Master Talk: Rise of Quant Funds — Alt-Data, Sentiment & Predictive Analysis
Join us for the conversation on “Rise of Quant Funds — Alt-Data, Sentiment & Predictive Analysis” with Mr. Anil Ghelani Head of Passive Investments & Products, DSP Investment Managers
Follow this story and more: https://www.outlookindia.com/
#Money #OutlookMoney #OutlookMagazine #OutlookGroup
Follow this story and more: https://www.outlookindia.com/
#Money #OutlookMoney #OutlookMagazine #OutlookGroup
Category
🗞
NewsTranscript
00:00 (upbeat music)
00:02 - Good evening investors,
00:24 and hope everyone's keeping safe and healthy.
00:27 Welcome back to our third episode
00:29 of the MasterTalk series,
00:30 where we learn about winning strategies
00:32 employed by fund managers.
00:34 So today's theme is going to be about rise of quant funds,
00:38 alternative data, sentiment and predictive analysis.
00:42 While the debate between value investing
00:44 and growth style investing rages on,
00:47 quant investing is slowly but surely
00:49 grabbing a piece of the investor pie.
00:51 For the average investor,
00:52 the best way to access quant strategies
00:55 is by investing in quant funds.
00:57 These are basically mutual funds
00:59 that use advanced mathematical modeling,
01:01 new age technologies like artificial intelligence
01:04 and data analytics,
01:05 and algorithmic trading
01:07 to identify and execute profitable trades.
01:10 A key point to notice here is that
01:12 since human intervention is minimal
01:14 in the decision making process,
01:16 these funds can optimally reduce errors
01:19 that stem from human emotional and behavioral biases.
01:23 In today's episode,
01:24 we will be discussing all things related to quant funds,
01:27 while specifically focusing on the winning strategies
01:30 employed by fund managers.
01:31 Sounds interesting, doesn't it?
01:33 Well, without much ado,
01:35 let's start this episode of the MasterTalk series
01:37 on the rise of quant funds,
01:39 powered by Refinitiv,
01:41 leading providers of financial technology,
01:43 data and expertise to keep you ahead,
01:46 and in association with your favorite
01:48 personal finance magazine, Outlook Money.
01:51 I'm Deepika Asthana, your host for the session.
01:54 Today, we have with us Mr. Anil Gilani,
01:57 Head of Passive Investments and Products,
01:59 DSP Investment Managers.
02:01 Anil has been working with the DSP group since 2003.
02:04 Previously, he served as the business head
02:07 and chief investment officer at DSP Pension Fund Managers.
02:11 Earlier in his career at DSP,
02:13 he led the risk and quantitative analysis team,
02:16 providing quantitative research inputs
02:18 and buy side credit research on companies
02:21 across various sectors.
02:23 He has been a volunteer with the CFA Society India
02:26 for more than a decade,
02:27 and currently serves as the vice chairman on the board.
02:30 Good evening and welcome Anil.
02:32 - Hi Deepika, good evening Vishu,
02:35 very warm welcome friends,
02:36 everybody who has joined for the session today.
02:39 - So let's start the session Anil, with an anomaly.
02:43 So over the last one year,
02:44 we have seen lackluster performance by quant funds globally.
02:48 However, in the backdrop of this performance,
02:51 Indian funds have outperformed benchmarks.
02:53 So let's hear your view on the reasons
02:55 for this supposed anomaly.
02:58 - Sure, see Deepika,
03:01 structurally when we talk about quant funds,
03:04 it's a slightly generic term,
03:06 and it can be used for several different types of strategies
03:10 which deploy quantitative style,
03:13 which could include certain strategies
03:14 which are momentum driven,
03:16 certain strategies which use some macro models,
03:20 some very different,
03:21 which use high frequency trading strategies.
03:23 This I'm talking about global markets,
03:26 like since you raised a point about global.
03:28 And structurally, all strategies and different managers
03:32 adopting these strategies,
03:34 always go through certain cycles or periods
03:36 of out and under performance.
