11. Unknown Unknowns: Risk and Uncertainty

The most intelligent code that you would ever find from a politician is from the Rumsfeld, right? You have known, knowns known, unknowns, and unknown unknowns. I mean, it is something that you can use in any scenario anywhere and it would be valued. Imagine if your child grew up in a Suraj barjatya style household, where everything was sing song and dance and party, the chances that an adverse outcome can completely topple the child, the chances increase.

Your job is not related to your stock market rate. You think both of them are independent but what we do not know inherently is the probability of you getting fired increases as your stock market portfolio, goes down. If imagine that I'm doing dollar dollar against rupee dollar against China as an e simpler example, right now in current benign environment world, you would think that China and India

slightly. Yes. They are Emerging Markets. Therefore they are correlated, but they are not Not so correlated. That is it slightly, you know, it's like a dog and a drunkard with a positive 20 percent correlation, right? Hello and welcome to data. Shatter the podcast on all things data. This podcast is a series of conversations with experts and Industry leaders in data.

And each week. We aim to unpack a different compartment of the Just in case I am your host, Karthik chassis that I'm a blogger newspaper, columnist book author and a former data and strategy consultant at currently head analytics and business intelligence for delivery. One of India's largest logistics. Companies. You can follow me on Twitter at Karthik s that is Kar. Thi KS and read my blog had no into da.com. That is n 0e. N e, hu d a A.com or opinions expressed in his podcast, belong

to me and my podcast guests. And it do not reflect the views of any organizations. We might be Associated, nothing discussing this podcast, should be taken as Financial or legal advice. The fundamental principle, underlying all analytics and data science is probability and probability was first invented, or should I say, discovered to assess the risk? So what is the risk? Can we quantify and measure it? How do we handle risk in life?

And as risk? Always back, today's guest on data chatter is below and see that over tea. Whoo. Co-founder & Investment advisor at Raleigh Asset Management, a global Arbitrage fund balance was my classmate at IIT Madras, where he studied computer science, but spent most of his time gaming, he then went on to IIM Ahmedabad when he continued to game heavily and graduated with a gold medal. He now runs a hedge fund but still spends most of his time gaming.

Moreover, he was one of the last traders to trade on behalf of Lehman Brothers, on 15th, September 2008. Who's Gus, you can. Imagine is a vast subject. And so, this is a long podcast. We talk about measuring risk problems, with too much measurement of risk. How risk can be managed and all that. We also talked about movies games, the differences between poker and Bridge and physics Envy. Enjoy. Welcome to data chatter.

So I will start with a loaded question on 15. September 2008. You were working for Lehman Brothers. You were probably one of the last people to actually trade on behalf of Lehman Brothers ever. So what's your definition of risk? It's funny. I mean, I will talk you over. What actually happened on that day on Friday. I remember closing. Some traits, in Brazil. I was massively long gamma. I have seen bear Stearns fall.

I position my books, you know, saying that there's going to be a lot of volatility but a Lehman Brothers is not going to collapse. So but the markets will be volatile and as position and on Sunday night in New York, I was trading and by the time news was not over there yet and I was

treating my gamma first. I trade my positions on the machine, then I had to shout it over to, you know, my colleagues in Singapore saying that, you know, by 10 million euros said, until meaning used to trade the gamma then initially, what would happen was the machine was stopped. Then the trader would take immediately. Give me the like they shout back the price, you know, over that

line and it was instantaneous. Then it would take about a minute and after that it was like three minutes and 5 meters like what's happening. And you know and are later is like nobody's taking our name and that's when I realized that. Okay, it's all over and you know, so that's like your first

it, you know, real. Like you join, you join the company which is like one of the, you know, top five in us and you know, consider to be next to Goldman Sachs and and in the way in the trading mentality and then one fine day and even like a month before, right? You see your comes on the floor saying that we will win we will do back and you know, a month later that thing doesn't even exist and you are out on the road. Look, fine. Trying to find a job, right? So nothing better introduction

to risk mean. In fact, When I joined Lehman in 2007, right on my first day, when I was on the trading floor FX Market, that's when the sub, you know, mortgage crisis started and my boss. Ed who joined, you know, he was the head of Exotics at that point of time. He said you should watch these markets. You will never see such kind of markets ever again. And you know that I don't know whether it was an omen or what, but I have seen worse markets over the next 56 years.

Okay, so let's talk about risk from multiple perspectives here, right? Having, let's just take your situation on the 15th of September 2008. One, your company had collapse that you were out of a job. They can is that like that's on the personal side, s on the

market side. I think you have seen a bit of a rollercoaster ride over the last one year and a bit when you were there with Lehman and so on and so you see the markets do lots of funny things and so on and so you would have one of the things you would have had to do as part of your job. I guess would have been to sort of manage risk in whatever. You can we find it. It's all.

So maybe we can use this to sort of talk about what risk means in a financial sense, what it means from a personal life, kind of sensing whether the two are related. We should think about them in different different ways. You know, it's funny you mention that because the way we think of risk in personal world is very different from the way you think of risk in financial as well as even economic sense, right? I mean, when we think about, do you want to take that risk?

You normally We mean, is the downside worth the upside, right? For us in real, in English sense. English, you know, risk is always to the downside. Everything else is about working. You don't talk about taking a risk, when you're buying a lottery, you say that, okay, I will lose one rupee or ten rupees and then I'm get a crore, right? So that's the way to think, but in economic sense and in financial world, it's funny. But risk is essentially any deviation from your mean.

Whether it's positive, I mean, you had an expected outcome of X. The actual outcome is X Plus 1% or x. Minus one percent. Both of them are risks and be the whole object to being that. You want to narrow the range of outcomes, the way the banks or you know, any person. So you want to reach a situation where the risk reduction is nothing but reducing the scope of outcomes or narrowing the range in which the outcomes can

happen. Right. So that's I think is the fundamental difference between the way we think of risk in the normal World versus the way you think of risk in the financial world. So in in real life, let's say, for example, you getting a big bonus is not a risk, but from the bank's perspective, the likelihood that there's some small probability, where you can get significantly more revenues, for whatever reason. That's a risk, which in some sense. You might be right. Yes, you might want.

Write your book, you would want to, you would want to trade off that small probability of getting a bumper profit. And for let's say, more assured smaller number, right? So, that is a risk in some sense. A casino is taking the opposite side. Where you might have to pay out a lot of money for one guy, but you are getting in a lot from smaller Nichols.

You're picking Nichols from everyone. Yeah. Okay. Okay. So from this perspective, right? I mean like I remember reading this tweet sometime last week. I probably sent it to you as well. I think it said that like the risk once created can never be destroyed, it can either be transferred to somebody else or it can be managed or it can be ridden out at the I'll probably link to that. We to the show notes. So can you talk about the creation and destruction of

risk? Is it something like entropy in a sort of thermodynamic sense? You know, it's funny that you mention that and I mean not many people are going to like it when I say this, but the whole field of finance and economics suffers from physics Envy. It is a great desire to associate certain principles and ethics to, you know, the way you want that method and Madness, like the whole scientific development which happened in 18th and 19th and early 20th

centuries. And you had this physics laws and mats and principles. And, you know, the Boyle's law is ranging from Newtonian mechanics to boys laws and everything. It was some method to that so-called Madness and we desire and economics as Acted and, you know, ranging from Samson's and you know, 1920s and you know, when you have the micro economic demand Supply graphs to everything. And coming back to your statement. That is actually, I don't think

it is true. I mean, I remember reading that and you know, I had a counter to it. I lend money to SK, right? I have created the risk. Mauro if s case gives me back that money, that risk is destroyed. Right? So that it is not like once created. It can't be destroyed. In fact, if you want to look at energy, energy is neither created nor destroyed. You had hold the Big Bang thing from energy to Mars and back to energy. It's like a closed ecosystem, right? The other thing I should tell

you, right? This is actually Lehman, Brothers is a phenomenal example of what risk. He's because, you know, on the Monday morning, when I went to the office and for one month, I just went to the office. No Scouting For What is going to happen?

