The ChatGPT stock portfolio tracker is now lagging the S&P. It was beating it but one of his holdings (Dollar Tree) got hit hard and dropped -17%. The portfolio is +1.9% since inception whereas SPY is +3.5%. God knows what Bitcoin is up during that time but it's much more.
Discussion
The thing is, any advantage AI gives you in the market, is going to be short-lived. Because eventually the prices will just reflect any asymmetric information advantage and the alpha will trend to zero.
You can bet that hedge funds are already using the most cutting edge AI models to squeeze out that edge. But it's an ever-moving target.
Yep, and I presume the dataset it's using is based on data up to 2021.
I suspect Ken Griffin at Citadel has already nailed the predictive algorithms for stock trading. Jim Simons too.
I don’t think they predict anything, they mostly just automated arbitrage.
Maybe they can detect large buyers and predict capital flows to front run?
But I doubt they really predict any market moving events at all, or stock performance, or technical analysis.
Suspect they are 100% concerned with high frequency arbitrage and detecting liquidity flows.
Could of course be wrong.
TBH I've never looked into the methodology. I just have a passive interest in it so follow the Twitter account. You're probably right. Seems more gimmicky than anything substantial.
I’m referring to Citadel and the big funds that stepped in when primary broker and investment bank prop desks were shut down in the aftermath of 2008.
No idea what GPT stock picker is doing? I would guess it’s based on value investing and not any sophisticated trading.
Again, I’m just making an educated guess. I don’t actually know.
Sorry, my mistake. I did wonder when you mentioned automated arbitrage. I have no idea on that either, though I do know Kenny’s fund is making bank
I read this 10-12 years ago. It’s an extremely niche book. But folk don’t really understand markets if you can’t wade through something like this. The market is the microstructure, not the macro stuff.
I couldn’t get through this book today, but at the time it was a revelation. Trading isn’t about predicting anything, it’s basically akin to fishing.
You just have to sit patiently and wait until you detect a big investor trying to move a lot of capital (either buying or selling) in tight liquidity and then you front run them in small volume for a razor thin premium. Get greedy and they stop the flow and you’re left holding someone elses baby. But most big funds will tolerate some traders leading them.
Eventually you can even end up communicating through the limit order book, using game theory.
The likes of Citadel automated the “fishing” and they scaled up by spreading it thinly across many instruments. AFAIK.
Turned it into an engineering problem.
https://www.amazon.co.uk/Arbitrage-market-microstructure-limit-order/dp/3838102622
That really is an extremely niche book. No digital version and zero reviews :)