03:38 So yes, maybe in certain segments
03:41 of this type of quant investing,
03:43 in the last one and a half years, two years,
03:45 some of these funds have underperformed,
03:47 whereas on the other side,
03:48 some funds have outperformed as well.
03:51 Like if we see some of the, in India,
03:54 structurally since you mentioned the comparable
03:55 between global markets and India,
03:58 in India, if we see today,
03:59 there are hardly five quant based mutual funds
04:03 in India currently.
04:04 Some of them have been launched even less than that,
04:06 like one year ago.
04:08 So structurally, if you see that space is very small,
04:12 as I said, within that,
04:13 depending on the strategy of that quant model,
04:17 let's say if somebody is following a momentum strategy,
04:19 somebody is using approach where they are focusing on,
04:22 let's say quality or value,
04:25 then depending on the underlying cycle,
04:28 some of these strategies will have underperformed
04:32 in the last year,
04:33 whereas some strategies have significantly
04:35 given good outperformance.
04:36 - Fair enough.
04:39 I think that is from that perspective,
04:40 true for all funds, right?
04:42 I mean, we should not look at it in silos
04:44 in terms of just one time period.
04:46 - Correct, one time and more so,
04:47 because when we say structurally at the kickstart
04:50 of this session itself,
04:51 that when we talk of quant or quant investing
04:54 or quant funds,
04:56 within that also,
04:58 there can be multiple very distinctly different strategies,
05:03 like a quant fund applying momentum strategy
05:05 will have a completely different risk and return trajectory
05:08 versus a quant fund applying value,
05:11 or for example, a combination of growth
05:14 and minimum volatility.
05:15 So that's very important to get clear
05:20 in the minds of our viewers at the upfront.
05:23 - Absolutely.
05:24 So let's, I think one thing let's try to focus on
05:27 is momentum strategies,
05:29 because trend following, identifying a trend
05:33 and then following it is,
05:34 I think everybody's a hobby in the market, right?
05:37 And momentum kind of gives you that kind of space
05:40 when you identify the trend and then you trade with it.
05:44 But from that perspective,
05:45 how do you think momentum strategies
05:47 are currently doing in the market
05:49 and going forward when maybe markets are not,
05:52 markets are never unidirectional.
05:53 So from that perspective,
05:54 how do you think they're likely to perform
05:56 in the future as well?
05:57 - In my opinion, momentum strategies have worked very well,
06:01 especially in Indian markets,
06:02 as well as typically speaking,
06:04 what we have seen in many developing markets.
06:07 So if you have a distinction between emerging
06:09 or developing markets and developed markets,
06:11 more often than not,
06:13 you're trying to see momentum strategies do pretty well
06:15 in developing countries or emerging markets.
06:19 In India, recent times,
06:21 in fact, we have seen a slight turnaround also,
06:25 not only pure momentum,
06:26 or it's a combination of certain other factors
06:28 which are at work.
06:30 Let's say about 18 to 22 months back,
06:33 if you would have seen momentum driven investing,
06:36 you would have been doing very well.
06:38 In fact, trend following,
06:39 like you mentioned, very classic word.
06:40 I mean, trend following versus let's say another,
06:43 belief is a mean, people who believe in mean reversal.
06:46 So what we are seeing in Indian markets,
06:47 because there was a huge polarization that was happening,
06:50 whereas in the entire market space,
06:53 not just large cap, but within large cap,
06:55 only a small set of quality stocks,
06:57 investors were just using quality.
06:59 So, so-called like, you know,
07:00 that nomenclature, Hrithik stocks or something like that,
07:03 names were only what were driving.
07:05 And if you were investing
07:07 with that momentum driven strategy,
07:10 but actually chasing the underlying quality
07:12 or the growth of that segment of the market,
07:16 then definitely your strategy would be doing well
07:19 for that time period.
