Can AI & ML Disrupt Investing? This Mumbai-based Startup Is Showing How

Upside AI is one of the first funds in India to use machine learning to make fundamental investing decisions. The company was founded on the belief that technology will make better decisions than humans over the long term since machines are unbiased and unemotional decision-makers. 

Founded in 2018 by Kanika Agarrwal, Nikhil Hooda and Atanuu Agarrwal, Mumbai-based Upside AI uses technology to understand, recognise, and buy companies that are not only fundamentally good businesses but are also in-demand stocks. After two years in development, in July 2019 Upside AI came out of beta to start offering its investment products under a SEBI registered PMS license. 

For this week’s startup column, we got in touch with one of the founders and Chief Investment Officer of Upside AI, Kanika Agarrwal to gain a more in-depth insight into how it drives AI and machine learning to provide fundamental investing.


Sign up for your weekly dose of what's up in emerging technology.

AIM: Tell us about your flagship products. 

Kanika: The flagship products essentially parse fundamental data of all stocks on the NSE, i.e. financial statements, price and valuation metrics, etc. The algorithms then, over millions of portfolio iterations, learn how to pick companies that are not only fundamentally good businesses but also in-demand stocks. Once it is done learning, it suggests a portfolio of stocks.

The Upside AI analysts then do manual checks on corporate governance to ensure that the companies are reporting financials accurately. We suggest a portfolio of 10-25 stocks to buy and then re-evaluate your portfolio every quarter.

Download our Mobile App

We currently run two products on the technology – Upside Multicap and Upside 250. Upside Multicap is market cap agnostic, and Upside 250 focuses on the top 250 stocks by market cap.

AIM: How is it different from other products in the market?

Kanika: The premise of Upside AI’s products stands on four legs: 

(a) eliminating human bias, so that calls are not based in market panic or euphoria

(b) focus only on the business performance of companies

(c) scan the market millions of times looking for investment ideas, and 

(d) dynamically adapt to the changing market conditions.

Differentiation from humans is clear – the use of a systemised rules-based approach to investing, which is not affected by bias and emotion. Further, because we use machine learning, the system is truly dynamic. It can learn and adapt to different market conditions and stock-pick based on (1) fundamentally good stocks (2) what the market thinks are good stocks. 

As far as other technology goes, today in India, tech is doing the basics, screeners, technical analysis, high-frequency trading etc. But the “holy grail” is learning true investing, which is the area we are focused on solving.

AIM: What are your innovative ways to use AI techniques? 

Kanika: The use of machine learning and AI techniques allows Upside AI to go through 50 million portfolio ideas every single quarter! This allows it to be truly dynamic and adjust according to changing market conditions in a manner which is simply not possible for human investors.

We were very clear from the beginning that we did not want to build a black box where we were not sure how the machine came to its conclusions. This is why we have not used neural networks, etc., to build the algorithms. So, the ML techniques we are using allow us to trace back the decision-making process. While we can’t go from front to back, we can work backwards with the tech we are using. 

Apart from the core algorithm, we have also built systems for pre-processing the data which allows for data clean up, checking for missing data, changing financial years, etc., which is a non-trivial process. In the post-processing, we are building a more systematic approach to corporate governance where the algorithm can flag issues based on multiple parameters for the analysts to examine. 

AIM: Can you tell us about upside AI’s core technology stack

Kanika: All 200,000 lines of code are in Java and Python that include the pre-processing and core machine learning algorithm. We have built our own servers to run our algorithms since they need to be powerful enough to handle the load. 

AIM: What does the future roadmap look like? 

Kanika: So far, we were structured as a PMS. A PMS is a specialised product that accepts investments of Rs 50L+ from high net worth individuals. We do not want to restrict this technology only to people who can invest Rs 50L+. 

Our retail investors should also have access to cutting edge technology that allows them to diversify away from humans. Therefore, we have opened up our research so you can invest as little as Rs 50,000 in products like ours. 

Over the next five years, we want to be able to reach 50,000 customers who can benefit from diversification into technology in their asset allocation. On the technology side, we will continue to be an equity shop but build out complementary products. For example, we are currently working on a macroeconomic model. The idea is to be able to attract best in class talent in machine learning that can keep us ahead of the curve as the country increasingly adopts technology in investing.

More Great AIM Stories

Ambika Choudhury
A Technical Journalist who loves writing about Machine Learning and Artificial Intelligence. A lover of music, writing and learning something out of the box.

AIM Upcoming Events

Conference, in-person (Bangalore)
Rising 2023 | Women in Tech Conference
16-17th Mar, 2023

Early Bird Passes expire on 10th Feb

Conference, in-person (Bangalore)
Data Engineering Summit (DES) 2023
27-28th Apr, 2023

3 Ways to Join our Community

Telegram group

Discover special offers, top stories, upcoming events, and more.

Discord Server

Stay Connected with a larger ecosystem of data science and ML Professionals

Subscribe to our Daily newsletter

Get our daily awesome stories & videos in your inbox

Top BI tools for Mainframes

Without BI, organisations will not be able to dominate with data-driven decision-making but focus on experiences, intuition, and gut feelings.