Arnab Pandey, Vice President - Raymond James
Arnab Pandey is the Senior Analyst of Fixed Income Derivatives and FX at Raymond James Financial.
Updated
On this page
Updated November 2026. This was first published on 22 November 2022. The idea is the same. Any number below is what the original piece said then. It is not a new count. This is a record of a conversation from that day. No new quotes were added.
Key takeaways
- Arnab Pandey is the Senior Analyst of Fixed Income Derivatives and FX at Raymond James Financial.
- He did his bachelor’s in engineering specializing in Applied Mechanics and Biomedical Sciences.
- He found out his passion for numbers when he got the opportunity to study statistics online from a course offered by Harvard University.
Arnab Pandey is the Senior Analyst of Fixed Income Derivatives and FX at Raymond James Financial. He did his bachelor’s in engineering specializing in Applied Mechanics and Biomedical Sciences. He found out his passion for numbers when he got the opportunity to study statistics online from a course offered by Harvard University. He later did his MBA with a concentration in Finance from UT and an executive specialization in valuation from Harvard Business School.
He is a passionate data analyst, and his love towards statistics and technology has led him to research and learn about unique investment strategies and quantitative analysis. He was awarded the ‘Best in Business’ honor during his MBA studies by the Beta Gamma Sigma Honor Society. He has led several global teams to solve some of the biggest economic challenges around the world. One of his research on India’s socio-economic development is supported by 35 nations globally and has appeared in national newspapers in India. He has significant exposure in the field of Private Equity, Buy-side Investment banking, capital markets, and compliance. Apart from his work in the capital markets, he is a nationally awarded debater (Parliamentary Award by the Government of India), an internationally acclaimed researcher, and a passionate blogger. He is also a columnist in various business magazines.
We discuss with him about capital market and how technology is changing the landscape.
Technology is now closely entwined in the business of capital markets than ever before. The COVID-19 pandemic accelerated the digital and cloud transformation in the finance sector. What are the major challenges you believe this sector is facing due to the sudden need for digital transformation and cloud platform-based delivery models?
I believe technology has always played a crucial role in capital markets since its inception. But the key factor to understand here is the rate of technological acceptance in the financial sector, as technology evolved through the years. If you analyze, you will see that it has been rather very slow. COVID-19 has literally forced institutions to re-think digitalization strategies without providing a lot of time to think about them. We often see institutions struggling due to that. While a few institutions successfully implemented digital changes, many of the financial institutions have faced problems due to the inheritance of complex infrastructure due to M&A activities. That literally raises the debate of systems integrations – legacy versus new technology. Then comes the problem of the right talent to lead the digitalization without increasing the cost of setup at higher levels. A study by BDO states that about 70% of financial services executives feel that lack of proper training and the required skill sets are the biggest challenge to a new digital initiative.
Due to the size of the banking/financial services business, the change takes time. It is not just about the fundamental technology that lags behind but also about the company’s digital culture. A lot of times, we see that departments in financial institutions work in silos. Each department has its own way of using a platform or system, they have their own procedure that might not be optimal at all. they still want to use it. There is no centralized mechanism through which you can educate people to use systems, especially when employees are too comfortable using the legacy systems. training them with motivation becomes extremely difficult. A recent study by the Boston Consulting Group explains that more than 80% of the digital transformation initiatives fail due to a lack of digital culture.
We see that teams and the management in the financial industry are forced to focus on the short-term goals rather than long-term goals due to increased regulations, cyberattacks, etc. Teams are working towards mending the problem they are facing today rather than thinking of problems that might occur down the line in 5 years. The attitude of “why fix something when it is not broken” is prevalent across the sector. Due to this thought process and the fact that digitalization requires serious investments whose ROI might not be visible in the short term, digitization tends to trump digitalization.
With the ‘work from home’ trend due to COVID-19, the risk of securing social media communications, the security of the IT infrastructure and data, and preventing rogue trading are the three most important challenges the financial services industry is facing. Tens and thousands of networked computers are connected in a banking portfolio, and maintaining security on such a massive scale is very challenging. Thanks to the flexibility and scalability of SaaS solutions, which provides automation and full cloud enablement to protect almost everything.
In 2021, as financial entities decide on a short-to-medium term business strategy, technology investment decisions will hinge on multiple considerations. What will provide the greatest impact at both inter-and intra-firm levels? What will best address each entity’s technical state and organizational culture?
In my opinion, even before talking about short or medium-term business strategy revolving about technology investments in the financial sector, firms have to understand that they are going in for the long-term gain. As I mentioned earlier, most of these technology projects’ ROI might not be noticeable in the short term. One has to project the future cash flows for a significant period of time to understand the net present value of these projects. Most of the time, these are positive values. People have to understand that digitization is not and will never be equivalent to digitalization.
Then comes the most crucial part for any company to succeed in digital transformation, and that is establishing a digital culture. To achieve this, the first step could be to simplify the legacy systems and start opting for artificial intelligence, process automation, robotics technology, SaaS infrastructure, and data mining models that can read through unstructured pieces of data. I know that the industry right now is pretty far from all this. change is coming, and we all have to embrace it; otherwise, we might become extinct.
