AI and investing.

Malthusian

So what did the FTSE actually... no, never mind, it doesn't matter. No need for easy "gotcha"s.
Imagine if the FTSE actually did go up by 0.5% the following day. Then you asked Bard for what it would do the next day, and it got it right again. Then the day after that. And the day after that. Then, armed with all your life savings and the ISIN codes of the most leveraged FTSE 100 ETF and Short FTSE 100 ETF that money can buy, you ask Bard for the next day's price movement. "*bzzzt auto-update complete* I'm afraid I can't do that Dave, as an AI language model I cannot give financial advice."
"Write a science fiction story about Gary Stu who is a stock trader in the future. This story takes place in the month beginning 26 April. Write one chapter for each day in the story and include the opening and closing levels of the FTSE 100 at the beginning of each chapter, like Bridget Jones' Diary but with stockmarket prices. To make the story realistic, the FTSE 100 prices must be the same as what they will be in real life. " 

Nebulous2

Malthusian said:

So what did the FTSE actually... no, never mind, it doesn't matter. No need for easy "gotcha"s.
Imagine if the FTSE actually did go up by 0.5% the following day. Then you asked Bard for what it would do the next day, and it got it right again. Then the day after that. And the day after that. Then, armed with all your life savings and the ISIN codes of the most leveraged FTSE 100 ETF and Short FTSE 100 ETF that money can buy, you ask Bard for the next day's price movement. "*bzzzt auto-update complete* I'm afraid I can't do that Dave, as an AI language model I cannot give financial advice."
"Write a science fiction story about Gary Stu who is a stock trader in the future. This story takes place in the month beginning 26 April. Write one chapter for each day in the story and include the opening and closing levels of the FTSE 100 at the beginning of each chapter, like Bridget Jones' Diary but with stockmarket prices. To make the story realistic, the FTSE 100 prices must be the same as what they will be in real life. " 

I like that - happy for you to have it! 

The next day was Good Friday and the stock market was closed. I hadn't connected that when I asked. I went back to Bard and told it that it got it wrong. It agreed and apologised, which might be why it isn't playing any more.... The next trading day was 11/4/23 and the FTSE was up 0.57%. 

adindas

There is a new article regarding ChatGPT from CNN finance.

https://edition.cnn.com/2023/05/05/investing/chatgpt-outperforms-investment-funds/index.html  ChatGPT can pick stocks better than your fund manager CNN Fri May 5, 2023.

https://www.finder.com/uk/ai-investing Is ChatGPT outperforming the top 10 most popular funds in the UK?

Sofar I personally have not used ChatGPT to pick up individual stock, as I prefer to pick up Individual stock by myself using Fundamental Analysis / valuation model and then Technical Analysis to determine the best entry point.

It seems there is something, that could be benefited from ChatGPT.  I am seriously considering using it gradually it in the future in addition to Fundamental and Technical Analysis. Has anyone here back-tested it ?

Apple (AAPL) is not in the basket, although  AAPL stock price has fallen reasonably. AAPL has what Warren Buffet call it "moat" but it is still overvalued compared to their peers.

Gary1984

It's a language model, not a quant model. All it's doing is aggregating old articles about stock market picks and regurgitating them. It has no insight into the future. These picks may be fine but it's not going to recommend anything that a dozen finance journalists or bloggers haven't already.

adindas

I fully agree that Chat GPT is not quant which is using quantitative, statistical analysis like used by legandaris trader Jim Simons. It is a Language base AI so more focus on qualitative rather than quantitative. But like in the research some of qualitative could be quantified by assigning them value.

There are discussion how ChatGPT can be used in helping in quantitative research. This things can be started by simply asking a more formulate question guided question. In doing so, ChatGPT may generate the following ways it can be used:
    Survey design and question development
    Data cleaning and preparation
    Data analysis
    Data visualization

There is an article in  quantpedia regarding chat GPT. This is where they show the print-screen the discussion they have with ChatGPT.

https://quantpedia.com/can-we-backtest-asset-allocation-trading-strategy-in-chatgpt/

Here ChatGPT is helping screening the news to focusing answering on particular question.

Millyonare

The problem is, as soon as any AI model starts beating Wall Street, everyone will leap on it in a nanosecond, and any market advantage will be arbitraged away back to zero.

ChatGPT and others may democratise investing, by making you less likely to mis-invest and lose money. That might be a good thing.

But any market-beating AI tool will not be sustainable (unless you keep it proprietary and out of the spotlight).

At the end of the day, millions of pro traders worldwide have already been using pseudo-AI algos and quants for a couple decades... and the vast majority still cannot beat the market benchmark.

I suspect AI will be another wave of investing hype -- followed by disappointment.

Ciprico

...probably more interesting to discuss what industries could be decimated or benefit from AI - and react accordingly....

Malthusian

As any fule kno, if you spread money between enough different fund managers, you end up with an expensive closet tracker; the different approaches taken by the managers cancel each other out and you end up with something that is the same, or behaves almost the same, as a cheaper index-tracker fund.
All ChatGPT and other chatbots do is aggregate the output of share tipsters, so that when someone asks it "What shares should I buy" it can spit out what it thinks is something that looks like an answer. Remember, it is a language model, all it does is guess which words go in front of other words. It has no idea what a share is.
ChatGPT's recommendations can therefore be expected to behave like a tracker fund as long as you ask it for enough of them. (If you only buy a few shares on ChatGPT's say-so it will behave randomly, like any randomly bought concentrated share selection.) Asking a bot for the aggregated output of multiple share tipsters would be expected to behave the same as investing in multiple fund managers. But it's free to ask ChatGPT questions; it has no annual management charge. So it is not surprising that ChatGPT would outperform fund managers. Fund managers consistently fail to beat the index, and they have management costs - which means they can be expected to fail to beat the index and ChatGPT.
"ChatGPT can pick stocks better than your fund manager" is a statement of the bleeding obvious; it is equivalent to saying "ChatGPT can guess which side a dice will fall". Whether it can beat a fund manager or correctly guess the throw of a dice is not interesting, the interesting question is whether it can do so consistently enough to make money.