How ICICI Securities’ advisors built an in-house financial advisory platform using MATLAB

Amongst the largest retail brokerage and investment firms in India, ICICI Securities Limited is a robust and fast-growing business. Its financial advisors spend long hours with data and growth curves to be able to deliver clear and well-articulated ways to their clients, to enable them to reach their investment goals. These advisors help investors in meeting their investment needs- varying between extremely personal and long planned investments like retirement, children’s higher education, or purchase of a home or business. To be in a position to offer sound advice, the financial advisors also need to collect information about the investors’ financial history - current investments, savings, expected income, and expenditures.

Collating all this base information, analysing it and deducing investment muscle and psychometric factors such as risk preference, requirements and the best suited plans- is a long and tenuous process. Poring over spreadsheets, manually entering data about various schemes, plans and options- it took days of study and accurate calculations to get best suited investment information for clients. Despite a good financial background and technical knowledge, they lacked a robust system that would enable them to provide better financial advice to clients and as they are not coders, creating a platform that was completely customised for their needs would be a challenge. However, there was no way out if they were to grow and provide better financial advice to more clients.

Build, experiment and modify

The critical ask was that the team wanted to develop the platform in-house, so that they would be able to make rapid changes. Says Abhishake Mathur, Sr. Vice President, Investment Advisory & Customer Service, ICICI Securities,” Financial analysts and investment advisors typically do not have the programming skills to code effectively. Yet if they do not code themselves, it is difficult to experiment, develop, and back-test new methodologies.”

“In financial markets, you need more ideas to understand the problems, coming out with the optimal portfolio. A lot of problems today are pertaining to getting some new ideas, trying to see them in practice, experiment and implement,” he adds.

The teams needed a tool that made it easy to build, experiment and modify the strategies, without the support of complex, incomprehensible codes that need software experts to run.

There were many issues that needed resolution, specially to make it easy for financial advisors in the Indian context. Mathur explains with an example, “we speak of long term, and the need to have a diversified portfolio. But in Indian context, how many stocks does it really take to diversify your portfolio and remove the systematic risk. That is common knowledge, it’s been written so many times, but how do you really run a simulation?”

With the primary purpose of getting an easy to use tool with readily available libraries, ICICI Securities decided on MATLAB, Financial Toolbox, and Optimization Toolbox, to create a robust financial advisory platform for their financial advisors. “Even though our team of investment advisors had limited programming experience, MATLAB being an easy tool enabled us to develop and deploy a full-featured automated financial advisory platform,” Mathur says, speaking of the ease of usage of these complex technology tools.

The Investment Advisory team, headed by Mathur, created various scripts to help them develop capital market expectations (CME) and an asset-liability matching algorithm. MATLAB scripts were used to clean time-series data from short-term debt, long-term debt, gold, and equity markets. Using MATLAB and Financial Toolbox they developed algorithms to determine long-term correlations in the assets and identify a set of efficient portfolios with optimal asset locations based on a customer’s profile information.

They then developed a MATLAB script to create asset allocation strategies by goal-risk profile and time horizon based on Monte Carlo simulation. The team went on to create an asset-liability matching algorithm using Optimization Toolbox. This algorithm enables them to develop customized investment strategies based on a client’s specific situation. To meet the challenge of allowing the advisors to use these algorithms independently, standalone executable applications were created using MATLAB Compiler. With this, the teams could institutionalize the use of new methodology to create financial plans which could be delivered as a report to the customer. These could be made available on the Track&Act platform on ICICIdirect.com, where it is automatically updated to reflect portfolio performance, current cash flow, and progress toward the customer’s goals.

Benefits abound

Using MATLAB as its advisory engine, ICICI Securities launched their Robo Advisory Platform Track&Act. A major enhancement to the company’s existing financial planning and advisory practice, Track&Act matches assets to liabilities and dynamically monitors each customer’s target asset allocation, savings rate, and portfolio progress. The platform’s advisory engine automates previously manual processes, enabling the company to offer personalized financial planning to its customer base

Today, servicing a base that exceed 4.6 million investors, financial advisors at ICICI Securities have brought down the manual computational process drastically, as also increasing its accuracy phenomenally. As Mathur says, “In the past, advisors needed three hours or more to manually perform asset allocation based on a risk profile. The automated system we developed in MATLAB produces a more reliable, goal-based asset allocation almost immediately. We have been able to significantly improve the accuracy of suggested asset allocation strategies within financial plans based on clients’ goals, as well as the suggested allocation for their surplus funds.” The platform now allows the advisors to run millions of simulations to deliver advice that is specific to a client’s situation. “The robustness and scalability of the platform that we developed lay the foundation for rapidly growing our advisory and financial planning practice—something that would just not be possible if we were using spreadsheets or similar tools,” he adds.

The algorithms that have thus been developed for defining optimal investment portfolios matching a customer’s current assets and future cash flows has enabled ICICI securities, can be deployed as a robust and scalable platform with MATLAB Compiler. For the users, MATLAB has proved to be a very popular tool for scientific templating and increasingly the libraries available are very pertinent to the environment. They allow easier solutions for very complex financial problems. Since it offers intensive libraries and extensive support, the tools became very easy to adopt for the financial advisors. In fact, Mathur points out that it is very easy to upgrade if one is familiar with the MS xls tool for computation.

This adoption also made scaling up very much easier than before.” For the team to scale up, whatever program is being made could be distribute within the team. All the advisors can use the library to create the function pertinent to them,” he points out.

In fact for the problem of financial planning – which could be actually an asset liability mismatch problem, some of the tools available makes it very easy. “A financial plan for our customers, it is technically like solving a 400 plus variable problem, it helps to optimise the amount of assets that the customer has, to reach his goals. That is what we do with MATLAB daily and that’s how MATLAB makes it easy to run simulations - program it, feed the time frame of stocks and estimate the optimal size of the portfolio. I think it’s a great tool for people in this industry to get new ideas, experiment with them, refine and implement them,” Mathur says.

The ICICI Securities investment advisory has been created from ground zero, and MATLAB has been an integral part of the process of new product launches. “Today, we have about over 25000 customers who subscribe from investment advisory services. A lot of investment engines that provide these services to clients are run on MATLAB. It is a core component of creating these products, so most of the tools that we have created inhouse and a lot of technology that has been used- algorithms, computations, all can be attributed to the system,” he signs off.