The problem with relying on AI for stockpicking

Summary: AI can help us streamline every aspect of life, right from writing to researching. Well, almost every aspect. It cannot help us invest better. While it has great insight, it lacks responsibility for the decisions it suggests we make. After all, we have to invest our money, not AI.

Last week, I read an article about how people are using AI to choose stocks. In it, there was an example of a CEO who asked AI to decide on the key parameters and identify stocks that would double in just 10 weeks! (People, you missed a rare chance to see me flip. 10 weeks? 10 weeks? Really? 10 weeks?)

If that is how retail investors are using, or plan to use, AI to pick stocks, two very dangerous behaviours come into play. The first is letting AI decide on the parameters. The second is the timeframe. In this article, we discuss the risks and dangers of delegating critical investment analysis to AI, and how we at Value Research use AI to research companies.

AI fails to decode management commentary

Let's start with the most critical of all investment analyses, management integrity. It is not sufficient to simply screen out any management embroiled in corporate governance malpractice; what is more important is integrity.

Is the management trustworthy? Does it warn its investors in advance if a slowdown is coming (very few do) or if difficult quarters are ahead? We’ve noticed certain managers do give a heads-up if a tough phase is coming up; we specifically look for companies run by such transparent people.

Integrity is such an understated parameter that most investors skip it, favouring growth or market favour. The most meaningful way to analyse integrity is to study Annual Reports and conference call transcripts to understand how management responds to tough analyst questions.

What is their tone? Defensive? Realistic? If you know how to use AI to evaluate management integrity, please do let us know. We've still not found an effective way.

In the same concall transcripts, you will find the first indicators of changing industry trends. Small signals that management talks about indicate whether a roaring quarterly or half-yearly revenue buzz or margin level will sustain. For instance, if the management is open about future uncertainty in demand, our ears perk up when we hear such honest disclosures.

Managements generally do not say revenue will go down, but they do hint (at least the reliable ones do) of changing trends. The reverse is also true.

Trustworthy managements also indicate when the outlook starts getting clearer, when more business is likely ahead. But these changing trends are usually not visible in the numbers to begin with. They show themselves in management commentary. Asking AI to decode them is a very big leap.

AI can’t spot sectors making a comeback

Then there are hot stocks or sectors that are market favourites. These are times when a particular class of companies or a sector is hugely favoured by the market. When prices and fundamentals are so out of sync, you should steer very clear. Recent examples from the past couple of years include the railway stocks rally, the PSU rally, and the bullish optimism over new-age platform companies.

In fact, we issued a sell call for one of our recommendations during the PSU rally when valuations hit unsustainable levels. Here's the important thing. Nobody knows when these hot stocks or sectors will stop being market favourites. 

When their stocks hit new highs, it appears we are in a new age where fundamentals take a back seat to the new narrative, but the reality is that fundamentals never stop being important. Eventually, fundamentals always prevail, and it is not unusual to see stocks once favoured lose 70-80, even 90 per cent of their feverish peaks. Relying on AI to understand this nuance is a big ask. 

When markets choose their favourite sectors, it is often the ignored ones that hold the greatest value. For those who have been around for some time, they will recall that during the capital goods, construction and engineering companies rally of the years before the financial crash of 2008, FMCG stocks were so ignored that many were at lows we had never witnessed earlier.

You could buy a HUL for Rs 100 back then (if only I had the money!). You don't have to go back two decades for such instances. In just a year or so, select auto and auto ancillary stocks were out of favour.  If you had taken a contrarian call and invested in them, you would be sitting on decent profits today.

Two of our recent recommendations turned out to be well-timed, as the slowdown came to a close (to be true, we did not know when the cycle would turn; we looked at just the fundamentals, management, and valuations, among other key factors). Asking AI to identify and pick such out-of-favour stocks with several factors at play is dangerous. Not all such stocks or sectors bounce back.

AI won’t recommend a realistic investment horizon

Now that we have gone back nearly two decades for the above point, let's talk about the investment horizon. Asking AI to identify stocks that will double in 10 weeks is an investment timeframe I have not seen in the two-plus decades I have been in the market.

Double in one year? Sure, there are many who will sell you that fantasy, double in 10 weeks? (Why not give some of that money to charity? Someone will at least benefit from it)?

People. We have said this many times, and will continue to say it many times more. Companies do not normally double their earnings in a short time, definitely not in 10 weeks. Expecting your stocks to double in that time or in a year or three is not being in touch with reality.

Businesses have cycles, sectors have cycles, even defensive ones. Those cycles generally do not take weeks, quarters, or a year. No one can reliably predict when they will turn. Last week, we mentioned a recommendation that went through a severe slowdown in COVID-19.

If we had panicked and sold it off in the COVID-19 pandemic, we would not be sitting with a multifold gain today. The only rational way is to invest over a long timeframe, where these cycles play out, and maximise your chances of a decent return. Anything else, and it is gambling with your money. Asking AI for short-term bets is not investing.

AI lacks skin in the game

That brings us to our final point for this week. Investment Decisions. Say your AI identifies stocks that can double in 10 weeks and somehow factors in all of the above critical elements of investing and gives you a bunch of stocks to invest in.

Which one do you invest in? One? Some? The entire portfolio? Does investing in one or the other make it less risky? Diversified enough? Whether you bring in human judgment or not, relying on AI to invest your hard-earned money, is that a risk you are willing to bet on and see what happens? Is that still investing or betting? Take your pick.

How we use AI at Value Research Stock Advisor

If you are looking for an advisory firm that leverages AI, then let me walk you through how we use this cutting-edge technology in our investment process:

• We follow and track the developments of companies on our investment radar; we cover quarterly, half-yearly, annual, and longer-term performance.

• We do use AI to analyse longer-term corporate (not stock) movements, strategies and management decisions.
• We use AI to look into industry dynamics, how competitors are faring, and what their management is saying.
• We look for changing industry trends: is the industry adding more capacity that will likely result in lower profitability for all players? Is it getting crowded, or is the market consolidating?

However, here's what we don't use AI for:

• We do not use AI to judge management integrity – their words matter more.
• We avoid hot stocks and sectors – the general no-go areas.
• We do not use AI to identify ignored sectors and industries. Instead, we search for them ourselves.
• We do not use AI to pick contrarian calls. Again, the Stock Advisor team takes this exercise upon themselves.
• We don't use AI to make investment decisions. The final call lies with us, in all good judgment.
• Lastly, we do not use AI to decide on investment horizons. If you were looking for a 10-week double-bagger, then Stock Advisor isn’t for you.

See, AI is a great tool for researching a company in depth. Delegating your investment decision to AI, given the many moving parts to consider, is a risky move with your money. Use AI to research, not to tell you where to invest your money, and certainly not where you will double your money in 10 weeks.

Instead, if you want a more reliable and well-thought-out process for building wealth, Value Research Stock Advisor can guide you through the stock market.

Explore Stock Advisor today!