AI in Financial Markets
Hi everyone, welcome to a financial markets webinar. I think we can wait for another one or two minutes to lick my writing many or many people are signing up for the webinar so we can wait for either one or two minutes and we are good to start.
Hi, everyone, welcome to air for financial market program webinar. So I'm nickel ready, I had the CIA for financial market program sales and throughout my career I've been to a complete deep tech program say in San Diego with more than like 3000 plus people into our deep tech programs. So coming into this program, before that, Vicki, I would like to welcome our professors Dr. Anand gentlemen and processor doc ready. So prefers side like you people to introduce yourself.
Hello, everyone. My name is Alan J. Rahman. I'm an ex hedge fund trader. I have a PhD in physics. And I was in academia for about six years before moving into, into the hedge fund space. And I started out as a research analyst, where I was using math and data science essentially to try and understand financial markets. And eventually it started building algorithms over there. And my soon my algorithms started performing well, and I grew and fund to, to a rank of a portfolio manager. I was a portfolio I was in the fund for about 10 years, when I left the fund, I was managing about $100 billion. For the past six years. It's been six years since I moved to India, this the fund I was managing was in based out of Seattle. For the past six years, I've been doing a variety of consulting, jobs that are based out of AI, I run a small consulting company, and although the largest projects that I get are in the financial markets, I also work in other sectors, healthcare and political analytics, and so on and so forth. I hope my experience in the markets that's that's what I'm hoping I'll be able to convey in the during this program. Thank you. Thank you. Hi,
good evening, everybody. I'm sadaqa ready. I'm a assistant professor in I am Calcutta in the finance and control group. So, my PhD is in the area of Finance. And my research interests are the financial markets, market microstructure algorithmic trading, etc. So, I am quite good at technical analysis, behavioral finance, market microstructure and various other cutting edge courses in the finance area and I offer courses at the PhD level. And at the MBA level, along with many other executive programs, so executed programs for various companies. We have open programs as well as a company level programs. And I myself I do a lot of investment and I actually have a keen interest in financial markets. And I've been in the markets for the past 10 years. And that experience my academic experience as well as practical experience. I learned a lot of things in this journey. And basically I would be discussing or I would be delivering lectures with respect to the financial markets, financial analysis, portfolio management, etc which will be helpful for you the basic level courses which will be helpful for you to build your AI in financial markets? knowledge. So that's how I will, I will be associated with this program. And I thank all my colleagues here, Dr. Anand, for giving me this opportunity. Also. Thanks, Nikhil and his team. Yeah.
Yeah, yeah. So I'll set the context. And later I'll hand over to the professor. So process will explain you in detail about the program curriculum at the same time, program application implementation and everything. So coming to the program context, actually, so the main reason behind launching this program in financial markets, like we have a lot of huge set of AI related machine learning related programs with nc talentsprint. But the objective behind designing this program, particularly in financial markets, we generate huge volume of data. And that's where actually we need exact insights, right. And to get that insights, when we are analyzing huge data, definitely, we need AI. And in stock markets, we have a huge number of variables, not just like four or five variables, and definitely in that context, definitely when we are making a decision of we have to analyze all the variables, we have to take a decision on the basis of all the way all the variables right. So in that case, traditionally, in non stock markets, what we are observing is most of the traders or most of the financial experts, they to take call on the basis of the data got generated in last five years or 10 years or 20 years, right. And even when we see in traditional strategies, most of it are based on aggregated data. And, and with less variables to be frank, and maybe a bypass performance and technical answers sentimental answers, a lot of sentiment market sentiments will affect our, like, buy items, like pipe our purchasing. Right? So and that's where actually we need AI to analyze all these variables. And that's, that's the main main objective in designing this program. This is a six month program where we'll be having the classes on Saturdays and Sundays, so weekly, almost like six to seven hours of teaching. So how we a transformation in financial markets and our and even including the areas like operations, human resources, cybersecurity and analytics, right? When we say financial markets, it's not just about only share market even we can see bond markets, money markets, derivative markets, even future markets, currencies, insurances and even when we do this program, it is not just confined to only Indian markets, it is for foreign exchange markets also right. So, and and if we see in India or also throughout our many of them, not just our Indian companies, many of them and see companies, almost 38% of companies are already using a in some capacity. So even if you see a lot of startup companies are also using AI machine learning almost like 20 to 30% practices already there on the market. And they started analyzing the data and they're taking a call on the basis of the data which got generated in last many years right. So, we can see a lot of companies over here a lot of the big companies like Bank of America or PwC or Citibank, Deutsche Bank. So, coming to the current industry changes are leading financial institutions like they already started hiring many chief AI officers and investing in a labs and incubators, like I already said many startup companies are also using AI and machine learning in their practices, right. So we have we are seeing a power banking boards are commonplace, right? So when we see the success stories of the one success story, most of the people know as JP Morgan Chase deployed here for document processing. So usually take 300,000 hours 2000 hours to analyze the data or process the data, but by implementing this UI, they reduced it to seconds. Right. And we are seeing the same in hedge funds and how they are detecting your humanly impossible market changes and top secret agencies also implementing gay for risk assessments also risk management also. Right. So I would like to hand over to Professor Alan J. Robin and Dr. sadaqa ready. So I would like like to throw some light. I would like them to throw some light on the program application cases at the same time. I would ask them to give some detailed understanding on the curriculum. Also. Over to you professor.