We didn't know searching for jobs, but we were watching the markets but we had nothing to do and I was very happy that it happened because on the other side on the weekends at Barclays and I know people because we spoke to dimmit, you know, and Credit Suisse and Barclays. They were worried because they had no clue what to do with their positions, with Lehman. Right, my risk was done. Right job loss and everything, but I knew the outcome and it is done.

Risk got eliminated outcome has happened for those guys. They didn't know our dick positions long, if they are sold and options, but they still short, or since Lehman got eliminated, were they did positions vanish and suddenly they became long optionality, you know, nobody had a clue. Right? And one of the most I would call it is like a back-end operational Innovation, right? Imagine. You had three Banks, a b, and c and you know that Banks do derivatives amongst themselves,

right? And a derivative with b.b. Derivative at C and C did with a. Now effectively, you had three units of derivative exposure. And for some reason bow, all of them off. So, let's say a soul to be, be sold to sea and sea salt. A now. There's technically, you know 3 into 2 6 units on banks balance sheets, right now, which also means that you have to post, mark to Market and all those things. What banks did is because they were worried about this Lehman thing, take the inventory, says,

there was a system. I mean, we're all the banks would post their positions, anonymously to have a central portal, and the portal would basically, then try to endure. It will also post their valuations for it. Right? You think it is worth X and some other bank thinks it's worth. Why? Right if X is less than y. Then you can net off. Right, but imagine it is not. Then you can. So they would basically in men do and try to reduce this counterparty risk, right? It was, it was not their pre,

you know, or rather. It was not into a big way. This kind of a reducing your risk by sending it to Anonymous portals and this thing and some banks actually did that very, well. They actually did funding arbitrageurs and all those things, you know quite well, but the point is that was one kind of a risk reduction, right risk was there one day before? This didn't exist the next day.

So, in that sense, know, I kind of disagree with the point that risk can neither be created nor destroyed, but it is true that while alive, and there is no, there is always a event which might destroy it. But while it is alive, you cannot, you know, there is limited formats. You can only more fit and create it. So the usual entropy principals who apply as long as it is alive. Yeah, I think what you mentioned about economics and finance having physics.

Envy. I mean, the first thing I I was thinking of was about how pretty much all of our quantitative Finance is it it's based on something like a heat equation, right? It's live in its they try to escape physics Concepts in pretty much every way possible among the other things. Do you mean the wonderful thing about statistical mechanics? If you think of it is that the movement of gas is it sort of follow it, always not sort of it

follows a normal distribution. A lot of physical phenomena, if you look at the distribution, it's always normal normal as it that bell curve distribution. What what that does to us, not just in finance or economics, but all of us in real life is that we assume that everything in life is normal. So, one of my favorite things is that, like, people think that there exists this thing called a middle class because they think

wealth is normally distributed. And so that everybody in the middle is, is the middle class, and then you have the Waiting for the rich in the lower tail for the poor. But if you look at any data you want that, like, for example,

wealth is not distributed normally or if you or even like sort of payoffs from different events are not distributed. Normally, even though we try to sort of suggest pretend that they are normal. So can you talk about this, think of, like, why is it that, like, we sort of always think of things are being normal and like, how do you deal with? How do you deal with long tailed distributions in terms of risk, and like who are, which is risky?

And how and what are the sort of some of the mistakes that people do in terms of assessing outcomes? Surely that is very interesting question in the sense that because as humans, I mean we came from a long evolutionary process, which is Primal right at the end. We are all monkeys, right? So and all nature, as you pointed out whether it is near normal or normal, I mean, Most of the distributions, follow the bell curve, right? Whether it is hides or you know, and so on and so forth.

So and secondly, what happens is, survival wise, you know, you prepare for the more you prepare for the mean, right? You don't necessarily prepare for the tail and which is why human civilizations have, you know, you have those Cycles, they are big, they grow, then they get wiped out because you're, you don't Normally, you expect things to be status quo, right? In in mathematical principles. You think everything is a Martingale? Yeah, can you explain articles for our business?

Sorry, not everybody can. Yeah. So Martingale is a funny mathematical process. Where the expectation is that for your future is you will be what you are right now, right? Expectation of X in any time in future is equal to X of 0 which is current time. Right? It means that and it also is one of the reasons why you have inequalities and all these things. It's goes very nicely and feeds

into the situation. Because if somebody gets lucky and has done, well, it is likely that that person will remain, you know, will continue to stay. Well, in fact, a certain amount of autocorrelation. But at least what it means is that people who have got lucky, right? That luck stays with them. It doesn't get normalized. Right? Whereas, overall, you kinda expect that things should get normalized.

It doesn't right. So this is where that jump from the moving from a typical normal distribution, where we think we are in a bell curve to Somewhere where you are, moving into a Martingale, where expectation of X is a future X is now to you have these long tail distribution. So, where the whole translation comes is the more fictitious the world is away from the nature where it's not exogenous variables, but it's an endogenous variables.

Right? Like I created economics because humans created a whole bunch of laws Way by which we created ecosystem, under which we are all talking to each other, exchanging goods and services and all these things right now. That creates the phenomena, which moves have, it's not normal, right? There are network effects. There are, you know, one thing leads to the other, like there is a super star in a movie or a pop singer. Who becomes popular popular bream brings in more popularity,

right? We know that it's equivalent to your old saying, Rolling Stone gathers, Mass, kind of a thing. Yet. I know I said it the other way around but the plot makes its each other there, the Bible, I think that like and also like in a popular Govinda song. I think we're like, if you have a lot, you will get more, but if you don't have anything like, you don't get much. So, exactly, right. So a lot in the ecosystem is

normalcy happens. When there is what you call Independent identical distribution, right? So the future event has to be independent of the current event. Then obviously, You will have that, normal distribution is coming it. But most of these things are, you know, it is like what it like you're going in the song, just go, it's a multiplicative effect. Right? So, people who are rich will get richer.

Even if they returns like, somebody who has one crore and somebody who has let's say 1 lakh even if the one black guy generated hundred percent return at the end of it, his net worth is 2 lakhs. And even if the one crore by generated, Return. It is, file acts. The Gap has only widened. It did not shrink, right? So it is this for, so in general, in life in finance is obviously slightly different, which you can talk later.

But there is deviation from your normal distribution because of these Auto correlative effects because of this endowment affects all these things will come into picture and then you have your typical Network effects, which come in, right? Your YouTube the popular guys will remain in fact, technology. Elegy. If anything is only accentuating this thing. So what it ends up is, you have lots of these tail distributions, right? And in finance, that is also true.

Because what happens is, Imagine today, stock market is there, right? Stock market moves down 3% or 5%, people will not panic, but it goes down to 20%, people will panic. And that Panic itself creates further Panic. Yes. Yes, right. So there is an acceleration effect, which happens Beyond a certain threshold. Right. A, you know, in your olden day in physics when we were younger and we studied, we had this pendulum swing where pendulum is pulled to the center.

But imagine that there is a pair, a pendulum where it goes to an extreme, but that itself pushes it further before it

comes back. Yeah. I mean, you can actually imagine that railing sort of like the Rope, kind of bending because of the momentum of the pendulum or something great, which should opinion, cancer physics book. So So that is what creates these long tails, right? I mean whether, you know, and we have many distributions which talked about it, right? Ranging from you know, exponential's to poisons do all sorts of things, which talked

about this. And in fact, one of the first things that even when in finance you have options when they were priced, they all assumed or normal distribution Brownian motion. And then you realize that, okay, if I do it that way, things don't work, but you don't adjust your model. What you do is you introduce, you know. No, workarounds to say that, you know, that model is becomes stops, being a model and starts

becoming a language, right? And then you have in volatility the Tails and uh details details are always more expensive than you know, the volatility of the Tails is higher. So all those things come into place and you know, and in the true, we all would like to have like in general why do people now like startups? Right startup is nothing but a long-tailed outcome where you know, you suddenly have Fact, even in coming back in finance. Like, on one side.