07:21 But as I said, that structurally, that particular cycle,
07:25 in fact, as we see the market becoming more broad-based
07:28 and not remaining polarized,
07:30 in that, like now last about six, eight months,
07:32 we have seen much more broader based rally,
07:35 as compared to what was there one year ago,
07:37 one and a half years ago.
07:38 So in that broad-based rally,
07:40 it could very well be that not always then
07:42 momentum as a strategy might be working,
07:45 but yes, again, it will have its own, you know,
07:49 comeback, you know, where it will do very well.
07:52 And just on the lighter side,
07:54 like you mentioned a good example
07:55 that very often people think that yes,
07:57 momentum investing is one of the best strategies,
07:59 completely agree, but equally good.
08:02 And there are equal number of followers,
08:06 you know, as a, let's say, algo trader,
08:07 or as a bond investor,
08:08 when you look at these two fundamental differences,
08:12 like, you know, when we even talk to our friends
08:16 in this circle, you know,
08:17 there is this, you know, different sometimes language,
08:20 and when you are nowadays not in a room,
08:23 but otherwise when you are in a room,
08:24 typically you get,
08:25 (indistinct)
08:27 these are all your trend followers,
08:28 and these are all the guys of the mean revision type stuff.
08:31 So equally enough is there for mean revision type of approach
08:36 because we all know, and we all understand that,
08:39 like there are cycles in the broader capital markets.
08:43 Similarly, even in individual pockets or individual factors,
08:47 there has to be a cycle.
08:49 And when that cycle is turning,
08:50 the mean revision will work.
08:52 Often, yeah, will work, exactly, exactly.
08:55 So like, even if you look at, let's say,
08:58 just the taking the same example,
09:00 like Ritik's stocks were doing very well.
09:03 Now, I'm not saying that now in this last six,
09:05 eight months, 10 months,
09:06 the Ritik's stocks have fallen considerably,
09:09 but I'm saying that the others,
09:11 even if you look at the large-cap space,
09:13 let's say, Nifty 50 stocks,
09:15 and earlier only those eight, 10 stocks were main drivers.
09:18 Now, even the other stocks within large-cap,
09:21 within Nifty 50, the other 30, 35 stocks,
09:23 they are doing more, better,
09:25 and playing catch up with the top 10.
09:28 - Right, right.
09:29 - So that type of a mean revision
09:31 and that type of a connect between how the change happens,
09:35 because very often when coming to the specific discussion
09:40 about quantitative investing,
09:41 and let's say one particular factor, like a momentum,
09:44 in certain situations, when that becomes very popular,
09:47 it results into positions that become over-owned
09:51 by investors who are using that same factor.
09:53 - Correct, correct.
09:54 - So it's very crucial and very important to understand
09:59 that you need to even have a second layer
10:02 of derivative analysis,
10:03 whereby you are monitoring the valuation of the factors also.
10:07 Okay, now momentum-driven strategies
10:09 or momentum-driven stocks,
10:11 what is the valuation metrics
10:12 of those underlying companies now?
10:14 Is it something going beyond certain threshold levels?
10:16 Then maybe it's a signal now
10:18 that maybe that's going to reverse.
10:20 - Absolutely.
10:22 I mean, I think your point with respect to what you said,
10:25 if everybody's following the same strategy,
10:27 it kind of becomes self-serving, right?
10:29 Because as more people buy it, it goes higher,
10:32 and then, I mean, then you're right all of the time.
10:35 - One is self-serving,
10:37 other is what I mentioned, slightly different,
10:39 is it becomes over-owned by investors
10:41 using that same factor.
10:42 So that could lead to a risk also that,
10:46 now if everybody's chasing that,
10:47 is it resulting into excessive valuation
10:50 of that particular factor or those one or two factors?
10:53 As I said, second layer of derivative analysis is required
10:57 when you're doing even momentum
11:00 or any factor-based investing.
11:02 Not just seeing that factor,
11:03 but looking at a second layer derivative
11:05 of the valuation of that factor also
11:07 becomes very crucial on a constant basis
11:09 we need to keep looking at that.