Thanks, Nikki. I'm going to share my screen now. Nicole, I hope you can see my screen. Yes, yes. Okay, thanks. So what I thought I will do is, instead of just talking about the general applications of AI in financial markets, which I'm sure you are, many of you're sort of already heard of this, I thought I will talk a little bit about a specific use case, one of the consulting projects that I recently did, right, and I thought that would be a good way, you'll see exactly where AI or machine learning is applied, and what kind of results it can produce. Right. And that will also give you a good sense of, you know, what you can expect to do once you finish this program. I'm sure many of you already know this. I mean, you have seen users of AI or machine learning all across in the financial space, the biggest innovations that you might have heard about are these idea of robo advisors. You're talking about? Many of you might have seen ads in the market about, you know, time this IPS. s IP is something as we know it no systematic investment methodology, where do you have every month you put in some amount of money, but time this IPS idea is that instead of always, let's say you're choosing to put 10,000 rupees every month into a mutual fund, it might not make perfect sense to put in that 10,000 rupees if a market is at all time highs, and it's likely to fall. Right. So what time is it idea is that you try to time the market in such a way that perhaps when the markets are too expensive or something, then you divert some of that SAP amount to bonds instead of purely equities. Right. And all of these products use a AI or machine learning to try and identify when the markets are likely to go up or when the markets are likely to come down. And the example that I was going to talk about is sort of related to that, right? In India, the mutual fund, when you're typically investing money in a mutual fund, you know, the the fees that average active mutual funds charge are pretty high, right? You have up to 2% fees pretty high fees, but as a passive mutual funds, such the fee charged are much lower. Now, as an investor, an investor always wonders, you know, am I paying high fees on these actual mutual fund and does it does it really realize do I get better performance because of that? Now, what has happened is that you find that the annual average annual outperformance has actually dropped. It was used to be 6% a decade ago. So there was value in putting money in active mutual funds versus passive mutual funds. But now that outperformance has sort of drop, in order for any of these funds to continue charging these high fees, they really need to use analytics to try and do better than what markets what passive fund is actually really. Now, recently, I got a project, I did a project where this particular retirement management fund, they were asking me the question that okay, I put money in. And this one is based out of us. And they our clientele, largely our retirement clientele. And they wanted to put money only in blue chip companies, right largest, the large cap companies in the US, the large cap companies are represented by this s&p 500 index. Most of the retirement, folks would have a large portion of their money in bonds, and some portion of the money they will invest in the index s&p 500 index. These are all blue chip companies. So they don't, they're not likely to drop, you know, 5% or 10%. On bad news, there are finding the company so you have good diversification in your portfolio. Now, the fund manager asked me this question that, you know, I know that when I hold a you know, instead of putting money entirely in the s&p 500 index, this is the s&p 500 index that I'm showing you. Or from 2005 to 2022. If you had put $1 million in 2005, and and held or in s&p 500, the money would have grown all the way to 5 million, right? It's pretty good returns, it's about 10% returns unrealized. But the fund manager was asking this question, right? It's well known in finance circle that if you diversify your portfolio in not just equities if you hold equities and some amount of bonds, like 60% of the portfolio you put in in equities and 40% in bonds, then your portfolio You will have a large, very much more smoother ride. Right, the red line that you're seeing is a portfolio that held only s&p 500 stocks, right, so entirely in equities. On the other hand, if you put 60% of the money in s&p 540% of the money in bonds, and you keep rebalancing on a quarterly basis, you your equity curve sort of looks like the blue line. So you had the the steep drop that the s&p 500 experienced in 2007 2008, when the financial market crashed, that would not have happened to your portfolio, your portfolio will be represented by the blue line, which has a much more smoother growth, your annual average return instead of about 10.5. It drops slightly, but the curve, the equity curve is much more smoother, you don't have to have a lot of sleepless nights. Now, the fund manager asked me, can we use machine learning to even do better? Right? I am in this case predicting I'm putting 60% of the money in stocks 40% of the money in bonds? Can I keep playing with it? You know, maybe when the market is not not that risky? Can I put 80% of the money in stocks, and only 20% of the bonds? And let's say when I know that markets are going to crash, maybe I can move some of the money and put more money in bonds and less money in equities. Right? Can I time the market? Right? Instead of having a standard 60 4040? portfolio? Can I time the markets? This was the question that the portfolio manager asked me. And, and typically, the way you would do this is you would look at some fundamental data about the market. Like for example, here, I'm showing you some data about nifty, even though the analysis we did was on s&p 500, just to set some context and giving you some nifty data. So if you look at the nifty price to earnings ratio, right, this is one of those things that people look at to understand if the market is expensive or not, when the price to earnings ratio is very high, it's typically implies that the market is too expensive, and it's likely to crash. And that's sort of sort of, loosely speaking, traditional interpretation. Now, if you look at this red graph, what you saw, what you see is that the value of this red graph is close to, you know, 2728. And if you look back into history, when has it been this high, it's been high this high very few times, once around 2000, year 2000, when the markets crashed, and the other time when this happened was shortly around 2008, before a big market crash happened, right. So when I see a P e ratio of that high, my prediction would be the market is likely to collapse. This is a very bad time to put money into the stock market. Right. But then there are other indicators people look at, when I look at, for example, the price to book ratio. But look at the price to book ratio. This is one of those things which talks about how expensive is the stock compared to the the book value of a company, right? And this number, when it is again, when it's very high, it tells that the markets are overvalued, when it's low, it means the markets are undervalued, if you look at the price to book ratio, the current price to book ratio numbers are not particularly high. Now, this is what will happen when you look at fundamental indicators. When you look at fundamental indicators, often you might not have very clear signals, you might often have contradictory data signals, right? How does one take this contradictory information and come up with one coherent decision? Right, this is the primary question people have the fund manager that I was dealing with was asking me this question. Typically we make decisions on Should I put more money in bonds or stocks based on many of these fundamental indicators, right? Can you do something better? Can we use machine learning to try and understand and predict more in a better way?