You have a hedge funds where they are trying to make a steady set of returns, right? So what they want is they want a narrower range of outcomes on a consistent basis and they believe that that compounding will carry on forward in contrast. If you look at VCS, all they are doing is that they are betting on long tails. Right, they will bet on 10 companies and say that. Okay, if one of them works and that generates 10x or 20x,

that's good enough. Yeah, actually, I was just thinking about it while you're explaining now, if you think about it in real life, most risks are sort of like highly skewed risks. So for example, there is this risk that like, you are driving from place a to place B. There is a risk that you might meet with an accident or there is a risk. There is a risk that whatever happens like there's an earthquake at your house. Abscess, and you lose your house.

It's always like, sort of the, and none of them. And in some sense, a lot of Life, which and life itself. Like you have one life, which means that, that, that is like sort of fundamentally alleged, right? So, this is really an issue. And also, on the, on the upside, on the upside, you in a, let's say, you start a company. That's a, that's a sort of a right-tailed, kind of a risk in a sense. Because like, in the worst case, you lose what you put in, in the best case you Make a lot more,

right? So so I think, like, in that sense. I think the real life everywhere. You. Look, you think you, you look at all these sort of, it's all long tail. It's all tail risks on either side of the tail and that makes it fundamentally different from Finance where they still is exists, but small compared to the volatility or standard deviation as you call it there.

And before we Sorry before we go ahead, I think like a we have sort of Digital ahead and so on the from so we can we sort of like stopped for a little bit and they say talk about things like tail risk, right? A left tail. What is volatility? What is uncertainty and so on and then we can sort of come back to come back to this. Don't see mean. Let me segue into that your question that you asked and start in this thing, right?

So, starting with it, is mean the most intelligent code that you would ever find from a politician is from the Rumsfeld, right? You have known, knowns known, unknowns, and unknown unknowns. I mean, it is something that you can use in any scenario anywhere and it would be valued and it is to understand risk in some sense, you know, disrespect. Like no knowns will be something like tomorrow. Sun will rise from the East. It's a known. Nothing is going to change it, right?

Mnsure few billion years later that might not, but it's a known known for is, you know, right. A Known Unknown will be that. I'm going to go. I have to go to my office tomorrow, but I don't know whether I will reach safely or whether I come back right? There is a probability right? There is a it is an adverse outcome can be there or something or Or I'm going to a casino and I'm playing blackjack, right? There's a probability of me winning or, you know, crafts, whichever way it is.

And then you have things like unknown unknown, right? Which is what you would call as uncertainty. So to speak. And the reason why you have clubbed it as there, as you didn't even know that this thing. I mean, for example, for all of us we can say that Source, covid-19 unknown unknown right now. It has become. In fact, if anything this kind of tells you that all these things are Spectrum's. Right there is really no

watertight compartments. Covid-19 Known Unknown to all of us except where, you know, except maybe a few people who wrote books on pandemics or who are doing research on it. And so on. Similarly financial crisis was an unknown unknown. For most normal people accepting to a few people in finance who actually watched it. So here you can again, see the distinction, right? So what is risk to one is in some sense and uncertain thing, too. The other, let me stop you

there. You said, what is risk to one? Is uncertain two together. So, what is the weather? Where are you drawing? The line between the risk and uncertain, or is it a spectrum? Again? It is a spectrum, but very crudely speaking unknown unknowns so something which is unmeasurable. Eight and, and, you know, or even not even known. Because if you don't know about it, you can't even measure it,

right. They follow in, you know, uncertain area risk, broadly speaks about things that you have some ideas, some estimate and let again taking the example, in December 2000. Not December. I should say in September 2019. Source covid-19 unknown unknown for me by December 2019. It started moving for me from being an unknown unknown to a, you know, Known Unknown. Like I knew of it, I was position for it in January.

I mean for in on a personal level like I started buying the sanitizers and Claude Claude masks and 95 in January right on my portfolio in my fund. I was positioned saying that, okay things are going to be blowing up this Thing. Because that's when you saw a China numbers. And, you know, I was, we were tracking that on a daily basis and it is there. But for still most people, it was not. Yeah, so I mean, I remember, I read my first orders and I spoke to people around.

I told the family and they were, like, what crazy thing are you talking about? Right. And by March, obviously everybody knew about it, started impacting our lives and so on. So the obviously what I mean is I'm trying to draw the line at unknown and unmeasurable stuff. And that is where I would call as uncertainty. Because even if you knew about it, what the hell, what on Earth would you do? Write it somewhere, it becomes measurable where you can take actions?

And it, and it, then it becomes risk. I mean, to be honest. I will be very honest with you, but it is risk and uncertainty and these things are terminologies, like God, you know, that elephant in the blind man story, when things are different. It's the same thing, but different things to different people, right, but I'm just giving you my perspective of how I look at it. Yeah. Okay. So you would so that brings us to like you were like when you can measure something.

You said you can do something about risk and when you Talking about doing something about risk. I think we'll, we can now talk about this concept that we know as hedging, which is sort of, like, sort of, I don't know if you were to call it as eliminating risk. I don't know if you were to call it as sort of reducing risk or doing something to take care of your disgusting. Broadly, we can, we can Define it as so.

Can you talk about hedging and like, the again, both have been pretty much this standard for every question. I'm going to ask you today. I need you to answer both from a life perspective and a finance perspective. I mean, that is a very good concept, right? I mean, inherently hedging is part of our life, whether we knowingly or unknowingly, we do it it. But when you hedge before hedging, is the other aspect, right? You have this concept, which is

expected value, right? Because hedging is an outcome, which is changing the expected. So, in a very, crude way, expected value is like, the, like outcome that you are hoping the outcome and then you have a range or a distribution of it. Based on which you are making a decision and hedging is a process which is essentially narrowing those outcomes, right? I mean expected value, I mean and to do that you need to First understand the principle of expected value first, right?

And the understand the concept that expected value the the same event has a different expected value to you to somebody else or even to you at two different points in time. I mean t taking an example, you know, where you it's a Saturday evening and you want to go to your Ends. You, you know, you could go and watch a movie with your friends or you could stay back and watch a movie on Netflix.

Right? And the expected value of these two is probably if you're an introvert, you might prefer the latter. If you're an extrovert. You may prefer the former, right? So that's where the same even has two different things. But independently the, for the same person whom I let s take an example of an extroverted person. If imagine that the second event had it with his or her romantic interest. Right? And the second one has more value, right?

So your expected value, inherently starts taking into account, your utility functions, your. You know, if we didn't have this dispersion in expected values, you would not even have a functioning market, and then you are, you have wrote between the buyers and sellers book, right? I mean, the buyer and the seller, have different value for the same asset. Otherwise, there is no trade, right? Of course, you I'm buying a Samsung phone because Samsung values my cash.

More and I value the for more exactly. Right, right. So now and economics, when they go into this and they say that this all agents are rational or these things, it is true to some extent, right? But where they do a mistake is they kind of say that, you know, every average person and they take an average person and then, you know, remember that in our Junior days we used to get this, you know, what is f of X and expectation of f of x. And F of expected X. Yes, right.

And these two would be functions which are not for some of them, they would end up being true. And some of them they are not true. Right? And that is, and that and we make a simplification that it is actually true. And that is where most of the problem is, comes in. Right? We have these laws and, you know, taking the adverse situation, write my law, my utility function from committing, a crime is different from a Psychopaths, utility

function from committing. Being a crime, I'm more scared about, you know, being imprisoned in suffering, the adverse consequences than let's say a person who takes Pride out of it. And therefore the deterrent for that person is very substantially different. Right? And this kind of we can talk about it later. But this kind of goes into the fact that why a uniform policy that you have will not be good enough to the say, apply as a deterrent for everyone, right?