11:11 - Right.
11:12 So Anil, actually, from what you're saying,
11:14 would it be fair to conclude that quant models
11:18 are able to optimally capture that space
11:21 between price and value?
11:23 Since you spoke about the second derivative level
11:26 of value capturing as well.
11:29 - Sure, sure.
11:30 So when we refer to value in a factor language,
11:35 it has slightly different context,
11:38 but yeah, otherwise you are right.
11:39 In that sense, if your model is designed
11:41 to work between these two in a suitable manner,
11:46 where you are assigning relevant weightage
11:49 to not just momentum, but even value,
11:51 then yes, it will be in a position
11:52 to capture that suitably.
11:54 But if you are working with a single factor model,
11:57 like I remember back in the day
11:59 when I was a young quant analyst in 2007,
12:02 eight, we used to run a paper trading portfolio
12:04 because in those days still it was not easy
12:06 to implement this in the real market.
12:08 So we were planning to do a paper trading first.
12:11 So we're using only one single factor momentum.
12:14 We used to select 20 stocks to go long,
12:15 20 stocks to go short.
12:17 But it was only one factor, momentum.
12:21 Now what happened, exactly what I'm saying,
12:22 at a point of inflection in the market,
12:25 when suddenly momentum as a factor changed.
12:30 So it took very long time for the models
12:32 to start recognizing that.
12:34 And in that interim period, two, three months period,
12:36 that particular paper traded portfolio
12:39 significantly underperformed.
12:40 Because we were overweight only the stocks
12:43 which were high on momentum.
12:45 And suddenly the entire cycle changed.
12:48 So hence my point was that you need to have a watch
12:52 on the other element, which is the valuation of any factor,
12:56 whether momentum or let's say when we are using value
12:58 as a factor.
12:59 (Sharlina laughs)
13:00 You should watch how the valuation
13:01 of those stocks are changing.
13:04 - Understood, understood.
13:06 Okay, so great.
13:07 Then let's talk a little bit about the models
13:09 since we've brought this into the conversation.
13:11 So I think one of the basic things about a quant fund
13:15 is the backtesting of data, right?
13:17 So here, there are two things I think we can
13:20 maybe go deeper into.
13:21 One is of course, we use historical data, right?
13:25 So in these models, how do we account for unpredictable
13:30 or the black swan events, so to speak,
13:32 because the models only take into consideration
13:34 events that have happened in the past, right?
13:37 And my second question, we will talk more about maybe data
13:41 in terms of data is the oil of the 21st century,
13:44 more tech powered, everything.
13:46 So what are the factors other than traditional metrics
13:49 that we now need to look at in terms of alternative data?
13:52 - Very good point.
13:55 So first aspect, like how you mentioned,
13:57 how does a quant model factor in this kind of
14:01 a black swan type events?
14:03 Like black swan event was made very popular after the GFC
14:07 and very beautifully captured in the book by Mr. Taleb.
14:12 Yes, you are right that structurally,
14:16 history hardly repeats itself in totality.
14:19 Structurally black swan events are such that they have
14:23 very, very rare occurrences, but when they occur,
14:27 the impact is catastrophic basically,
14:30 but being that they are very rare to occur,
14:32 whether it's a quant manager or active fund manager,
14:35 individually tracking stocks or sectors of teams,
14:38 it's extremely difficult or crazily expensive
14:43 to prepare for a black swan event
14:44 and hedge your portfolio for that.
14:46 So what happens structurally is that if today,
14:50 sitting in the month of let's say November
14:52 or December of 2019, some data points would have shown
14:57 that in March there's going to be a sharp correction,
15:01 would it have been possible that somebody will hedge
15:03 that portfolio very difficult or very expensive
15:06 because sitting in that period, if you try to buy out,
15:09 let's say any hedge in the portfolio
15:11 or any long dated put options or something like that,
15:13 it's going to be very expensive.
