How AI Is Changing The Investing Field

The Wimbledon got over not so long back. Mark the difference between the two pictures on the next page (what the centre court looks like at the end of the tournament, and what it looked like 20 years ago). Now, the wear and tear is only near the baseline, but back then, the grass was worn away near the net as well. If you go back a little more in time, the difference would have been starker.

What changed over the years? “Those changes (in the court) are evidence of a change in the game itself: from the dynamic serve-and-volley style of the 1970s and 80s to an increasingly static game dominated by powerful serves and returns from the baseline, a shift that began around the turn of the century,” writes Frank Jacobs in Big Think’s article, “Do these pictures prove that tennis is dead?” Not only have the courts changed, even the racquets have undergone transformation—from wood to metal to titanium.

Something similar happened in hockey, too, our national game, where we have barely won a medal in the past 50 years. India, and later India and Pakistan were kings of the field hockey heap from the 1920s to the 1970s. From 1928 to 1960, the Indian men’s hockey team remained unbeaten in the Olympics, winning six golds in a row. They were also the only team ever to win in the Olympics twice without conceding a single goal in the entire tournament. The dominance continued till the mid ’70s. The subcontinent teams’ fluid passes, superb dribbling and bewitching stick work ran circles around the Western teams, enthralling crowds worldwide.

And then suddenly, India and Pakistan became part of the also-rans for decades. Why?

The developed nations struck back: they changed the field of play. Literally. By introducing astroturf, or artificial grass. This technology innovation changed hockey forever.

Astroturf offered a flat terrain, allowing fast passes with precise ball control. Fast sprinting and extreme fitness became essential. No prizes for guessing which players were fitter: those from the developed nations!

Their superb fitness was the result of better training and nutrition. Also, video recordings, unheard of in India, were brought in to the play. Kinesiology, the science of movement, came in; time-release nutrition became order of the day.

At all the levels of the game, from grass to training to nutrition, science and tech took away the edge that the “traditional” players had enjoyed.

Bottom line: Once the playing field changes, the option of continuing to play by the old rules and with the old skill sets is gone. If you choose to do that, you will be left far, far behind.

What better example than Bjorn Borg, who is still the joint record holder for the most consecutive Wimbledon wins? But when he tried to make a comeback with his trusted old wooden racquet a few years later, he could not go past the first round of the tournament. The game had changed by then.

It is the same as an auto company saying today that they will stick to the traditional combustion engine and not look at electric vehicles or other green technologies. We all know how that story will end for any manufacturer who does that.

The Playing Field Is Changing

Advanced tech is doing to investment management what it did to sport. Traditionally, investment decisions were made only by the human mind. If you put your money in a fund or portfolio scheme, your “human” fund manager painstakingly analysed company and industry data in order to decide where to invest. Think Warren Buffett and Peter Lynch.

The traditionalists, which is still the bulk of the industry, still do this. They attend conference calls. They read annual reports (at least you hope they do!). Almost all investment management practices today, remain frozen in a 1940s-90s time warp. It’s all touchy-feely old school... still, and under-performing for most part.

But there is seismic change afoot.

Why Human-Only Model No Longer Works

A large part of what made the traditional model work lay in getting additional or unique information by meeting company managements. It was a game of information arbitrage. Large fund managers could sit in a closed room with company officials and get information.

I have done plenty of that myself. For decades I met companies, went around the countryside to visit steel manufacturing facilities in Jamshedpur and Tarapore or aluminium plants in Renukoot, auto-ancillary units on the outskirts of Chennai and Delhi NCR and pharma companies around Hyderabad. I visited umpteen offices and plants, trying to glean that extra bit of insight that would give me an edge.

Hand on heart, I would say that it was the most fun part of my job as an equity researcher!

However, the finest output is not about what is most fun for you individually but about what works the best... in this case, what results in an optimised portfolio.

The edge that fund managers had in getting information from companies has been regulated away across the world—information availability has been made uniform. Information arbitrage is gone! Earlier, conference calls after an earnings announcement used to be accessible only to a chosen few. Now all call transcripts have to be out in the public domain, by law.

Now the problem is something different—there is a surfeit of data which is humanly unmanageable. This is where machines come in.

To handle tonnes of data, you need advanced computing power for data-crunching. There are mathematical models that dispense insights at speeds unimaginable. They can analyse more securities and more data points in each of those securities than is possible even for large teams of humans.

Also, they can do this consistently and without bias, something that is impossible for human beings to do. Machines ‘learn’ quicker and better than humans. Crucially, machines admit mistakes and learn from them.

AI and machine learning (ML) are transforming portfolio management. An expertly constructed quant ML model can read millions of research papers, balance sheets, conference-call transcripts, social media feeds, and analyse auditor’s reports and management commentary. It can distinguish between good and bad accounting policies. It can granularly analyse ratios, in time series as well as in cross section, across thousands of companies. It can analyse reams of data, discern patterns and linkages among stocks across the world.

Of course, a machine needs to be programmed and trained. So, one quant model is not as good as another, but that’s another topic.

By Devina Mehra, Chairperson, Managing Director and Founder of First Global