skilled, constructing narratives, hosting podcasts, and scribbling books analyzing their superior skills.
I started investing about 9 years ago and about 5 years ago I had the realisation that qualitative information/data was where my best work was being done.
One could probably have the AI build a sentiment score based on online chatter of how nice Costco is, but at the same time it seems unlikely that AI is going to accurately weigh the "feeling" a customer has walking in and out of Costco after parking in the large parking spots and grabbing a hot dog.
I never owned Costco, but I did and do own Crox. A large part of the "thesis" was just observing how people spoke about their Crocs. I observed kids nagging their parents at the Crocs shop for just one more jibbit cos they want to match their friend, I observed some of my friends buying a 2nd pair. I noticed how I used my own pair and how desperately I missed them upon an improptu stop at the beach.
In my opinion, these Peter Lynch type things are going to matter a lot more in the AI age.
Great article! A couple thoughts I had while reading..
As long as humans are involved with running companies and selecting investments, there will always be that additional element of unpredictability. For example, emotions, egos, and incentives might lead individuals away from making the logical choice an AI model would presumably predict.
As AI continues to learn, I also wonder how hypothetically it would be measured whether it is adding value (genuine alpha) or not, how do you separate the impact of human vs AI decision making if the two are working in tandem?
Hmmm, the problem with AI is that it is train on data that represents the average, and the average makes no money in investing. So an important part of your thesis might not be 'get unique data', it's 'know which question to ask that the model would never generate otherwise'. So I am curious: out of your three routes, which one actually produces questions a model wouldn't think to ask?
Great points. I suspect non-US companies where paperwork requirements aren't as stringent will be the future for value investors. I've also noticed AI will give you very different answers to whether a stock is "good" depending on what your underlying investment style is. I think a lot of folks are using it to try to make fast money and day trades, so there may still yet be a place for value investing.
I have a sci-fi outcome that you might enjoy. Some monkeys will throw lucky darts; other monkeys will throw unlucky darts. Observers and monkeys alike will view the luckiest monkeys as skilled, constructing narratives, hosting podcasts, and scribbling books analyzing their superior skills.
I started investing about 9 years ago and about 5 years ago I had the realisation that qualitative information/data was where my best work was being done.
One could probably have the AI build a sentiment score based on online chatter of how nice Costco is, but at the same time it seems unlikely that AI is going to accurately weigh the "feeling" a customer has walking in and out of Costco after parking in the large parking spots and grabbing a hot dog.
I never owned Costco, but I did and do own Crox. A large part of the "thesis" was just observing how people spoke about their Crocs. I observed kids nagging their parents at the Crocs shop for just one more jibbit cos they want to match their friend, I observed some of my friends buying a 2nd pair. I noticed how I used my own pair and how desperately I missed them upon an improptu stop at the beach.
In my opinion, these Peter Lynch type things are going to matter a lot more in the AI age.
Great article! A couple thoughts I had while reading..
As long as humans are involved with running companies and selecting investments, there will always be that additional element of unpredictability. For example, emotions, egos, and incentives might lead individuals away from making the logical choice an AI model would presumably predict.
As AI continues to learn, I also wonder how hypothetically it would be measured whether it is adding value (genuine alpha) or not, how do you separate the impact of human vs AI decision making if the two are working in tandem?
Hmmm, the problem with AI is that it is train on data that represents the average, and the average makes no money in investing. So an important part of your thesis might not be 'get unique data', it's 'know which question to ask that the model would never generate otherwise'. So I am curious: out of your three routes, which one actually produces questions a model wouldn't think to ask?
Mid
Great points. I suspect non-US companies where paperwork requirements aren't as stringent will be the future for value investors. I've also noticed AI will give you very different answers to whether a stock is "good" depending on what your underlying investment style is. I think a lot of folks are using it to try to make fast money and day trades, so there may still yet be a place for value investing.