Whether the markets are likely to fall or not? And in order to do this, we are going to take a very the data driven approach. The way what do I mean by data driven approach? This is a kind of approach all algorithmic trading guys have done for decades, right? I want to make when people create algorithmic trading strategies, they use past data and from the past data, they try to understand patterns and then they validate these patterns and they say that if I see this pattern, and if I hadn't placed trades based on the pattern in the past, would I have made money or not? And based on that they make these decisions, right this approach is called as back testing. Now, back testing has a lot of issues that are the thing is one kind of idea, you might think is that you might say that I might say that I will buy only when the P e ratio is low. And you know say when the dividend yields are too low, then it may be back, you know, you might come up with some idea like that you formulate an investment strategy and you test it, you look back in history and see in the past two decades, how would you have done? Right? This is sort of the back testing way of doing things. But there are problems with doing it in a back testing way. It's effectively what you are doing is curve fitting, you are looking back in history with sort of, you know, with a clear with, with a, you know, sort of the you'll have a 2020 vision when you obviously look back in history, but it's not clear in 2010, you would have made the same decision or not. Right now, just because you saw it seen that in the markets or mood in some way. Now, you might say that this strategy might have worked well. But what would have been a decision you would have made in 2010? Without knowing without the benefit of hindsight, that is a difficult question. And machine learning helps you to solve that. What we do in machine learning is we do not necessarily rely only on domain expertise. Often what happens with domain expertise is that you are relying on some intuition, someone has told you that when the P e ratio number is below this number, it's very good when P e ratio number is about this number. It's bad, right? That is based on some amount of experience, it might have worked for one time period, but might not necessarily work for other time periods. And so any many of this intuition are often prone to biases, right? The better way of doing it is entirely through a workflow process in machine learning that we typically call as walk forward, walk forward testing, right, walk forward testing is the method something that we will talk about in detail. When we learn about machine learning, as a part of this program, the core, the core idea of walk forward of building a machine learning idea is this, what we do is we collect data for every single quarter in the past, this is exactly what we did for the project. So for example, a quarter ending in 2015, the previous quarters returns, the stock returns were so much the volatility of the stocks was so much and the dividend yields was so much so on and so forth. We collected all the data and for that, that particular quarter, and we found out what happened to the markets in the following quarter. Right. So this was a quarter ending in the March. So what happened from April to June, we found that the market actually went down. Then we collected data for June, the quarter ending June. And we collected data and we found out again, what happened to the market in the subsequent quarter, we found the market went up, we collect this kind of data for every single quarter. And you ask the machine to find patterns in the data. And the machine goes ahead and finds patterns. And we make sure that this pattern does is not overfitting. It does not look necessarily. It does not make decisions, looking at what happens in the future. But instead it looks at data only that's available up to that particular date. We used a whole bunch of indicators, we looked at technical indicators, we looked at not of economy based indicators, you know, how is the federal government pumping in more money or less money, what are the interest rates and so on and so forth. These indicators are often called us macro economic indicators, right macro economic indicators, and we will talk about what these indicators are when we are discussing the course when it during the course right. When you put in all of this indicators, what this machine learning algorithm did was it came back and it was able to predict which are the periods which are bad for,
you know, holding stocks. And when I saw it made decisions, it said that this particular period is a neutral period, it's neither too risky nor particularly bad, this particular period is very bad period. This particular period is very good period. And it was making this decision before the onset of the period, not after the onset. It made these decisions before the onset of pain. Now machine learning algorithms are not perfect, right? It cannot predict everything in the future. What it's doing is it's trying to compute in a probabilistic way, what are likely to happen, what are likely to or what's not likely to happen. Now, one thing you might notice is it completely missed the pandemic. It did not know that pandemic was going to happen. Of course, nobody knew pandemic was going to happen. Right. If you look back over the data that was available in December end of December quarter, none of us would have guessed December quarter of 2029 2019. None of us would have predicted The first quarter they would have pandemic would have happened, the machine learning algorithm also does not predict it, because it doesn't have that feature in it. But in spite of that, when we applied this machine learning algorithm, the predictions if you follow what this machine learning algorithm had said, and if you invested in SNP and bonds accordingly, you know, invest more in s&p during good periods, and more advanced during bad periods, when we do that, what we find is that our performance ends up being better, instead of being 10%. Before now, we have gotten marginally better, we are about made about an extra 1%, which is actually quite substantial, right? If we have made about 11.3% return. So instead of getting ending up with a portfolio of 5 million, we are ending up with a portfolio about 6 million. We then ask the question, can I even decide which sectors are likely to do, right in a given index, there are multiple sectors, and we use the same machine learning idea to figure out which sectors are likely to go do good. And this algorithm again, was able to go ahead and start allocating, you know, energy