So now, for me, my actions is, I'm now going to narrow these outcomes, right? I want to reduce the like if I want to go to a movie. I want to go to a movie, which I think I will enjoy, right? I will unlikely to take a, you know, being let's say not a are three kind of a person. I am probably going to go to a generic movie, which is likely to have a moderate range. Right? Whereas let's say that the art movies tend to have a wider

range of outcomes. We're one of them to be either extremely good or it can completely go over my head and I won't even understand. So that naturally means that I will go and up going to a artsy movie, not a Nazi movie, but a normal movie, right? So hedging is nothing but you are trying to die, reduce the range of outcomes, so that you have greater assurance that you will not. And it feeds into the point that our utility functions are unfortunately not linear, right? A negative reaction.

The pain hurts us more than the joy. Actually, this utility function reminds me of, like, I think cubs, we have also in some of my past lives, have done some work on utility functions and using that to construct portfolios and things like that, but it reminds me of Behavioral economists who kind of like once I had drawn a utility function. It is very hard for me to sort of think about it. What is what is behavioral economics?

Because C4 fundamentally utility function is like, x axis is how much you have lecture income, or wealth or returns or whatever y-axis is the utility. And we know that it's a sort of a, the, it's an increasing function but increases with the - diminishing slope, right? So I don't have to try to mix or concave. I think it's like a, that's

something I never understood. So based on this, which means that if I have 100 rupees, if I lose ten bucks, then I'm going to The feel a lot more pain than I would if I were to gain 10 bucks study from hunting, right? That's a, it's a sort of fundamental, sort of sort of everybody had.

I mean, I may not have liked the exact shape of the utility first curve for me, might be different from it is for you, but but the fundamental the nature of the curve that it's an increasing function, that it's non decreasing and that the slope is decreasing. I think that's two. 22 out of

everyone, right? That is true for everyone. Absolutely. I think I mean there are very few instances where this would not be like, I mean other than compulsive gamblers and few other areas, I think in general it is true. But again, this is where, you know, this is made actually the good aspect of risk management that you mentioned whether in real life or whether in financial Market comes in because net net Equity function.

The way we learn in economics is like a simple single dimensional axis, but risk overall is like a Three dimensional things in many ways, right? For example, if I were to be sick and I needed money money right now. The sickness is defining, and the utility for money is far greater because it has a long-term impact on my survival, right? Whereas, if you are a guy who is like happily sitting on, let's say some Surplus cash flow,

right? That 50,000 Rupees that, let's say your your driver, who is sick and was you and whereas for you. You who is like, does is marginal, you know, it is probably going to be in your bank and owning a 4% on FD, right? That is the value that it has so substantial, you know, the same

thing can have substantial. Don't both of you have a similar utility function, but because you are in a different on different dimensions, whether it is in time or objectives, or Etc. The same object gives you different utilities and this is where risk management happens, right? So, we all have left tales in real life, right? All scared, We Buy Houses, we

buy, we have jobs. They're worried about our families, not all of us. Always have dual employment, or not offers even jobs, which give you a guaranteed secure employment. Like there is a reason why government jobs get, you know, even for a government job of a sweeper, you'd find one is 2000. Kind of applicants or one is to 10,000 applicants, right? Because there is certain security that comes with a government job with a pension, and all those things right? In some sense.

That is like, the risk free. Rate, it's like a risk-free job, right? And yeah, so coming back you do so that this multiple times, you know, rigidity that helps you in diversify. You can actually transfer risk from one person to the other person and you people have in finances. Let's say that I am a guy who has as a Trader when I used to trade options and derivatives in general. I don't manage Delta risk. I am always Delta hitched, right?

Which means that I do not. Are about the direction of the underlying asset. Most of the risk, which I would manage was either volatility risk and correlations risk. That is the risk that I was comfortable with and I would manage that right. But I was transferring the Delta risk to my spot desk, who is comfortable managing it, right? So in some sense our utility functions are different and we are able to transfer in real life.

Insurance is like that. All of us have loans, all of us have made Health in issues and liabilities, which we have, we take long. Term liabilities, you know, in hope that you're all live long, and you will all prosper and Etc. But as you said, right, we only have one life and in fact, it's a very different problem. I'm whether if this, right, I mean, you are not is ergodicity, but we all have one life and we are very scared about it. So you kind of need to hedge that aspect of it, right?

So if you take a housing loan of about a crow and your income is about 15 lakhs per annum you want to make Sure that if something to untoward happen to you tomorrow, that loan burden doesn't come back and bite your family, take an insurance. Right? And why is it? Because the insurance person, you know, for him? It's a different risk altogether. He's taking your risk and converting. So this is where your original problem, right?

People are able to convert risk from one, type to the other type, right and bpay because it's a left tail risk. Despite its expected value. Let's say is low. Tea and is only about thousand or ten thousand, right? We end up paying twenty thousand because we are so risk, averse and we, it's the adverse event will hurt us so much more, right? All of us have spent all our money. I mean, we can give the best event on risk is happening right now, is covid, right?

I mean, we don't trade on probabilities, vitro, we trade on our fear of the probabilities. Yes, yes. Yes. Yes. So, yeah, so that's, that's, that's essentially how risk management works and hedging Works, which is you want to keep those things that you are comfortable managing and you do not want those things which are in finance obviously because you measure it even in finance, their lots of unmeasurable stuff and you will see very frequently that despite the claims of Finance industry.

They always surprised right and reason being that none of this them are good enough to First, and the second third, fourth order effects, and these systems, like, I mentioned earlier, right, the tailed systems are those where the second and third order effects are again, talking about utility functions and because they sort of flattened out, I think insurance is, of course, one of the one of the made that the rate of Returns on insurance is low, is one of the things, one

of the outcomes of the galatea are flattening utility functions. One of the other things is that, like, if you have a certain event and an uncertain, if you have a certain pay off, And an uncertain payoff and the to have the same expected value. You would be willing to pay more for the certain payoff, or for the more certain payoff than for the uncertainty of. So in some sense.

I guess, one thing that we have, sort of, all of us have internalised is that when there is risk, you need to just be compensated for the existed to take on the risk. You need to be compensated just for the existence of the risk. Absolutely know that that that that is absolutely true. Right? Imagine.

You had a job offer, right? From a multinational company, which probably is paying you X and second one is your have a job offer from a start-up which probably is paying you less than x. And in the multinational companies, you have your standard utilities and you know, fringe benefits, corporate benefits. Good lifestyle, Etc. You did the whole ESOP structure is created to make that happen so that the, in reality, the expected value of salary from overall, or at least, perceived

expected value. Because we never know what is true. Expected value. Anyway, right? The perceived expected value from your standard company is X. And the Y that you get from your startup. Y has to be greater than x. Otherwise, you won't even consider it. Right. It is a different thing that a more risk-averse person will need far much higher y than let's say you who might be less risk-averse.

Eight, which is why you would see all people, in startups, end up being those risk-taking people. It's a natural outcome. And which is why organizations, which are slow moving and sturdy. There isn't that? It's like birds of a feather flock together, right? So, there is a natural outcome of each person's utility functions. Whether we realize it or not, it inherently is what happens. Yeah. So I guess it's like not only does risk have a price, but the price of risk is different for

different people. And I varies from risk to this, coming risk of income has one price for somebody risk of

something else. Like, I don't know, like stock market is return on investments will have different prices for different people and and so on, I guess so. I know that you are a gamer. You are a massive gamer. I know you pretty much spent most of your college Life gaming and so on. And I think there are some games which are like coming, we had another episode of couple of months back where we had this guy.