15:14 And structurally no model can be structured
15:20 or created to factor in this type of very out of the way,
15:24 like a more than a six sigma type event,
15:25 like black swan or what you mentioned.
15:28 So in some form or fashion,
15:31 often for such kind of situations,
15:34 any quant strategy would tend to underperform,
15:39 but structurally there are various other areas
15:44 where a good simulation can be built in
15:47 and protection can be created.
15:49 Like might be that long term fall down in November,
15:54 one could not have predicted,
15:56 but it's one of topic very close to heart for me,
16:01 earlier in my career, I had also been a risk manager.
16:03 So thinking of risk is also very dear to mind
16:06 when you're investing.
16:07 That another similar animal,
16:10 which was made famous after the global financial crisis,
16:13 after black swan was the gray rhino.
16:15 It was done by a very dear friend,
16:18 Michel Hooker, we hosted in India last year
16:20 for a conference.
16:21 So where Michel Hooker classifies gray rhino
16:26 as an event which is open for the world at large to see,
16:29 not like a black swan, unexpected.
16:31 Today when person sees a swan,
16:33 you are expecting only black color swan.
16:35 That's how you are wired to think.
16:36 So your models are built on your mind,
16:39 even for a personal event,
16:40 anything is structured as a white color swan
16:43 only I'm going to see.
16:45 But rhino gray, it's gray only.
16:48 I know it's coming, it's going to be gray only.
16:51 So it's an event open for the world at large to see.
16:55 And only thing is that it's a human bias
16:59 which sometimes thinks that, let me just avoid that
17:02 because okay, even if I'm seeing a rhino standing
17:03 far from me, not necessarily it's going to start charging
17:06 and coming running towards me.
17:07 Let's say I'm somewhere in the north of the country
17:10 where rhinoceros is very commonly found animal.
17:13 Maybe it's going towards that side
17:14 where there's some food line for the rhino.
17:16 But if my model is well aligned,
17:20 maybe sometime in December, January,
17:22 when there was data points coming out
17:24 that in certain countries like China,
17:26 like connected to China, Hong Kong, or some Japan,
17:29 some other countries,
17:30 number of cases of the virus are increasing.
17:33 What can be the next year of impact for that?
17:36 Okay, very small chance,
17:37 but there's a chance that it can carry on
17:39 further to some other countries.
17:41 Exact impact of the magnitude that we saw
17:44 might be very difficult to predict,
17:45 but in small, small ways,
17:48 these kind of instances or these kind of occurrences
17:53 could lead to some more relevant preparation
17:55 in the portfolio.
17:56 So maybe during the month of January,
17:58 maybe one could have taken some near term kind of protection
18:01 not anticipating that in March,
18:03 we are going to see such a sharp kind of a reaction
18:05 in the capital markets and in the physical real economy,
18:08 but some wiring, some changes
18:10 definitely could have been incorporated
18:13 if you have got your structuring correct.
18:15 So coming to your original thing
18:18 and to conclude this angle,
18:20 I feel very strongly whether for a bond model
18:24 or for you and I taking any decision in personal capacity
18:26 or active fund manager doing a stock by stock research,
18:29 factoring in a black swan event
18:33 and preparing for it or hedging for it
18:35 is extremely challenging and usually insanely expensive.
18:39 But instead of that,
18:40 looking at smaller pieces of information,
18:43 which could be gray rhino type instances
18:44 of gray rhino events in some way can be more relevant
18:48 and more practically implementable.
18:51 Yes, black swan events also are important
18:52 to try and identify and capture,
18:54 but structurally, especially in this kind of a scenario,
18:58 very challenging.
19:00 - Understood.
19:01 Agreed, actually.
19:02 I think what I, my biggest takeaway from this
19:06 is that it's important to recognize the risks
19:09 and not always possible to judge
19:10 the magnitude of the risk, right?
19:13 So like you said, in January,
19:15 let's recognize the risk and take some proactive measures
19:18 to mitigate the impact,
19:19 not really fully understanding the magnitude of the impact.