sectors, the blue line blue region you're seeing is the energy sector. The red one you're seeing is that material sectors, it started saying that what percentage of your portfolio should be materials, what percentage of portfolio should be energy, what percentage of portfolio should be financed, and it kept changing it based on the indicators based on these macroeconomic indicators? Remember, we are not doing day trading, we are adjusting our portfolio once a quarter once in three months is all you're touching this portfolio, which is what a retirement person a retiree would like, not do day trading, right. And so this is the kind of algorithm a mutual fund person, the mutual fund portfolio manager would be interested in. When and we did the same thing, can we add gold to the portfolio, and we were able to be asked algorithm to predict, which are the good times to hold gold and which are the bad times to hold gold. And at the end of the day, when we put together all of it, what we were able to do is we were able to create a portfolio that that did much, much better, it had an overall return of about 14%. And again, we are not doing day trading once a quarter, I look at the portfolio and decide what percentage of my portfolio should be gold, what percentage of portfolio should be materials, what percentage should be financed, so on and so forth. And I do this adjustment once a quarter and I manage to my portfolio, I managed to avoid drawdowns in the portfolio much in a much more smarter way. This is a product that we created about six months back and the manager is trading with it, it's right now running in the test phase, the manager has put in about $14 million in this. And this is the kind of idea where you are using ml to add value to a traditional portfolio. Now there are a whole bunch of things, additional things you can do. You can in fact do sentiment analysis, you can look at data from Google and find out and add look at the the current sentiment in the data in the markets and use that to add to your portfolio. Right. In order to do that you can't do it manually doing it manually is not efficient way. You can write algorithms, which can look at the all the tweets that are going through being published in Twitter, and try to grasp the sentiment of those tweets, and use that as an additional signal for your investment portfolio as well. And this is something that can be done. And in order to do that, again, you need a good knowledge of data science. Right? Now, this is what the field of data analytics does. What is data analytics, it's basically a bunch of mathematical ideas from a variety of fields that have been brought together right. And now data sciences are actually now taught as a separate specialization right now. And this is being used in all sectors, every single sector has been touched by machine learning or ideas from data sciences. Right? I would say a lot of these low hanging fruits of what what are the easy progress you can do has already been done, right? From now on in order to make real progress, domain understanding and domain specialization is essential in order to make further progress. And this is something that everyone has recognized, right. All MBA schools now
offer a one semester course on introductory course on data science. Right? And this is it's because they're realizing that data sciences are actually absolutely important. What ends up happening is in many of the funds that I have worked, worked with, there are two this is white divide between The two kinds of specialists you find in these firms, there are people who have very solid understanding of traditional finance and find financial markets, right? But often not. So such a good knowledge of the data analytic methodology. Right? So and to, to add to the capability of the firm, what these funds do is they go out and hire people with data science background background. Now, date people with data science background, typically understand algorithms, but they do not quite understand what's happening in the financial side of things, they do not have domain knowledge, right, what we are planning to do in this course, is essentially do teach data science together with the requisite financial background. Right. And this is, I think, one of the key things about this particular program, right, I am a practitioner, I've been working in the industry, I'm my background, per se, is not in finance, I, as I mentioned earlier, I have a PhD in physics, I worked for 10 years in the hedge fund, and essentially, I've been looking at the markets entirely from data analytics perspective. And Professor sadaqah actually, as is an expert, now, who comes from very solid academic background on financial markets, and, and he has traded a lot on a personal capacity as well. And so, what we are doing is sort of we are bringing both aspects together, right, and putting this together from entirely from a practitioners perspective, right, and this is what our course looks like this particular course, we realize that a lot of people who are joining the course, might not be might not have good amount of programming knowledge, they might not have a solid understanding of mathematical methods that are needed to understand data sciences, and some of them might not have understanding of traditional financial markets concepts, right. And so, we have divided our program into two separate segments, so to speak, the this is the outline of the foundation modules, which will run about the first half of the course, where we will teach you basic ideas of the data sciences, right understanding of what is probability, how do you look at what is linear regression? How, what is the basic idea of supervised learning? And how is this and what is what do you mean by optimization and so on. And these concepts are something that we will be teaching, not with general textbook examples that you typically find in a data science course. But examples that are very much applied examples and oriented towards the finance domain. Right. And there is a section that's that Professor Dr. Reddy is going to lead on where he is going to give you the requisite back background that you need, in order to understand things that are typically start taught in a sort of a first year of the first semester of your MBA course, this is what we typically cover on the foundation. And then separately, the second part is the advanced modules, where we talk about advanced machine learning methods, and also advanced financial concepts. I'll pause at this moment, and then I'll ask Dr. Zucker to talk a little bit about the finance models that he's going to be teaching.