Rahul dravid, who runs a Montessori School, who spoke about how one of the ways in which you can introduce analytics to young kids is, by introducing them to games. He was like one of the statement. She said in the podcast, is at Blackjack is absolutely compulsory for 60. Let's I agree to that. Let's talk to. I know that clicker. We'd like you also like sort of, I do you play a bridge. I don't know if you play poker and so on. But let's talk about games and the risk in them.

What we can learn about risk, some games. And what is the limits of what we can learn from? Like a situation such as games. You know, I mean the point that you're, you know, the previous podcaster has just mentioned in terms of risk and understanding games. It's actually bang on and it is see. I mean, it's not just I mean, I'm going to take a little detour, you know, that all our body has chemicals, right?

And we have been we have different chemicals for different things, your serotonin, oxytocin. And during stress. I forget the name of the stress hormone right now, but It starts with dopamine serotonin, oxytocin the end. The bad one is the bad. One is something. It's not idling, but it comes from the same region and it's it is thing. Yeah, so there is a stress hormone. I think it starts with k. I just can't. It's at the tip of the tongue and I can't remember, right?

So what typically happens is your flight or fight instincts, are generate these stress hormones? And the point is It in mother's womb. There is a reason why they say that mother has to be taken care of. Well, they should not get stressed and so on, because what typically happens is when the stress hormone, when it gets produced the baby's, it adapts to certain level as a base and that starts hurting future.

So inherently all of us, we don't like uncertainty when there is uncertainty, when there is lot of water, you know, things there is chaos. I mean, which is why. When I mean, there is this word we get Just right. Stress is triggered by anxiety, uncertainty, and all these things. And the ability to handle stress is not uniform, amongst all individuals. It is a very wide Divergent things, right?

Which is also leads to the things why some of us, you know, Depression hit someone the same event, make somebody depressed. Whereas somebody not and there's also a lack of understanding all these things fit together and in some sense, introducing your kids to uncertainty at a The rage is like a vaccination, right? You should think of it in that

fashion. Right? Imagine. If your child grew up in a Suraj barjatya style household, where everything was sing song and dance, and party, the chances that an adverse outcome can completely topple the child, the chances increase. Right? And you and I, we had friends whom we lost because of such outcomes, right? I mean, I mean, that is one where we know that is a specific group of people. And we have spoken, you know, many times outside, you know, about this event.

So so so mi, do you uncertainty introducing games, which have uncertainty, I feel, is a way of vaccination games by themselves, are could write, unless again, is what I call a complete information game, right? Which is things. Chess where you can actually compute boo, but even chess does not become complete information for a child because he's not a computer. He's not a computer, right? So she would have information is very large. So he's very large for them. So it for all practical

purposes. It can be considered incomplete information, right? Because you can't guess and so on and so forth, but the point is all these games help you to handle uncertainty because you we are all taught especially in your moral science and other things. Things do good. Good will come back to you and everything is almost like a definitive statement. We are not handled probabilities.

The fact that if I say that tomorrow in an election, some party is going to win, 300 seats, you know, if it is greater than 300, you are proven wrong, if it is less than 300, you're proven wrong. I mean, you are an election forecaster. So I'm sure you got these feedbacks, right? And what people don't understand is, there's a difference between talking about the mean and talking about the outcomes. Expecting that doing the same thing will lead to result to the same actions.

Again is something that you would want your child to not. You know, it's actually risk management for a longer life to introduce them to that concept, right? Because natural world, is you cycle fast. The cycle goes, faster, you break it. It's consistent. Most of the world. We mechanical world that we live in and interact with on a day-to-day basis. As Unser remove like imagine financial markets, right? We consider somebody who is Right 60% of the time to be, you know, generating, awesome

results. Now, imagine if your car starts 60% of the time, would you even own that car? Right? So there is a huge expectation difference on what we think on a day-to-day basis. What we think was is this thing and that because we are so used to certainty. We kind of think that the same actions should repeat and especially this happens with

human interactions and humans. Seeing an overall in events that, you know, the things like your Butterfly Effects and all sorts of things, which bring back the uncertainty. So games, I think, which have uncertainty are absolute even starting from your Snakes and Ladders, right? They are good because there is no like not. It's not like it is luck and you're not most people currently focus on chess and other things which have skilled, but actually

it is not skill. I think that is more important. You want to go for. It's not efficiency that you should aim for. Her in childhood, especially Ukraine for robustness. And that is risk management. Right? Robustness is nothing but your ability to handle risk. Well, thank you currently are suffering from covid because we've been trying to be too efficient. Yes, it is. I'd risk management. Yeah, they got a video every

day. You see the paper that some Automotive manufacturers is slowing down because like of the chip shortage. And that's because of, of the just in time, over the last hundred years, right? Like if we had more stocks, if we were more Fact, in some sense. You'd have a lot to receive less risk and be a lot more robust. So what are the things you told about how? Like, by introducing some risk to kids early, you sort of, inoculate them against this great. So, I'm reminded of this line

from someone of scallops books. I don't know which one it is a, it's called possibly Black Swan where he says, the countries where I would least expect a political crisis, or a coup or something her countries, like India and Italy. And some other places. He might ease because these are countries which have like you have Alexis you have high political instability all the time. Because you have high political instability. It means that you are like sort of.

You have a lot of short-term volatility, which means that you are sort of hedged against, like, big changes. Another example of this is I used to ride a motorcycle. I used to have a Royal Enfield and I had gone on a tour to Rajasthan in 2012 and their people taught how to motorcycle on Sand. And what infield guys told us is that you should always hold a bike loosely. You should never grip it too tightly and let it wobble a bit.

Then if it wobbles a bit, you know how much it can wobble in it. It won't wobble enough for you to fall here it. Absolutely right. What you are doing. Is it you're converting things moving things from unknown unknowns to known unknowns, right? And if you have known unknowns, hopefully your robustness in handling, these things matter like that is why I like it and games like games like on an investment side. So that is on real life.

In real life, I think people should introduce Blackjack is a perfect example. I think that is simplest and most elegant game that you would do. I like slightly complex games. And, you know, for example, Seven Wonders is a good game where there is certain amount of Randomness and, you know, I played with my kid and it's beautiful and they get excited and and and so on, but I think Blackjack, I mean that's a beautiful example, right? And the emphasis on things like chess.

I would, if anything I would argue should be lesser. ER, than compared to these blackjacks. I mean, you should play chess. That's a skill game. But overall you want more and more of these things. And there's also a certain difference right now, coming back to the other hat that you asked me to join, which is on a financial hat, right? So, there are two kinds of games, right? One is what I would call Al classify as poker. And the other one is, I would call as Bridge, right? In a poker.

Yeah, sure. You have an edge. If you know, certain things you can keep your accounts and so on. But a lot of edge comes from behavioral patterns, right? Your ability. So there is a significant role of individual and individuals personality which comes into play in in that game Bridge. On the other hand, especially your duplicate bridge that you know, we all used to play, you know, it's a different there.

There are opposite. So you have more meth So there is a method to the madness, right? You're trying to be consistent. You're trying to replicate, you know, uncertainty, but you are trying to replicate a certain process over a longer period of time. And you think that overall that narrows the outcomes, right? So again, they will meet, there is a Midway, its Spectrum.

It is not uniform, but on the right hand side in Bridge is what I would call more and investing style and left hand side is what I would call more trading Style. And you know, it actually feeds into that. I have seen historically in in my experience like most of the Traders whom I would not you know, who know the things. Well, the Traders are. Those people who actually know their behavioral and behavior of the other people around them.

They know the behavior of the market like they trade by gut right and on the right hand side. You have Kwan's who sure there is obviously a basis in everything but they're you know, the Renaissance and all those things ranging from quants all the way. What you're doing is you I like my as a Quant. I like more trades. In fact, I like short term trades more trade because then I'm moving back my distribution and narrowing the outcomes, right?