Digital culture in an company does not really mean just having digital products, services, and interactions via digital platforms. it also means empowering employees at all levels (especially operations) with technology. Empowering employees with the right technology, knowledge, and support will lead to a change in technological acceptance, boost morale and enhance communication and interaction both internally (to employees) and externally (to customers). This will lead to delivering faster results and better decision-making because digital companies tend to be less hierarchical. Employees will prioritize making the customers’ experience better, as digital companies use a closed-loop feedback mechanism. Empowered operations staff will always put them in the customer’s perspective for effective feedback because indirectly, that will make their lives easier and improve their performance.
The mindset shift from the top to bottom level employees in an company to make digitalization a part of their everyday core functions is the need of the hour. From the board members, to the risk management teams, everyone should be cognitive about what technology they are using and what are the risk and reward towards each of the systems and modules. Even for the people in the front office and business in general, the over-reliance on a software vendor or IT department for each of their technical problems is a thing of the past. A comprehensive technology school should be established as a part of the digitalization program, where the right talent could teach and lead the technology innovation across the company through knowledge transfer and hands-on training.
This year has accelerated innovation to reduce risk and improve resilience across all sectors. how did technology influence innovation in capital markets.
Capital Markets, in general, has seen unprecedented growth over the last five years in commemorating different kinds of technology and digitalization techniques. The market is also looking forward to some fascinating upcoming amalgamation of mathematics and tech-based products and algorithms at its disposal. I find it really interesting to talk about technology and capital markets, and can talk for hours. for the sake of the reader, I will keep it short and crisp.
One of the most disruptive and important innovations that the markets have experienced is cloud computing. The innovation in this space is amazing, and the choices are vast because the FinTech industry is literally competing with each other in this space for market share. Competition in the market and cloud providers embracing open-source solutions together with their own proprietary software is good for the consumers. Services and products offered become less and less expensive. One of the interesting advancements in the cloud computing arena in capital markets has been the integration of time-series databases with parallel streaming of activities in milliseconds to viewers linked to exchanges such as Nasdaq, NYSE, etc. This has literally changed how companies perform and view activities in the markets.
Technologies such as distributive ledger mechanism (blockchain) have great potential towards making processes transparent, secured, and more error-free than ever before. Someone’s output is someone else’s input, and if there is a mismatch, you cannot proceed further without fixing the error or without agreeing with a consensus. No central validation is required, and cryptography enhances the authenticity of each block. What does that mean if you introduce this mechanism in financial operations? Reduce costs of errors, sure shot outputs, increased empowerment, trust, transparency, and structure in the processes.
The middle office and back office heavily rely on Robotic Process Automation (RPA) and Artificial Intelligence (AI) nowadays. RPA has the ability to eliminate human error in structured, repetitive jobs, enhance speed, reduce operations cost, and increase profitability by quantifying the amount of jobs done in a period of time. Now the question is, what if you need decision-making – can RPA alone do the job? But if you introduce artificial intelligence to the RPA, you might be able to train your algorithm to parse through unstructured data as well. The algorithm can be trained in such a manner that the robot learns and adapts to the process autonomously and then see the results.
Did someone say Smart Contracts!* Heaps of paper-based legal documentation can be very well replaced with Smart Contracts. Smart contracts are basically programs that verify, facilitate or negotiate the performance of a contract without themselves being legal contracts. Distributed ledgers might be enhanced with smart contracts, and there is research going on as we speak if that system can mimic service agreements and credit support annexes in the derivatives world. If that happens, it will provide extreme transparency over contractual terms between dealers and counterparties in the markets. But yes, Smart Contracts can only work on black and white structured data, so it might fail when decision making over a contract is needed. So, there are some risks associated with it as well.
I was reading one of the studies conducted by EY the other day, and it stated that the average amount of customer data that is not used by businesses is a staggering 80%. Think if we could access this unused data at what level of significance a predictive analytics module can project your outputs to? It’s like you will be able to predict the future performance of a portfolio with an outstanding level of significance. Here comes Advanced Analytics – a process with which you can make precise and powerful decisions by using extensive data sets. It encompasses a number of techniques, but the most well-known are predictive analysis, sentimental analysis, behavioral analysis, and data visualization. The only problem with this mechanism is that it requires a huge amount of data to provide better outputs, and the cost of parsing useful data through structured and unstructured data is usually high. So, here is a place where we need innovation!
Capital market firms are adopting to automate bond pricing and risk management by using AI predictive analytics for better pricing and liquidity in the market. Do you think/believe that AI-based algorithms can enhance bond trading performance for the firm?