Thanks. So friends, if you see as Professor Arnold rightly mentioned, the program is a blend of practice, from the data science requirement along with solid foundations in finance. So, it's going to be a very good program, because we are going to bring in For example, we have target from the finance background without technological experience or the data science experience, and on the other hand, we have people having a different type of experience. So, in this scenario, this program would build the you know, requisite concepts on both ends and you know, it will make a comprehensive program for all of you. So, if you look at the finance concepts which are required, so we are going to have all the finance concepts built into this 18 hours which are given Here, starting from understanding the financial statements, and using the financial statements for equity valuation, or bond valuation, or business valuation, because if you look at what prosound has given the example that He has given, it's all about how you understand the markets and generate ideas. Somebody asked a question in the chat box, asking that if everybody knows what we are doing, then that's not the thing. One thing that I understood from the markets is you should be different from others, then only you can beat the market. And you should always be under the radar. You should think innovatively. So how do you think innovative, everybody knows how to how to see the charts, everybody can understand what is PBE? Who is investing in the market? But in my charts, can I make a setup, which is different from others, where the big fit is, don't see the order book and understand my setup. That's what is the difference which comes on. So that to understand to build those ideas, you need to have a solid background, a solid background is in terms of understanding practically, for example, what is the importance of a debt instrument, versus an equity instrument, when to have for example, what was Ron has mentioned, a beautiful example, just given a practical example, which has given me something which looks very simple, but very, very powerful. So we try to give you the solid background and give some practical examples of how to use them how to when to use them. What are the implications of each one of that, for example, if you want a very good example, which I see, for example, good traders in the market, what they do is they invest all their money in bonds,
use the bonds as collateral, and right options, where they aren't 6.5% there, and around 3% here, and they make hefty returns, what we do is they put all your money in equity or options. And you're losing that 6.5%, which is huge. So that So things have to be understood. And over a period of time, we should have our own innovative thinking, that's what I feel, personally, you should have your own innovative thinking and thinking is what we are inculcating with some live examples with the machine learning you feed your ideas to the machine, feed your ideas to the algorithms used based on the back testing. And then I think what hedge fund managers, the amount of return that they generate is all because of the innovative ideas. So unless we are not saying that we will teach the innovative ideas, you know, like you don't spoon feed them, because that's not the way it works. But we will give you all the information which is required to build a solid foundation for you. So in this regard, we're going to cover various aspects of the required things. For example, if you're not having the finance background, we'll give you whatever they are not a MBA kind of a finance concepts that they give. So what we do practically in what we need practically, how do we trade in options? How do we look at options market? Why do we always Why do all all the buyers and option markets failed? Why only the writers with just 10% knowledge that the buyers should have they always been so how to beat this all these kinds of things, which we'll talk when I talk about, for example, the derivative section I talk I'll talk about that, for example, I'm talking about portfolio management, what is portfolio management traditionally, and how it has changed now, how the behavioral finance if I if I see the B portfolio modern portfolio theory right now, it is entirely entirely it has changed and how it is related to machine learning how you have to use with advanced, you know, programming that we are doing. So all these things are built here. And I hope and we also cover we also cover something on the market microstructure. Remember friends one thing that market microstructure is all about how the individual traders are behaving in the market, and how they are changing the trading altogether, how the big traders are thinking how the small traders behave. So what is the advantage that the big traders have or the small traders, so how a small trader should think like a big trader, well, even though he's having a smaller capital. So these are the things that I'm going to discuss in the market microstructure and then the behavioral aspects all your technical analysis is nothing but the behavior Finance is a byproduct of behavioral finance. So how do we build behavioral strategies? How do we have? How do we understand the trading psychology? So all these things are covered here in the program. So basically, I'm not the data science guy, and I'm not I, but I'd be very, I would say that, it's going to be very helpful, because trading something manually and writing code and executing that is something powerful. So it's going to be a very good program. And it's a one of its kind in India, and started by NSE Academy, nobody, you know, you cannot match NSS technology and as the Academy's, you know, strength in these things. Well, and we are also as maybe, we'll throw some light on that, we are going to give you access to the NSE data, nse trading book and order book data to work on some back testing strategies. That's what we are planning. But maybe you will get some more information in the future. So this is all from my set on. So can you you can take out.
Thanks. So I, I see that there were a lot of questions, I think we can get to the questions. One of the questions that I see in front of me is, What language is going to be used for the MLS portion? We're going to be using Python for coding, and this is, if you're not familiar with Python, that's okay. Before the program starts, there is a bridge module where we'll be giving you the required programming back background. And I have to add this as well, I, as I mentioned, I'm I'm a I'm not a computer science guy, my I'm not a programmer, per se. Time. I I learned Fortran a long, long time ago when I was doing my 100 graduation. So but I have been coding. Because thing is whatever you need to code, right? It's the syntax, part of it is easy. The biggest part of it is the EDS, right? Once you are able to formulate an idea and have and be able to clearly express it in an algorithm. The syntax part is the easy part. And 90% of the time, I'm always googling and I'm trying to find code snippets to accomplish those things. And the beauty of Python is that for doing data science, lot of packages are available, and many of the things that are commonly done tasks that are already packages that help you accomplish it easily. Right. And so the pattern is going to be very, very powerful pattern. It's a wonderful tool to accomplish what we are trying to do here. And nickel, do you want to sort of go over different directions? And then we can have an audible professor?
So we have a couple of questions. Really excited to see that questions also, because a lot of participants are showing a lot of questions. So so one common question, which we are hearing is like what kind of programming language is required? And do I need to have any kind of programming background to do this program? Because as I'm from a core finance background? So that's kind of one common question which we are getting another question which we are getting is I am from a core finance domain. And I don't have any kind of understanding towards stock masari stock market investing even though I'm from a finance domain, but I don't have any kind of understanding to stock markets, but I am really inclined towards learning this program. So how this program will help me out and how I'm going to get benefits by doing this program. So would you like to take the question?
Uh, sure. Yeah. So this is not all that uncommon. So this is one of the unfortunate fact of our educational system is that many of us have done, you know, people who come out of engineering would not be actually that comfortable about using all of the core ideas in practice, right. And that's, that's the stuff and one of those unfortunate facts of our educational system. And this is something that we acknowledge we accept. I have actually, I teach also at some of the US universities, but there's a Ph. D. program. I was recently teaching for a Ph. D program in one of the French University and people who are doing a doctoral level program on in finance, in finance, struggle with basic ideas regarding stock market. And and I understand that right. And this the, that was one of the learnings from teaching that is one of those things that has gone into creating this program, right? We are planning on starting from the basics, right? What we plan to do is, all of us have learned probability in seventh or eighth grade, right. But we are going to assume that none of us will remember that right, we are going to go back and start revisiting probability and y covering probability we will be using stock market examples, we will start thinking about stocks, we'll start thinking about daily movement of stocks and look at distributions of daily returns. And from there, we will introduce both Mac and the required knowledge of Finance. And I believe this is sort of the unique selling point of the post that we are going to start from the basic and cover in a very applied way and with sort of a laser like focus on financial markets, right. And Professor Dockers module is will definitely give you the required background for in terms of thinking about investments, right? Yeah, yes, you have a minor amount of finance background, but then the kind of ideas and concepts that are required for investments, that's different, right. And Professor Zucker, is going to emphasize on that Professor, Zachary Otto. Yeah,
as I mentioned, the idea is not the textbook stuff, especially coming to the finance concepts, we'll use even data. When talking about these things, I'ma give you some assignments, using them historical data, to understand the behavior back test. So it's not even if you're having a background in finance, you will see that there's a big difference that we see because that's how we build the course. And as I told you, you will have the data in NSE data readily available, which you can use to learn for example, to construct a portfolio to optimize a portfolio, how do you optimize a portfolios in Python? So as simple portfolio, let's say a five asset portfolio, how do you do that? How do you test how it has performed? So that understanding that you have will definitely help you and to you know, strengthen your concepts? Yeah.