So that's the investment hat on the games and but playing uncertain incomplete information games where the same sequence of events. Don't lead to same outcome. I think for the kids. It's it must So actually I have this Parks, multiple thoughts for me. Actually one thing where when you spoke about Raiders and quants being partial to Bridget poker am reminded of this story,

again going back to the weekend of 13, 14 September 2008, where apparently they were trying to contact Bob Diamond. I think who was the CEO of Buckley's and he couldn't take the call of the other Bankers because he was busy playing Rich. That's the story in too big to

fail. And yeah, I'd also like in some way like in Bridge The uncertainty that you deal with is the uncertainty with the of the lay of the cards, once the cards have been laid irrespective of who, your opponent is. I mean, some might like to finish more than the others. Some might like to play for something else, but more or less you're playing the board rather than playing your opponent's. While in poker. I guess you're playing your opponent's far more than you're

playing the lay of the kites. Absolutely that. That is true. Right. I mean also the number of cards like in a bridge, all 52 cards are on the deck right around the table. You have it and you pay or on the air or on the table, right? And also, I mean, you know, that the bridge Rue timing, right? A typical Bridge games takes about 7 minutes and they say that the first three minutes, we'll probably spent in bidding.

And then they say that, you know, once the lead happens on the dummy comes down, that is when the declarer need supposed to think, then that is when a maximum amount of Means a lot for everybody because everybody is thinking but afterwards in a professional, you know, the cards the next Forty thirteen cards, which go happen in a jiffy, right? And you also decide it's not like in Bridge, you always pay the probability. In fact, we are taught that in

Bridge, you play to win, right? Even if it is a low probability. There is a one way to play. You just play that because you're playing to win but that is subject to actually stop you here because you need me to this other concept which is of what I call is finite and infinite games. It's a beautiful book by the Written by a guy named James casts. He was a hero energy and I think at Harvard, and it's a short book, but a beautiful book.

So basically the concept is what the title suggests is. The finite game. There's an infinite game in a finite game your you only capture you're playing games one at a time. All that matters is for you to win the game. In an infinite, gave the objective of the game is to just continue to play. So it's some things in Bridge, for example, even if I don't know, like, even if I let us say, we are playing a sort of a event or something, even if I go seven down or whatever.

It doesn't matter. I start the next hand at zero while in poker, if I lose all my money, now, I can't play the next time, so it's a sort of a infinite game in that sense, right? It is in fact, that's why you change the way you play Bridges. The, you know, there's Swiss pairs versus duplicate. You play slightly differently, right? The risk of going fi down is much bigger in imps in, you know, in duplicate bridge and it

will hurt you. Whereas, you know, my 14th ranked, if I'm let's say it, 14 pairs, the 14th ranked will be 14th, ranked and, you know, there's a difference in pairs and there's also, you know, where there's a Common Board against which you are compared. So that influences, right? That you're playing strategy changes based on the game.

And the finite and infinite actually works very well and it is Aspect in fund management where if you don't know matter, I mean imagine that I have two returns 1 is minus 100 percent under in second is a plus 100 percent. And in contrast, let's say I have a plus 10% under - 10 % are thematically, both of them healed the same. But geometrically it is 0 because it doesn't really matter. You you're dead in one case.

So a lot of fund management goes into the first Android hedge funds in some sense originally were Like you got to have that survival thing first before, you know, if you are alive you can eat another day, right? You don't need a feast every day. First thing rule of any game is to survive. Now you have finite games and you have infinite games and Let's ignore agency conflict and all those things. And suddenly you will find that fund management and all those things which are multiplicative

in nature. You would find that these are you know, where you Defensive because you want to survive and you don't just play to win. But now contrast that with a agency conflict comes in where in a hedge fund or in any Bank, where the guy is downside, is limited to the salary that he has been paid, right earlier. And even that was not the case, they would not, they would know clawbacks, but currently there are clawbacks, right? So you the same.

So what they would do is they would take an enormous amount of risk because inherently what they would convert is, they would convert this into An infinite set of finite games because you have a job. You would go, you would play you play to win. If it doesn't, you get fired. You go find another job. So you are just sequentially, buying options, which is good for you. Right? And that's how you make money.

You are effectively gaming, the system by converting what appears to be. If, you know, the finite not necessarily an infinite game, you're converting it into a bunch of finite infinite game or rather in finite, finite games. And you know, Grabbing value, that is how most of the money was made in finance from. Let's say it 1990s all the way till 2010 Stephen probably, now little even 2020s.

So, one of the other things I was thinking about, I keep thinking about risk and logical fallacies, right? Sometimes I think, I think one of the things that we, as humans sort of don't do very well, is to, in, a lot of cases is to assess the risk. Some of the times we sort of, sort of, like, we use small samples. They will, once I went once I invested in the stock market, And I lost ten percent of my money. So I'm never investing in the stock market. Again, you look at small samples.

You look at selection bias, you look at all these logical fallacy. So I will singing it will call context of our recent discussion on finite and infinite Games movies to a pretty bad job of teaching us risk because most movies by definition are finite gains the whatever. The plot of the movie is it it's like maybe the hero is in trouble, the plot of the movie is for him to get out of trouble, by the end of the movie. There's a finite game for this it finish.

Which You to get out of which means that you take all sorts of bliss. Can you do all sorts of sort of heroic things and stuff and you get out? But what they don't tell you is the, what happens next II mean, unless there's a sequel of course by itself. In that sense. We don't do a good job, CT. Uracil it movies. Absolutely not.

I mean movies will teach you that romance is probably one of the most beautiful things in the world, but what they don't tell you is, I mean other than a few where there is a certain amount of friction, which always exists, you know, no matter what between any two, people friction have incomes but in Because you won't even leave that, that happens, right? Or they would make you believe that is one hero who will come.

And, you know, we like to imagine, see there are originally or anywhere stories movies. They're all meant to be exaggerations. I hate, but what we mistake is often exaggerations create a certain amount frequency of that makes it kind of normalize this things, right? It is your most common thing other waiting is right. Imagine. The very widely used code man. Bites dog.

Was his dog bites man. Right now, if imagine if I sample, if somebody is samples all the newspapers historically, they would find that most of the news. Would always be there will be bad news, right? Or will be man biting dog and somebody will not get a real distribution of how often dogs bite, man. And the simple reason is that they are not meant to give you information. They are meant to trigger some reactions in you. Right here meant to trigger your

interest. We confuse them for being providing information and being representative of a real world, which they are not. Yeah, I'd actually this get Is it social media? Because at least the newspaper, which is why I sort of off late rely on broadsheet newspapers to get menus rather than social media because they at least they give you a distribution. They talk about everything that's happening in the world. They might be on average - but

overall, it's hedged. But if you go to social media, they'll be one topic of the day. And like you think that's the most important thing in the world today and tomorrow will be something else. Exactly? Right, so that it completely distorts your perception, right? Which is all the more important that Not get. It is very important. And, you know, the biggest challenge in risk management is actually not managing risk. It is in knowing that you have risk.

That is the fundamental and most basic thing, which most people. I mean, even in people who are supposedly, professional Risk Managers, they would always tend to get short sided be can, you know? Because he thinks like it is the second third multi order effects, which come back like the Algos when social media was invented. It was Invented for something. Then it became something.

It is aimed to gather clicks. It is aim to gather certain and we will see the impact of on the kids generation and all that the attention span DK, which you will find, right? It's all will come to us in some years, down the line. Now, there are some good things about it. Obviously without that, they will not be, but they're all again. When we speak of risk in real world. We normally talk about the downsides, not about the upsides, right?