19:23 - Yes, yes.
19:23 And as I said, the difference being in November,
19:26 there was not so much information yet available.
19:28 Hence, I was saying that at that time,
19:30 it can still be thought as a black swan.
19:32 It's unexpected right now.
19:34 Normal case, little bit virus in all,
19:36 it was seasonal thing.
19:38 Many times it happens in those part of the world,
19:40 unfortunately, I mean, whatever you think of it.
19:42 But by December, January, it was already coming out.
19:44 It's a very unique, different strain of virus,
19:47 very challenging to cure and curtail.
19:51 So it had become a grace at one stage
19:54 before it actually hit us.
19:56 That's the difference between,
19:57 like why I said, as an example,
19:59 October, November, and then December, January,
20:00 like the distinction between the two.
20:02 That by the time then that information is available
20:05 and open for the world at last to see
20:07 and react if you are able to react.
20:09 That is a distinction.
20:11 Those events, yes, certainly some models
20:13 can capture suitably and assist the portfolio managers
20:17 to take correct, protective action.
20:20 See this again, this last point I'm saying,
20:22 not could not be only for quant managers,
20:25 but it could even be for quantamental,
20:27 like active managers using quant strategies
20:29 or active manager using risk management
20:31 incorporates in corporate risk management
20:33 in their investment framework,
20:34 quantitative analysis in their framework in that way.
20:37 - Certainly, because risk is everywhere, right?
20:39 So we all need to deal with it.
20:42 Great, so you spoke about information, right?
20:45 I mean, I think that is the basis,
20:48 the foundation for everything all fund managers do,
20:52 whether it's quant or passive,
20:53 any fund manager, information is important.
20:56 So from that perspective,
20:58 what kind of data is sacrosanct to quant models?
21:03 And even more importantly, now,
21:05 we're living in a world where social media
21:07 and sentiment analysis is becoming more and more important.
21:11 So how do we incorporate this into a model?
21:14 - Very important point, Deepika.
21:16 And structurally, yes, you're right,
21:17 that one needs to have very high quality of data.
21:22 And nowadays there is a large amount of ancillary data
21:27 or what often we call in technical language,
21:30 like in investing world as alternative data,
21:32 or all data we call it, or some people call it big data.
21:36 All data is a more slightly,
21:39 commonly used word in investment industry.
21:43 So one is this type of financial data
21:46 officially released by the company
21:47 in terms of their company filings,
21:49 balance sheet, P&L, cashflow, et cetera.
21:51 And based on your expertise as a portfolio manager
21:55 or as a person tracking the company,
21:57 you can mix and match some ratios
22:00 which are captured from only balance sheet,
22:02 some only from P&L,
22:03 some from a combination of balance sheet and P&L
22:04 and cashflow, so as to give a better understanding
22:06 of any financial shenanigans or any red flags.
22:11 So that is your expertise,
22:13 how you ensure you develop a very risk model,
22:17 I'm sorry, very robust model
22:19 to capture the company's underlying fundamental
22:22 as well as technical analysis.
22:24 And then this other part of the data which you mentioned,
22:27 social media or now with further enhancement.
22:31 Today, just yesterday I was chatting with somebody
22:36 and some young friends in the industry.
22:39 Very often I talk to them about the CFA network
22:41 and they were saying,
22:42 (speaking in foreign language)
22:46 I'm there but not very,
22:48 (speaking in foreign language)
22:51 How trends are evolving?
22:52 I said, yeah, I will maybe try to look it up.
22:55 So what I'm realizing,
22:57 and this is going to be the future
23:00 of even the way in which investment management
23:03 will use this kind of data,
23:06 is that how emotional data
23:09 or emotional aspects are being captured.
23:14 When I mean by emotional,
23:17 what I'm saying is that,
23:18 what I spoke in the first part is the financial data.
23:21 I'm the company, I'm sitting in my headquarters in Delhi
23:24 or XYZ, these are my factories,
23:25 these are my, or whatever my outreach.