So we have other question of, should we have understanding towards technical analysis or to do this program? Like how do you compare both like AI in financial markets and technical analysis programs? like is that monetary, to more technical analysis also to this program? That is one question which we got got from the audience? Yeah. Yeah.
So you as a as Prasanna mentioned in his presentation, let me tell you one thing, all whether it's technical analysis, or behavioral finance, or trading psychology or investment analysis, you need to have everything you need to know everything, to be a successful trader. And I don't say that, you need to have all of that you need to have some idea of that. But you can build your strategies on on any of these aspects. For example, if I believe in technical analysis, and I understand technical analysis, I can build this or I can build my strategies, just based on technical analysis, if I am, if I believe for example, there are investors in the market who don't even look at charts, they don't believe in technical analysis. So, what they do is they use the investment analysis, the portfolio theory, which they believe in and the building. So, there is not nothing which is essential, basically, for the program, because we are building the program from scratch. So we are going to introduce all that is required. And then whichever you like, whichever you think that fits your ideas better, you can take it for. Thank you. Thank
you. So we have another question of this is regarding job opportunities, like because we have a lot of people who are stockbrokers and portfolio managers who want to know more about this program because they want to switch their career towards AI. Ai related to jobs in in many stock broking companies. So the question is like, what kind of job opportunities will I get by doing this program?
Sure. So I one of the things that so if you have been looking for specifically looking for job opportunities. In you know, any generic website, JOB, JOB search website, you do, of course see a lot of jobs that are there in, in the finance domain, specifically in data science area in finance domain. But I would encourage you guys to just take a look at E financial careers, right, this is one of those websites, which specifically talk about financial careers ever job opportunities in finance, right. And you will be shocked to see the number of jobs that are they're asking specifically for people with data analytics packet background, right? This is now sort of, you know, the, it's come to the level that you know, how, when people apply for a job in, you know, in more or less any domain, they expect you to know, Excel, like, they expect comfort level in Excel, right? Why Excel, because you need to handle data, it doesn't matter what your job description is, you need to handle data. Data Science now is, is become so democratic. Now, there are plenty of tools that are available, which make this data analysis easy. And this is more so in financial domain, right? Regardless of whether you're going to take this course or not, you need to be comfortable with data, right? That is an accepted, right, what we will be doing is we are going to give you a much more stronger tool, which is Python, Python will allow you to do all kinds of not just visualization analytics, regarding with regards to financial data, once you go through the course, you will be able to start doing predictive analytics as well, right, you will be able to make forecasting some of the things which are very valued in business, right, this is what I am saying, just at the end of the program at the end of the seminar, you can just check the financial careers.com. And you will see the number of jobs that are there for in the financial domain specifically asking for data analytic skills. Right. And this program is basically preparing you for those types of jobs. And these are all I have to tell you, this is the main these are there's a reason why people call data scientists as sort of the the next big thing, right, I mean, one of the most highly paid careers are in data sciences. Right. And we also know, you know, people who are already in finance, also know that finance, again, is one of those highly paid carriers, you put all of this together, then you are a great you know, job candidate. Right, you are definitely adding to your skill set. So, yes. In terms of jobs that are plenty that are available. Yes.
Thank you. Thank you, professor. So our one question. Another question, which is very interesting, is like, How much time will it take to build our own like AI model, my own AI model after this program? And we have another question, which is related to the same because I'm from finance background, and when they compete with a guy who is from company to data analytics completely come from computer science background, how can it compete, because in the same program, we will see people from it background, and we will see people from finance background even a half of the same question. So how can this people from finance can compete with a guy who is from complete technical background?
Yeah, no, absolutely. And I think this is exactly where someone from this kind of back background would be valuable for a financial company. I have personally worked. So I mentioned that I have a consulting company, and I take students from so I also teach, you know, machine learning and AI. And I have taken students from that particular back background, more, but these students do not come with financial knowledge, right. And most of the time when I'm working on this problem, and I have certain tasks to do, they are really handicap because they know coding, they understand some amount of data signs. But when I tell them that I have this portfolio, and I want to try to find out, you know how to immunize the sport portfolio for these kinds of shocks that is going to come right now. They don't know anything about what I'm talking about. They don't even know. They don't understand how you think about portfolio diversification, so on. And each time, I find myself spending a lot of time educating them on the finance side of things, right. And I've seen this in the firms that I've consulted with as well, right. And this is why this program is this the exact reason When we thought about this particular program, creating this program, because we wanted to create this set of workforce, who have both the knowledge, knowledge of both finance and data sciences, that was the intent. That's exactly the intent. So,
Professor Alan, could you throw some light on the capstone project, which we are providing the program, like, most of the participants would like to know about the capstone project? Also, because it is completely hands on, right? That's, that's when they get that hands on experience.