So that is definitely right. So these are all factors. This is the fallacy of you observing and thinking that you are a representative. Right? I mean I once tweeted about this angle, right? I would be a bad Equity analyst because I am in one of the standard ways is, you know, you want to find what people will consume, what people will this thing etcetera.

And I mean, you know me and you know, you be so you know that as a people we generally don't like there is very little that you would want and you would fallacy is that you think most of the people in the world are Like you. And I imagine animated when I started, Facebook was listed and I saw it fall, I said it is true because I would never spend my time on Facebook. I still don't spend my time on Facebook, but that's a good stock to buy because a lot of

people who do that. So using yourself as a representative sample of the world, using your, you know, that your experience is somehow. He's The Ensemble experience and

I think we should really talk about ergodicity after when we use the word Ensemble, is these are all the standard things that we In risks and these are the things we do mistakes and then therefore the ability to manage risk completely, you know, goes Haywire. Right. Now our God is it e is actually a principle. Which is it's a it's a complex thing, right?

It is again, it is you're talking amount about F of expectation of X versus expectation of f of x. Now, just add one more element called, time into it, it and understanding ergodicity, really? If you wince, you understand, it. It is a brilliant concept. And tells you how, why, and how can risk be managed and risk? Be transformed and so on. I mean the most obvious example I will take is as follows.

Imagine that there is a sum game happening in a carnival where they are playing Russian roulette' and you know, they're giving to do, they'll take one dollars for a bit and two dollars. If you come back alive, right? Then you get two dollars if you come back alive. Now if I had a clone army, like I'm the Star Wars guy and I had a clone me. All I would do is I would keep keep like every instance round. Let's call it. Six is equal to one round and every instance.

I would send, six people go each of them. I'll give one dollar. Five of them will come back and I will get $10. So I'm making a four dollar profit and I would repeat this ad infinitum, right? So if you ask me, what is my expected profit in one round, it is four dollars, right? My long-term expected value is infinity because I have a clone army and I can keep doing this forever right now. Is the equation? I'm not in Star Wars world and it's just me.

Now. You ask me if I will go and do it. The answer is no by your long-term expected value is 0 after six shots. You were even if your luck favours and but 60 short, you go to die. I hate. So once you die, it doesn't matter what your money is, your it's zero. So the point is the time average for a single person is different from The Ensemble average, right, which is a cross sectional average. Courage and it is this difference, which is what you

would call. Like. My experience on time is very different from a cross-sections experience and the utility functions on all these things, right? And coming back to the original point. We have only one life if we have one life. The value of that life is immense for us for the insurance company. You and I are same as any Tom Dick and Harry in any part of the world. So they have some correlations

between based on some geography. Is and Healthy Lifestyles and so on. But otherwise what you are doing by buying insurance is you are converting your time average risk and translating it into an ensemble average risk for somebody else. Well, it's over. Now. I think one of the ways in which I think, people try to manage this cutting, right is to kind

of start having. I mean, typically when you want to manage something, they say that you cannot manage something that you cannot measure it. So you start measuring it and

then you come up with certain Matrix, so too. For the lack of a better term. And with all puns intended. Let's call it risk Matrix. Okay, which is I think I think it was initially started at JP Morgan in the mid-90s and then it was spun off as a separate company, which introduced this call concept of VAR. There is value at risk, which is that Why, what is the maximum amount of money that I can lose in the left five percent of the time or something? It is something like that, right?

So can you talk about risk Matrix in a generic term and the concept of VAR and what happens when you have a matrix like this to measure risk? Now, this is actually one of my favorite topics in the sense that when people and risk and they start talking about it. And, you know, in the fund, we get these, we need to present to our investors and clients. And, you know, I take pains to say that these metrics while they are visible is actually not our key things.

The key metrics in our fund is various other things. Like, you know, we have to manage liquidity.

We have to manage regulatory, which not necessarily do not have a quantifiable things, but there is a great Fascination about measuring things, which is good when there is a logic Merit behind it, but what typically happens is, once you measure something, you also get comfort with something and once you get comfort you then think that the measure is be all and end all you know, it's one of my favorite laws, which is that good Hearts law, right, which is once use what gets measured

starts getting managed and it also starts influencing your behavior value at risk. The way it is, measured is very simple. It is like you take your Leo either you run Monte Carlo, simulations, or you run historical simulations or you use implied volatility, is there. Many ways to slice the cat bread, right? So but what they do is they say that okay, 95% of our what is the downside risk that happens

once in 20 days? And you want to have that as a measure because that in some sense gives you okay. If that measure is increasing it means you are taking greater risk and that is correct. And it's a good metric in that fashion. Now what typically happens is you think that 95 measure then you think that that is somewhere, it gets normalized, then people. You don't realize that actually that means that it is probably the Minima in some sense, right? It also assumes that there is a

normal distribution, right? So sometimes what would happen is your 95% of our gets hit in three consecutive days, and then your people is like it is supposed to happen one. In 20 days, I mean, why exactly once a month, you know, why are we having three kinds of things? But that's not the the so that is where from you're having a metric, not understanding the metric, and the limitations of it comes into place. Right? And so, that is a single biggest

challenge. I mean, whether it falls in the fallacy of under, you know, or whether it is in understanding, lack of understanding, whichever way it is. So you have these metrics and you don't and then you realize a post an event that this war by itself. Ridiculous or not useful, right? And it's actually not, the problem of VAR bar is doing what it is supposed to it. Is you who didn't understand? What were its limitations, Right? Is coming back to the quants.

You need to know as a model, when a model. So what's been somebody use me. I mean, we do have internal points and we do model building. And one of the first questions I ask is not what are the returns and what are these things I asked when will you fail? Right. So remember our favorite XKCD comic, right, which is you stir and of data and then you get the results that you want. So there are two kinds of modeling, right?

One is you stir the data and then you get a result second is you have an hypothesis and then you build a model and you get a result. Right? The first one is a random via things but Algos use that so typically you want to have risk Matrix which compensate for it, which is those Horse, which are data-driven. You want them to be high sharp, right? So high frequency quantile goes will have. Hi Char because High sharp is essentially saying that I have low volatility and higher returns.

So I'm getting better risk, adjusted return, but it is not really better risk. Adjusted return it is like because the way risk is measured is volatility and volatility is measured in short term duration, which by itself will not capture tails. You know, that standard deviation is good only for a normal distribution, right? I mean, you have Distribution, which don't even have standard deviation exactly, right? So, to compensate for that, in

some sense. It has become like a proxy black-scholes where it becomes a language, you sharp becomes a language rather than a metric by itself. So if there is a high frequency Quant, where I don't know when it doesn't work on, it is being driven by data. I want a high shop. I want to maybe three or a six. Depends on different methods and logic but you want that? And the reason is effectively, you know, that there is a hidden risk that you are. Measuring and it will break

down. It will cost you money. And you want a compensation for that in contrast. Let's imagine that there is a model driven, like a bridge water or, you know, you know other people who build it. There. There is a certain element of robustness, which comes in two models will feign inherently, you know, that these are the things which don't work. It already has high wall in some sense. High wall is a measure of robustness if you see the

difference over there. Right, so it's ironical, but in some sense, it is a measure of robustness. You would think that because it is high wall. You would have seen all possible scenarios and because it is model driven or whatever it is. And therefore, you will have a lower sharp threshold for those things, right? We all invest in equities, and equities have a sharp ranging

from point 5 2 .8, right. Whereas, if I were to invest in a hedge fund, I probably would demand a much higher Sharpe because there is an unknown risk element over there. Right inequities. It's visible. It's this thing and so on you're talking about unknown risks. Now. We started talking about the 2008 financial crisis because you were sort of involved in it, in some sense like you're on the floor there, but it's putting those two together soon after the crisis.

I remember one, one of the popular magazines either, Vanity. Fair or wired or someone, they published an article called the formula that brought down Wall Street, and this was something called the gaussian copula or I think this was, I this brings me to this concept of correlation, which I think we have not spoken about.