23:28 And this is my financial data
23:29 I'm generating for you to analyze.
23:31 Now, that company's users, vendors, blah, blah,
23:36 everybody is absorbing them.
23:38 And then they are putting on social media,
23:39 oh, wow, wonderful product.
23:40 Oh, no, this I didn't enjoy, blah, blah, whatever.
23:43 That is what we call as emotional data.
23:46 So the classic combination of emotional data
23:50 with financial data
23:51 will ensure best financial performance.
23:54 - Right.
23:55 - How emotional data will be captured today
23:57 is happening, but it's still very, very challenging.
24:00 Because today, if you have to just capture data
24:02 from social media handles or do some web scrapping
24:04 and try to see how many people have said good things
24:07 about let's say Maruti's new car getting launched,
24:10 for example.
24:11 - Right.
24:12 - Or let's say, for example,
24:14 if one wants to see that where are the kind of new
24:19 developments happening in electronic vehicle side.
24:23 So suppose you want to see Google trends
24:25 of how many people are now searching for electronic scooter
24:29 versus petrol fired scooter, basically.
24:32 So that can give good inputs to the portfolio managers
24:38 that how I can ensure or incorporate this type of
24:42 alternative data or in some form,
24:48 like unstructured data to form some judgments.
24:51 Now at the current situation,
24:54 at least as per my limited understanding
24:56 to incorporate this type of unstructured
24:58 and alternative data into a quantitative model
25:02 and run it is very challenging.
25:03 Because it's unstructured as we are discussing today.
25:07 It's very challenging because you need to have
25:09 specific rules being coded and established,
25:13 which can't be so easily done
25:14 in this kind of unstructured data.
25:16 But certainly it will be useful going forward
25:21 in terms of understanding better trends,
25:23 in terms of understanding investor or consumer trends
25:27 about certain products, certain service,
25:29 certain category of services,
25:31 certain shift in consumer behavior.
25:33 If people are trying to capture the early shift
25:37 or change towards electronic vehicles,
25:40 maybe these are the good trends to observe.
25:42 Not just Google searches,
25:43 how many people are searching about electronic scooters,
25:46 but social media trends,
25:47 it was how everybody are reacting.
25:49 - Reviewing it.
25:49 - People have written, "Oh, I have to buy
25:51 an electronic scooter."
25:52 And others are in their friend circle are saying,
25:54 "Oh, how was it?"
25:55 And he or she is giving good review.
25:57 "First class, boss.
25:58 No cost, easy to charge.
25:59 Go and buy, go and buy, go and take a look at the dealer."
26:03 So those are the trends which can help in picking up
26:07 more information and incorporating them in your judgment
26:10 or decision making process.
26:13 - Sure.
26:14 So then coming back to this bit about unstructured data,
26:17 because it's becoming more and more relevant today.
26:20 It's not just in investing,
26:21 but I think across sectors data,
26:23 unstructured data is becoming important.
26:26 From that perspective,
26:27 do you think quant fund managers or quant funds in general
26:30 can going forward increase the use of technologies like AI,
26:35 maybe data analytics to make more sense
26:38 of this unstructured data?
26:39 - Yeah, definitely.
26:41 So in that, the AI as well as ML,
26:44 the machine learning and artificial intelligence,
26:46 they will become more and more powerful.
26:49 In fact, in many of the global markets,
26:50 it is really being done in a very big way.
26:53 There are large teams specialized in this kind of area.
26:58 Gradually, I think definitely get incorporated
27:01 and enhanced over here also in India.
27:05 In the sense of utilizing that and making sure
27:13 your current models or your current way of investing
27:16 incorporates that, that's the key.
27:18 Like today, if you are given large set of data also
27:22 about this kind of social media trends or Google analytics,
27:26 what will you do with that?
27:27 And how you will incorporate, that is the key.
27:30 - Right, agreed, absolutely.
