In an interview with ETMarkets, Agarrwal who has over a decade of experience in public and private investing both in India and the US said: “We feed in fundamental data for all companies listed on the NSE and macro data like GDP, inflation, liquidity etc. since when (mostly early 2000s) these data points are available and then run millions of iterations based on proprietary frameworks.” Edited excerpts:
Can the AI system adapt to external changes such as COVID or a war situation?
The key feature of our AI/ML system is the ability to be dynamic and adapt to changing market conditions.
In addition, because they don’t possess any of the behavioural biases that plague human beings, machines are able to be ruthless with their decision-making; basically, they optimize for picking the portfolio that has the highest probability of maximizing returns.
Changing market conditions are of course communicated through changing prices, fundamental data, and macro prices.
We don’t feed in current events data because we think it is very difficult to separate noise from signals. For example – when Trump was tweeting about aluminium or China etc. there was no way to know if any of it would have any material impact.
Read Also: Role of artificial intelligence in investment
ETMarkets Smart Talk: Atanuu Agarrwal explains role of artificial intelligence in investment
Atanuu Agarrwal, Co-founder, Upside AI who has over a decade of experience in public and private investing both in India and the US said that fundamentals-based investing remains the domain of humans in India, but that will change over the next 5-10years.
How is risk factor mitigated or controlled by the system?
We have built-in multiple risk mitigation layers –
(i) As mentioned earlier, we do not use any leverage – no F&O, margin etc. Neither do we engage in any shorting of any kind. So, we are not going to have flash crashes, margin calls, etc.
(ii) We have dynamic volume-based filters that eliminate all illiquid counters – we do not want to be in the position of being stuck with some scrip that is continuously hitting circuit levels and we can’t get in and out of reasonably quickly, and
(iii) We restrict exposure to any signal company to 10-15% (depending on the product) at cost and any industry to 35%.
What role does Machine learning play in portfolio building?
I think it is important to understand the difference between AI and ML. AI is the ability of a machine to replace human intelligence and be able to perform and enhance human-like tasks. Robots, chatbots etc. are all examples of AI use cases.
ML is a subset of AI, where the machine has the ability to learn from data and draw inferences from it without explicit human assistance or rules.
We use ML because we are specifically looking for the machine to rely on objective data and eliminate human bias from investment decisions.
We feed in fundamental data for all companies listed on the NSE and macro data like GDP, inflation, liquidity etc. since when (mostly early 2000s) these data points are available and then run millions of iterations based on proprietary frameworks.
It’s simply not possible for a human being to crunch this amount of data and draw inferences from it.
So, to summarize, ML plays two roles –
(i) It eliminates human bias, and
(ii) It is able to parse and make decisions using millions of data points and iterations.
(Disclaimer: Recommendations, suggestions, views, and opinions given by the experts are their own. These do not represent the views of Economic Times)