Yes, I cannot agree more to the fact that AI applications can significantly enhance bond trading performance in firms. AI can be extremely helpful because the bond market is considered less liquid than the stock market. There are various reasons such as trading over the phone, over the counter (OTC) trading, huge trading sizes of bonds that a trader can trade, and of course, the fact that there are no direct announcements of the current market price of bonds, to name a few. You can say that there is a liquidity issue with bonds. The factor of transparency in information between a stock and a bond is a matter of concern. You can see the market price of a stock almost in real-time. there is a 15-minute delay in actually accessing the market price on a bond. This is where AI can play a vital role in Bond price predictions.
With big data analytics and machine learning algorithms, the price prediction in Bond desks can be significantly improved. Algorithms can use large historical data sets available and work on various variables such as trading pattern, trading history, issuer fundamentals, ratings, and other market economics. Although the algorithms can be designed to find trends in the data and transform unstructured raw data into usable structured data, it is imperative to employ machine learning techniques to improve data mining processes. Collaborative filtering systems and hybrid recommendation systems can be used to analyze the trader’s trading history and recommend to the trader as to which bond to buy and which to sell.
Some hedge funds and financial institutions use chatbots to provide assistance to customers. The same technique can be designed using smart online platforms such as IBM Watson to create chatbots that can help traders as to which bond is a good buy and investment-worthy, thereby reducing the risk in the portfolio and providing greater risk-adjusted returns, and reduce human error.
Machine learning algorithms can also be used for excellent interpolation and extrapolation in filling missing yields if the market is illiquid. Think about an algorithm that trains itself by observing hundreds and thousands of historical corporate bond yield curves. It can incorporate the idea of how the yield curve should be, so when there are missing points due to illiquidity in the market, it can interpolate or extrapolate the data using historical trends and predict the yield curve, thereby predicting the price of the bonds. There is a research paper that focuses explicitly on this – titled ‘Machine Learning for Yield Curve Feature Extraction: Application to Illiquid Corporate Bonds’, written by Greg Kirczenow, et al.
Can you talk about quandamental investing, and what are the major challenges associated with it?
Quantamental Investing has pretty much become a buzzword nowadays, especially among hedge funds, wealth managers, and trading firms. Although it sounds very complicated, which it truly is, we can understand it by simply breaking the term Quantamental into two words – “Quantitative” and “Fundamental”. So essentially, the use of quantitative methods in deriving insights using extremely large amounts of market data and then applying the fundamental approach of investing in taking the final decision of investing in Quantamental Investing.
Quantitative investing refers to applying and using complex machine learning algorithms over a vast amount of structured and unstructured data available in the market to derive insights, patterns, trends, etc. As computers are involved in this case, algorithms can be faster than humans in analyzing trends and better predict market movements or security pricing in the future. Quantitative strategies such as price arbitrage are used to scan the market, find out securities, and then compare them with similar securities and determine if there is any price difference. If there is a price difference, you can capitalize on that differentiator, known as arbitrage. Factor investing is another quantitative strategy where the models/algorithms try to find out specific economic factors that could influence the pricing of security (macroeconomic factor, microeconomic factor, style factors such as market cap, etc.). Again if there is a difference in price among identical securities, you can capitalize on that. Algorithms are faster, more error-free, and not biased like a human brain. So decisions and options can be generated within seconds which can be acted upon and get you better risk adjusted returns.
Fundamental investing refers to the traditional bottom-to-top method of stock picking by analyzing the balance sheet, income statement, cash flows, ratings, etc., of a company and then deciding on the health of the company before investing. Although this process is one of the best processes to pick stocks/bonds traditionally, this process can be slow, and it is nearly impossible to find out and capitalize on opportunities that exist all over the market as only human brains are working behind it.
By merging the two processes and binding the computer’s computing and analytical power with the human’s rationale and decision-making ability, a really good ensemble could be created to get the best out of the market. There are some big Wall Street players and hedge funds which have already implemented Quantamental Investing in their approach, and have generated competitive returns. But with all the good comes the bad too. It is not that Quantamental Investing does not have its problems. First of all, you will need extremely large amount of data sets, to actually predict the markets. A lot of time, a huge amount of data sets can yield no results as they can be corrupted or not appropriately mined. That calls for the right data mining processes – which machine learning might solve with time. Then comes the problem of cognitive ability. An algorithm can analyze data and get you the trends. the human needs to take the final decision. A human can make a wrong decision even after given the best alternatives. Machine learning algorithms with sound cognitive abilities to mimic the human brain are way far in the future. We also have to remember that humans build these algorithms and strategies, so the results derived by these algorithms are somehow linked with the developer’s intelligence and what it can do. Highly skilled analysts need to be hired to code these algorithms. Significant upgrades in internal systems are required to cope with the additional strain on resources caused due to big data analytics. Thus, this requires substantial investments even in the beginning stages of development.
Questions and answers #
What does that mean if you introduce this mechanism in financial operations? #
Reduce costs of errors, sure shot outputs, increased empowerment, trust, transparency, and structure in the processes.
Now the question is, what if you need decision-making – can RPA alone do the job? #
But if you introduce artificial intelligence to the RPA, you might be able to train your algorithm to parse through unstructured data as well. The algorithm can be trained in such a manner that the robot learns and adapts to the process autonomously and then see the results.