Bard or ChatGPT: How far can they change your trading/investing life?

Chatbots can't be completely relied on to point you to the best investor to emulate. (Photo by Miguel Á. Padriñán/Pexels)

With Google set to unleash its very own Bard to take on Open AI’s ChatGPT, there’s talk once more about the disruption that is about to set in.

While it can crack entrance exams to Ivy League schools and for medical licensing, can these language models overturn the way trading or investing is done?

Since Bard is yet to be made available to all, we try with ChatGPT.

Also watch: Can a command get ChatGPT to break its own rules?

Coding

First, let’s start with trading algos. Some online demonstrations had shown ChatGPT generating codes. This had led people to imagine that they could request for a profitable algo, and have one pop out from the other end, freshly baked. ChatGPT isn’t there yet.

If an investor asks for a code to execute a simple trading strategy by providing certain indicators — for example, place a buy when the stock trades above its 200 day moving average (DMA) — ChatGPT can generate one. But, it won’t come without errors. To correct the errors, the user needs to do the following — run the code that ChatGPT provides; if there is an error, then feed the error back into ChatGPT; the AI will give a corrected code, which then needs to be entered into the trading software. Repeat this till all errors are eliminated. For simple codes, this is possible, but for more complex ones, this can be time-consuming.

According to Shreyas Bandi, a veteran trader and algo writer, if you have a clear idea of the system (a complex strategy) you want to build (including the values of trading indicators) then ChatGPT can help with coding the basic blocks. Then, the user needs to work on integrating these blocks for a more complex strategy. Or, if a user wants an outcome and wants to refine his/her code, the user can ask ChatGPT what can be done better. Overall, it can help a trader code more easily, but it cannot help a person become a better trader.

“It can help translate your idea into code, but your idea is what is going to make you money,” said Bandi, succinctly.

ChatGPT is also limited in testing a strategy and cannot be relied upon to be factually correct, according to Bandi. “While it has vast data, it does not have specific data. For example, historical NSE data to test for a simple opening range breakout strategy,” he said.

Quick summaries

Investors need to source and process a lot of data on their portfolio companies, and ChatGPT could be of great help here, said Shyam S. A long-term investor, who has grown his investment at over 15 percent compounded annual growth rate (CAGR) over 10 years, Shyam uses ChatGPT extensively for summarising investor call transcripts and research reports. He shared with Moneycontrol a shortened version of a concall transcript of specialty chemicals company. The ChatGPT-generated version captured the operating metrics, and the reasons behind the rise and fall of these variables this quarter, the company’s ongoing capex spends, and its future outlook in about 500 words.

He said that he also uses AI for a quick understanding of companies he may be interested in. For example, on the companies’ business models and revenue streams.

Who’s the boss?

To test some of its features, we asked a few questions.

First, can ChatGPT help a newbie figure out who the best investor is, to emulate his/her style? Not reliably.

When asked, it named those who have delivered long-term outperformance ― Warren Buffet (average annual return of 20.5 percent since 1965), Peter Lynch (29.2 percent between 1977 and 1990), George Soros (30 percent since 1969), Ray Dalio (21 percent over four decades) and Carl Icahn (24.7 percent since 1968). So, why is there no mention of Rakesh Jhunjhunwala though, who delivered an average annual return of 25-30 percent? It has to do with the way ChatGPT relies on data it has been trained on. If the data is biased, ChatGPT’s replies too, will be biased.

That said, it can argue in each investor’s favour, if you pit the merits of one against another. For example, when asked why Warren Buffett is a better investor than Jhunjhunwala, it pointed to Buffett’s long-term value investing style. When asked the other way, it pointed to Jhunjhunwala’s superior risk-taking ability. So, it comes down to how you frame the questions.

Stock suggestions

If anyone is worried about ChatGPT creating havoc by giving stock suggestions ― as it isn’t a SEBI-registered analyst or advisor ― rest assured. It does nothing of the sort. It is boringly proper in its replies. When asked for a recommendation, it responded saying that the stock market is unpredictable, that different stocks can be attractive at different points in time, and that it would be best to consult a financial advisor. When asked about a specific stock, it gave a similar response ― that choosing a stock depends on a person’s investment goals and risk tolerance, and that he/she should understand the risks associated with the company before making a decision.

Marketing material

In the stock market, there are many scamsters who promise unrealistic returns from their strategies or investing calls. Their one giveaway is poor content on their website and badly-worded marketing material. With the AI bots, they could plug this gap. They can now have readily-available, authentic-looking material for their bulk SMSes and social media posts.

For example, we asked for an SMS to advertise a trading call and it gave a short and compelling one, with a promise of great returns, clear entry and exit points, and all for free. It ended with a call to action: “Get in touch to learn more”.

Also read: The great chatbot race| Tech biggies battle it out

That said, what the capital market needs right now is a compelling argument for investors to stick with it and not migrate to fixed-income assets, which now promise higher returns. So, we asked ChatGPT to write a love note to an investor. It wasn’t impressive.

The best the bot could offer were a few half-hearted pleas ― “I want to ask you not to leave me behind”, “We’ve seen many ups and downs together….” and “Please don’t leave me for fixed income - I need you” — and some platitudes about equity giving great returns. No poetry, no data.