Right? Absolutely. So that is actually going to be the key portion. And I, I, I'm one of those guys, you know, who spent in college, during the lectures, you know, most of the time, you're not really working that hard, you know, you ease off during college, college is meant for having fun, and so on. And come exam time, that is when all its lates, and complete panic, and you are trying to read everything together and try to do stuff together at the last minute. But you know, the idea of when you are preparing for the exam at the last minute, and you're reading everything, and I try to, you know, Master everything in the past two days, three days, then suddenly, a new kind of understanding happens, right, because all the concepts that were taught to you for six months, now you're reading all of them together and trying to put them together. And this is when suddenly pieces start to fall in place. Right now, the capstone project is precisely meant to do that capstone project is going to be there are a bunch of different different options that are available, you guys will get to pick and of course, if you have your own favorite problem, you're welcome to bring data and that print favorite problem also there. And what you will be doing is you will be putting together pretty much all the concepts that you learned and ideas from you know, probability, how do you start with analyzing the data? And you would have learned different techniques, some from supervised learning, someone clustering, someone optimization, how do you bring in all of these pieces together to accomplish this final business goal, right? That is when your true understanding actually happens, right? So we hope that you will attend all the lectures, you Don't slack off the way I used to do it during college. But it's not important just that you attend lecture, the capstone project is when you put in put together all of this thing, and that is when real learning happens. And plenty of options are available for the capstone project. That there's some of them is again, like the the particular project that I talked about the case study that I talked about, which is a allocation case study, where you are trying to allocate across assets, right? This is sort of an evergreen project, right? Different people handle it in very different ways. Right, what we did was we offered one solution to it. I mentioned, we haven't used anything about sentiment at all, in the solution that we offer, the same project can be handled entirely by looking at sentiment analysis. So data from a Twitter streams are available, right? You can look at that, you can extract that and look at those sentiments and see, can I look at the sentiments that are coming from Twitter streams, and do leverage analysis, right, perhaps add more money towards cash, more money to bond instead of equity, or maybe change it do the other way based on the sentiment and see how that portfolio performs right now, there are really 100 real world things that you can do with data science, and that is where real knowledge, real learning is going to happen. Thank you. Thank you.
So we have one more question from Ashutosh, do we have to train the model basis the market capture sector sector of the securities, since the classification may skew of the training of model
could do? Very, very good question. And so this is where individual creativity comes, right? What we are here is not to give you this magic algorithm that suddenly gonna make money, right? That's not the case. We are here to teach general ideas. We are here to teach you one way of thinking, no, this is like you know, when you do people ask this question all the time, right? I mean, if you have this magic algorithm, which can predict stock market, why are you even teaching it to us? Right? You'd be making money on it. The thing is, I don't believe there is a magic algorithm, right? Each one of us bring it so we are building in this knowledge base. We have this knowledge base, you take different ideas and construct it creatively on your own. Right. The way I think about this is like carpentry school, right? There is a people you know if somebody wants to become a carpenter Someone you know, the way you teach them is you talk about each individual tool, you say that, okay, this is a hacksaw. Here, this is the advantage of using a hacksaw. This is what it can create. And this is a screwdriver, and this is a hammer, and so on and so forth. Now, just because 10 people came from the same carpentry school does not mean that the final product that they create is identical. You can some people can create magic out of the exactly same piece of knowledge, right. And this is entirely where your creativity comes. When I was part of the hedge fund. I. The thing is that we were doing algorithmic trading, and we were created. We were always working towards trying to create new algorithms. Why would you have to try create new algorithms if the old algorithms are supposed to work forever. The reality is old algorithms don't work forever. The as more people find out about some particular hater, hidden opportunity, within a year or so the opportunity disappears, right. And this is where your creativity comes, you have to always be on your toes keep exploring new opportunities and and what data science brings in it's it's all it gives you a toolset, you gives you a way to think about the problem and come up with your own solutions, right. And that's what the capstone project will help you do, right you you are able to think about a particular problem and use the skill sets that you have learned as a part of the course and put them together and solve it for them.
Thank you. Thank you.
So we have another question from Mr. Dinesh Kumar, sahoo. How machine learning can help to avoid big fall what we saw in COVID situation? Do we have something which can help us to avoid investing in that time?