So, because I think what happened then was that, like I think the gaussian copula had been used to sort of, like, in some way, quantify correlations between different assets and then the regime change the correlations change like that,

led to the failure of a lot of models and things like that. So can we talk a little bit about correlation? I mean, This is a tricky topic. I mean it is also not necessarily so I don't know how deep in depth. We want to go here, but see effectively everything is about modeling. In fact, especially in financial things. There is a behavior that is a pattern that you want to model in subprime crisis.

You had your packaging Securities together, you're saying that the Securities Behavior will behave in certain fashion. So already there is a this thing collateralization and giving cash flows. And then you knew have another package where you were putting on top of it. So what you are doing is copulas as you know, I mean again for this thing is our functions in a correlation functions between two variables, right? But now they are in some sense normalized.

Gaussian copula has our simple functions. What they don't have is they don't take into account. They are in some sense. There's a randomized correlation function, right? With the normal distribution and blah, blah, blah, but what typically happens is again. This is where understanding the Right, when initially, when you spoke about how do I understand? Tailed distributions tailed

distributions. Are best understood in understanding from figuring out our, their Auto correlative effects, are their Network defects, right? Whether it is from the VC world or an fa, you know, my hair Financial world or whatever. It is, need to understand that correlation has ask you, like, how volatility has asked you or volatility is, you know, considered heteroscedastic. And you know, it is auto correlative nature, low wall brings low wall and high, volatility brings highwomen.

Volatility right? There is a correlation effect in similarly correlation or the correlation. Also has asked you when so-called risk assets. They all tend to go down together. Is it logical? Probably not, but does it happen? The answer is yes, right. So imagine to give you an idea. I mean, we priced options. We I and Lehman and Barclays. I used to trade this multi currency options, right? If a multi currency option. Hide strikes, which are what are called at the money, which is

current spot level. The correlations are different, right? They are X like dollar. If imagine that I'm doing dollar dollar against rupee dollar against China as an simpler example. Right now in current benign environment world, you would think that China and India slightly. Yes, they are Emerging Markets. Therefore they are correlated,

but they are not so correlated. That is it slightly, you know, it's like a dog and Drunkard with a positive 20 percent correlation write that kind of a behavior is what you would expect and you would price. But if somebody says that, I want to price an option between dollar China at 7 and dollar rupee at 80 together, right? If I use today is correlation.

I am massively underpricing it because if China goes to seven dollar China goes to 7, it is very likely that dollar rupee will be at 80 so conditionality, right? I mean we never spoke. About this Bayesian thing. So in all these things are, there is a conditional, probabilities conditional, correlations. All these things, come into place and we all model Things based on it. It's like, you know, I have a hammer. I need to find a tool.

Or I have only this skill set and therefore, reduce me. I need to find a nail, right? Sorry. I need to have a hammer and a, to find a nail. So that is your challenge and you don't price that correlation skew because you didn't have tools about it or you don't even think about it. Right? And it is that what causes models to collapse? It? Is that which causes a risk you, which is what even in real life. It happens, right?

And they say it in Hindi, right, who privilege of data has suffered Market data, which is essentially when it rains, it pours, you know, in the other direction, when you are in crisis, you would find that con you always get more and more things which acts 2008 crisis, right? It happens. It happens not because the do didn't know of the distribution or anything. It happens because the pain the aspect all those things are all

linked together, right? It is a natural phenomenon and it is this which causes always the pain to Wall Street and you know, even in real life even for us, it causes entities. This avoidance of this worst case where you know, which causes a makes us all to be defensive which is what in some sense. We are all the market Falls 20%. We see a 40%, it actually might make sense to invest a 20 percent correction, but we are always I worried about forty

percent, right? So our Primal brain, you know, the fight-or-flight reactions they all trigger, which is why you would want your money to be managed. Ideally. I mean, by in a systematic or methodical fashion where there are behavioral anomalies that you can overcome right and mean you know as well maybe spoke about it many times, but that's the, I mean, that's the broader principle about not being able to see how correlation itself

has a smile. This problems and brings things, may brings down things and I think as you rightly mentioned, it has a we forget financial markets. I think even in real life a lot of things are like lot of our risks have to do with correlations and with correlations with change and you kind of like the and when they change you don't know how to handle it and then you sort of get exactly and so on. And we use current expectations to risk the correlations. Right?

Dre. You're having good times today. You're having your job is not related to your stock market. Eight. You think both of them are independent, but what we do not know inherently is the probability of you getting fired increases as your stock market,

portfolio goes down. So there is a positive correlation or negative correlation which way you want to think is there and you kind of have to manage it which is why when if you had a good financial advisor, in fact, you know, one of the things is if you have a financial advisor, ideally you should hate him or her right? Because it's like a his that person is supposed to be like a doctor making you to go on a diet making you to do things

which you don't. Like like put force you to take insurance, which will never see any value. It said money down the drain force, you to take medical insurance or force you to keep like one or, you know, two years of capital just so that you survived right now, you will see that capital. And you would think that, oh, if I put that money last year in stock market, this year, it would have doubled, right? So you should hate that

financial advisor. But ideally, your financial advisor is supposed to be protecting you from these correlation Tales of correlation. Skew, if you really have a good one obviously agency. Conflicts come into picture. That's a different thing. But you have a truly good one. That's what that person is supposed to do. Possibly. This has been a fascinating conversation.

I think we've been talking for a long time now, and now, as a closing question are like, I mean like let's say you're a young person who is for whatever reason interested in the risk. Later. Your you've mastered black jacket 6 poker at 9, GJ 12 and so on liquor and you think you're a, you're very good at risk understanding risk, its own, like now after the global financial crisis, I guess the overall Financial Tree, which is what - mostly hosted this

coefficient. Is that sort of like diminishing now, so if you are, if you want to make a career in Risk, what you do, so the broadly two ways in looking at risk one, is you can be a risk, professional yourself, right? We're all organizations. Legally compliant spies everywhere. There are mandated. They risk teams need to be independent, and there is a tremendous value in having a good risk person in any

organization, right? And, and I am not Just about finance that is now, we mean, if you hear any of these podcasts about Supply chains recently, you would find that somebody the job of risk is like to get gpus to operate your sass farm, right? That itself is a risk and you could not procure, because there was no supply chain. There was in availability in the markets. So, risk management is manifold Right. Medical insurance, medical area. Is another thing.

So there are lots of areas. We speak about Enterprise risk, Financial Risk Logistics. And so many areas where you can do, right? And in Financial Risk, obviously specifically, you have requirements like frm and, and so on, but in and separately, obviously, as a Trader, you know, to fader is, being risk manager in and out, right? You don't live by external world. You inherently have to be a risk, manager yourself.

In fact, I mean, it might be not widely, might want your popular opinion, but actually think very good. Predators usually, end up being very good Risk Managers. And you can see the top hedge funds, the risk. They have are usually extruders who kind of done quite well, whether it's Millennium or, you know, anything etcetera. So career-wise, I think risk is one area where there is greater

awareness. There is in fact, the word risk now has much wider and it so to speak and even if you were to be a start-up, in fact, what should happen is? It's like computers right now. You no longer mean, computer science. Engineering is kind of a programming skills is now, Now necessary for any job right. Similarly risk. And in fact, I will go one step back in some sense understanding data. Eight, which is a precursor for understanding risks, right?

He's a necessity in any job anything, right? It's a skill that it is not like Risk jobs will value more. In fact, if you were CEO of a company and you understood your wrist, when you can add far more value than somebody who doesn't thank you for listening to data shatter. If you like this show, please leave a comment, share and subscribe to the podcast. You can find this podcast on Apple podcasts Spotify or wherever else you go to get your podcasts. Once again, this is Karthik

signing off. Thank you.