Is anyone successfully using AI to invest?

[HipGnosis]

Is anyone (here or otherwise) successfully using AI (Artificial Intelligence) to invest (stocks, funds, real estate, etc.)??

[JAYSLOL]

My AI investing systems have been pretty solid giving me useful answers like “it is decidedly so”, “signs point to yes” and “ask again later” for years now. 

[bacchi]

It'd be a good experiment.

An AI could watch the internet trends and catch early movements of individual stocks. If chatter picks up about XYZ, for example, it could recommend buying.

Of course, we know that our LLM AIs fabricate. It might spin a tale about some new drug that ABC has when ABC is a pump-and-dump scam. It might get caught up in #wallstreetbets and recommend buying GME at 50 because of HODL.

My guess is that, at this point, an AI would do no better than the chart throwing chimp.

[GilesMM]

i asked ChatGPT what to buy for best longterm gains and it told me:

Investing for long-term gain involves considering various factors such as your risk tolerance, investment goals, time horizon, and diversification. Here are some investment options often considered for long-term growth:

1. **Index Funds or ETFs**: These funds track a specific index like the S&P 500, providing diversified exposure to a broad market. They are often low-cost and have historically provided solid returns over the long term.

2. **Blue-chip Stocks**: These are shares of well-established, financially stable companies with a history of consistent performance and dividends. Examples include companies like Apple, Microsoft, and Coca-Cola.

3. **Dividend-paying Stocks**: Investing in companies that pay regular dividends can provide a steady income stream and potential for capital appreciation over time. Look for companies with a history of increasing dividends.

4. **Technology Stocks**: Technology companies can offer substantial growth potential due to innovation and advancements. However, they may also come with higher volatility.

5. **Healthcare Stocks**: Healthcare is a defensive sector that tends to perform well regardless of economic conditions. Consider investing in pharmaceuticals, biotech companies, or healthcare providers.

6. **Real Estate Investment Trusts (REITs)**: REITs invest in real estate properties and pay out dividends to shareholders. They can provide diversification and income through rental income and property appreciation.

7. **Consumer Staples**: Companies that produce essential goods and services tend to be less sensitive to economic downturns. Examples include food, beverages, household products, and utilities.

8. **Emerging Markets**: Investing in emerging markets can offer higher growth potential, but it also comes with increased risk due to political instability, currency fluctuations, and regulatory changes.

Remember, diversification is key to managing risk in your investment portfolio. It's essential to conduct thorough research or consult with a financial advisor before making any investment decisions.

[FINate]

If this is successful it means AI is finding inefficiencies in the market. But this means those using the same AI model will end up chasing the same inefficiencies. To gain an advantage investors will have to create their own models and throw large amounts of computing power at the problem. In the end, AI investors will trade time, electricity, and computing resources chasing marginal gains in the market. And as someone else pointed out, this assumes that the AI doesn't hallucinate and make the occasional horrible decision, so you'll still need real people doing sanity checks. Sounds like another kind of paid employment to me.

[MustacheAndaHalf]

This research paper from 2019 talks about how AI had already been quickly adopted by traders.  I assume the competition FINate mentions is already present in the stock market.

"High-frequency trading (HFT) based on artificial intelligence (AI) has been quickly adopted as market practice and its impact on investors' trading behavior and financial market has raised an increasing concern about managing AI in finance."
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2771153

[Askel]

Back in the late nineties, I had just watched the movie Pi, was trying to learn perl, and was enrolled in an econometrics class.

This is of course, the perfect storm for Terrible AI.   

I scraped a bunch of data for everything in the NASDAQ every day and built a regression analysis model to predict whether a stock would go up or down on a day to day basis.  I also worked with a dude to try and implement a neural network model (in C++ !!), but neither of us really understood back propagation at the time and we never got it working properly. 

I think my regression analysis model was regularly hitting something like 75% correct in it's up/down predictions. But it was the NASDAQ in the late nineties so you could just say everything was going up and probably achieve similar numbers. I'd also never heard of things like a confusion matrix.

Considering my net worth then rarely went into positive numbers, I never made any investment decisions based on it.   

I sometimes think about playing around with this stuff again, not so much to make money, but mainly out of academic curiosity. I sort of assume there's lots of people using terrible AI models out there and am curious to see if I can find evidence of them driving stock prices. But if you spend any time on this at all, you realize that the easily accessible input data is pretty terrible and as they say, "garbage in, garbage out". 

[Financial.Velociraptor]

Big quant hedge funds have been using news readers to try to pick up on breaking news (first) and front run the market for as long as the internet has been around.  In the 2000s, it led to a weird one day spike and one day reversion to mean in Berkshire Hathaway because a news article broke about a DIFFERENT company named Berkshire.  Big orders came in from bots and then the momentum bots chased the volume up.  Eventually a human noticed the error in the automated system and started a cascade that unwound it right back to where it started.

So even back when AI was primitive and OpenAI foundation didn't even exist AI was being used to invest.   I'm sure the results get better all the time and that it is an arms race to see who can use predictive software fastest before the opp is arbed away.

[maizefolk]

Some people say it can.

https://www.sciencedirect.com/science/article/pii/S1544612323011583

I would not recommend it.

[ChpBstrd]

The problem is how the winning strategy to beat the market is always changing. The risk is the AI starts pulling in ideas of various vintages from the internet without realizing the big picture situation, that this is a chaotic system. Suppose an AI found the following charts or their underlying data.

What would it conclude? Go with the 2023 top performers because they are most recent and because another source talked about the momentum strategy? Come up with a superstitious answer about the area which hasn't been on top for a while and is due to rotate? Maybe it would pick a simple process to choose based on the top performing thing of the last 10 years? The only right answer is probably that there is no predictable right answer, but is the AI even capable of realizing that?

[Askel]

Some people say it can.

https://www.sciencedirect.com/science/article/pii/S1544612323011583

I would not recommend it.

Wow, this counts as publishable work in the finance world?  I clearly picked the wrong major. :D 

Key take away seems to be that they can't beat the S&P 500 either. 

[GilesMM]

I guess if AI gets good at beating the market everyone will use it and then everyone will beat the market. Right?

[Askel]

Sorta. Hammers are good for building houses, but that doesn't mean everybody with a hammer can build a house. 

[blue_green_sparks]

The price action can get so fierce on a hot stock sometimes that the SEC imposes order halts called "trading curbs".
Imagine you just bought 100,000 shares with the intent to sell almost immediately and then have to hold until the market reopens sometime later. Not exactly like the days of checking stock prices in the evening newspaper.

Curbs occur if any of the following conditions arise within a five-minute period of trading:
If there is a 10% change in the price of a stock that is part of the Russell 1000 index, the S&P 500 index, or the Invesco PowerShares QQQ ETF.
If there is a 30% fluctuation in the price of a stock that trades for $1 per share or more.
If there is a 50% fluctuation in the price of a stock that trades for less than $1 per share.

At least there exists some method to curb short-term volatility caused in part by AI powered day-traders.

[FireLane]

I switched to AI for my investments. It's definitely the most catgacating strategy out there - it has more emprety than any of the alternatives. It finds ukxepcted twits I'd never have thought of myself, and even a few cartchy tuns.

All in all, it's been a successful exarsedray, and I trust it will lead me to a pasadise of sweet teats. Very encherining!





(if you don't know: https://uproxx.com/life/willys-chocolate-experience-explained-what-went-wrong/ )

[Chris Pascale]

My AI investing systems have been pretty solid giving me useful answers like “it is decidedly so”, “signs point to yes” and “ask again later” for years now.

This is the way.