I, you know, all I can say is that there have been the fund that I created. This particular algorithm for the product that I created was a product that which they wanted to use for the the retirees, the retiree clients, right. The retiree clients were people who did not want to trade on a frequent basis, when you trade on a frequent basis, your tax liability is really high. Right? So these are clients, they want to make a decision once in three months. So a decision was made at the end of December, what should be my holdings for the next three months. And next time you're going to touch that decision is only in the end at the beginning of April, right? every three months, you think about it, when you're looking at that kind of timeframe, you cannot respond quickly to things that are happening around you. So that portfolio does not recognize it does not know in December, that there is going to be a pandemic, that's going to happen in the next three months. Right. But what this algorithm does do is when sorry, let me just show this. So what this algorithm does do so this algorithm actually missed out detecting that time the pandemic, right, and so the state missed out, it didn't warn with the with the red line, saying that that's a danger zone. But what it was able to do was after the pandemic happened, the recovery period, it actually said that this is a lower risk zone, you should put invest more money at this point. Right? most fundamental investors did not believe that they actually thought right after pandemic, they thought, you know, the, we still haven't found the vaccine. Why is the stock market moving like crazy, right, the unemployment is at Sky High, and so on. But the ML looked at the data. And I don't know what it recognized, it recognized probably the fact that government was pumping a lot of money, it recognized patterns. And it said that you know what, this is a load of spirit and go ahead and invest. On the other hand, if we were looking at strategies that were operating at a much shorter level, perhaps those strategies could have reduced leverage during the pandemic, right, but it depends on the time period, right. And this is what I what we are going to teach is concepts, not necessarily one spoon fed solution that works for across all the time. When once you learn these concepts, and perhaps the next pandemic next, next kind of shock event happens, you might be able to try and tune these ideas so that your portfolio is sort of insulated from those kinds of shocks. Thank you, thank
you, professor. So another question is how deep we are going to teach the concepts because by seeing that curriculum, the curriculum is in depth and there are many topics we are going to cover in the program. So the question is about like how do we are going to teach
so That is a very valid question. And the answer is, is, is that it's not going to be so deep as if you had taken, for example, a master's program in data science, which are run for three semesters, it's clearly not going to be that that be right. They will be way I am thinking about it is as a practitioner, right? My there are a lot of amazing innovations that's happening in both the field of finance in the field of data science as well. Right? And I don't know, all right. But what ends up happening is that when I have a task, when I have a business task, that is the time you sort of jump in, and you jump in and look at what are the current tools that are there, what are the current ideas that are there, and you pull in all of those ideas at that time. Now, I am able to do that, because my basics are strong, what we are trying to do is exactly that, we are trying to give you a reasonably strong basis. And instead of just talking about the basics, in abstract, we are planning to show you how to use this basic concept. Right, what will happen is when you get a job, when there is a new problem that comes in at that point, you will have to read up more create more concepts. But the thing is, what we have done is will be we have given you enough basics to be to face the problem confidently and learn the tools that are needed at that time. Right, clearly, a six month program will not go into a depth that a master's program, which is one and a half years, two years long, will do but this is a very focused program. It's mix us what in my opinion is sort of best of both worlds, right? decent amount of financial knowledge for the markets, decent amount of data science in order to accomplish most commonly required tasks for the job.
Thank you. Thank you. So another question from the banker, Mitra there are software's developed nowadays, that is all machine learning algo Sindhu I, for the course be redundant for future, especially the Python or coding when things move to gi.
Um, so I, you know, I admitted to you that I'm not a program, right, I, hey, I can program I'm not a programmer, right. And this would be a welcome development, if, you know, I don't have to program. But you know, what, I don't think we are going to be there for another decade, at least, right? Most of these tools are pretty good for checking things quickly. Right. But when it comes to actually solving a particular business problem, the amount, it doesn't have enough flexibility to handle a task that a particular business needs, right. So I as I mentioned, and I have a consulting company and like I encounter these projects, right? For me, my job is going to be lot more simpler if I can just use one of these GUI to develop it. And I still haven't found something that was going to reduce the overall development time. The key problem, always I found is in that translation problem, which is the business task, what ends up happening is that the business user gives you a problem in English language, saying that you know what, I have this, I want to create an investment scheme that is going to minimize tax impact. Right now, that is English statement that they have given. Now, how do I convert that into a math statement? Which I can plug into my GUI and get an answer, right? That requires you to take concepts, take that business statement and convert it into the mathematical concepts, and then find a way to program it. Right. And that is where the real real deal is, right. And that's what the basics of these concepts, this course these kinds of courses will be. I don't think we are there yet where a full GUI interface can do all of this magic right away. I don't think we are there yet. We will be still some some time. But that is not necessarily going to make you redundant. Someone with the basics is now going to be all the more empowered to solve business problems, not less, your value is not going to go down just because a tool came up you are the person who still needs to solve the problem. It's not the tool that's doing it.
Thank you. Thank you, professor. So we have few more basic questions where I can answer a few people are asking about the course fee. The course fee is three lakhs plus 18% GST, but we have a couple of scholarships. We have women scholarship, we have started scholarship funding professional scholarships also. So I'm giving you a number of you can just call up to that number and you can just get on the program later. details and scholarship related Also, and few people are asking about the bridge course also, we are going to start the bridge course shortly once you enroll for the program, and it will be in the weekends only and where we are going to give some basic understanding towards Python language right? And six months program classes will be on Saturdays and Sundays only. And what are the questions we have? I think that's it. We don't have more questions on this. So you can just call up to the number which we gave in the chat itself. And oppressor Would you like to take this question one of the participant is asking, after finishing this question after finishing this post, can the mentors provide guidance if you are interested in starting a startup company?
I'm happy to help thing as I'm as I said, for me, as I for my consulting company, I mean I used my students because what I find is that freshers always come up with better ideas, right? When people and I love interacting with students, and I'm sure a process Docker would mirror that. That thought as a part of Cal I'm sure he's mentored a lot. Lots of students in this area.
Oh, another question from one of the participants who is a fresher I think he's asking about the freshers. But you need to have at least one year work experience to do this program manager not. So yeah, I end the program. Sorry, the webinar recording will be provided in next 24 hours. So we will email you. You can just have a look at it again. And if you have any questions, you can just email us also you can call us at the number which we have provided. So yeah, that's it. If you have any other questions you can just ask. We are done with the questions. Professor. We can wind up the session. So yeah, thank you. Thank you, Professor Allen, Professor dagoretti your Thanks. Thanks for your time. Thank you so much. Thanks. Thanks, everyone. Thank you
Watch the entire interview here https://www.youtube.com/watch?v=yWLfzDHbW1U