27:32 So then, I think let's also understand,
27:37 so we know data plays an important role.
27:38 We understand that there are various strategies
27:41 that quant funds can follow and risk is also imperative.
27:45 One more thing that then I would like
27:47 to really understand from you
27:48 because you have so much experience here
27:50 is how does one go about building robust quant models?
27:54 - Sure, see, I think in some way,
28:00 it's not like something very like eureka moment,
28:05 like Archimedes was that, like suddenly in the bathtub,
28:08 suddenly some idea came into the mind
28:10 and jumped out and ran towards his workshop
28:11 and did it like that.
28:13 So it's not like suddenly some quant manager
28:15 is sitting on his or her desk
28:18 and she thinks of some unique idea
28:20 and then runs to the computer and starts coding that model.
28:24 Structurally, when you look at that,
28:26 there are certain predefined buckets of information data.
28:31 And based on that, how your investment style
28:38 or your investment philosophy,
28:39 you want to incorporate in that model, that's the key.
28:44 So if I want to in a very simple way,
28:48 define first of all, what are quantitative parameters
28:51 or what are factors or factor investing?
28:53 When we say factors, they are nothing but the risk
28:58 and return characteristics of each individual line item
29:01 or each individual financial security.
29:03 So the risk and return can be sliced down
29:08 into multiple segments.
29:10 So, or in a very simple example,
29:12 whatever food we eat, you can slice it down,
29:15 okay, this food has given me X number of protein,
29:17 X number of carbohydrates, blah, blah.
29:19 When I'm approaching a dietician,
29:20 that's in preparation for the marathon in Mumbai next year,
29:24 she has told that, okay, X number of carbohydrates,
29:26 Y number of proteins, and Z number of whatever,
29:29 I mean, I'm not a dietician, sorry.
29:32 Then each and every food that I plant during the day,
29:35 I will not look at the item,
29:36 (speaking in foreign language)
29:39 what item I'm eating.
29:40 I'm going to look at,
29:40 now this is what I'm eating for lunch.
29:42 So I finished X quantity of carbohydrates, Y of this.
29:45 I stopped talking about,
29:48 you know, I had two chapatis and one sabji.
29:51 I'm saying, okay, I finished X protein and Y this.
29:55 So how is it happening?
29:56 There is something which is defined top down,
29:58 okay, you need to have X only protein and carbohydrates,
30:02 for example, then everything you are doing,
30:04 sliced down everything else
30:05 and only picking up that much part.
30:08 So like that, before a quant strategy
30:10 or a quant model is defined,
30:11 suppose the particular,
30:13 okay, these are the two factors you have to consider,
30:16 for example, that you have to consider quality,
30:19 growth, and value, for example, just I'm saying,
30:21 three factors you should consider.
30:23 For any simple investing, what would you say?
30:25 If you buy the company,
30:25 it has got a good quality of earnings.
30:28 And it has a good growth potential of those earnings.
30:29 Earnings quality is there,
30:31 it should have growth potential.
30:32 And that company is available at a reasonable value.
30:34 Or simply put, if I was to define a basic principle,
30:39 any, as I said in the beginning,
30:40 a quant model can be one or multiple factors, basically.
30:43 So let's say we define one model with these three factors.
30:46 Now, everything depends on that,
30:51 what, how you define the individual parameters
30:53 of those factors that for a quality,
30:57 what are the factors you want to see?
30:59 You want to see how the balance sheet stands,
31:00 how the P&L is, how the cash flow to earnings,
31:04 certain ratios are captured,
31:06 then the growth part, and then the valuation part.
31:09 So that's how it typically happens,
31:12 it's a top to bottom type of a thinking
31:14 and capturing the relevant data
31:16 and building up on your model.
31:18 Like how a successful model can be built up.
31:22 But first you define what are your core investing philosophy
31:26 or core investing rules or principles.
31:30 Then you dissect them like you do for a food particle,
31:33 food item into different, different factors,
31:36 two or more or one or three or whatever.
31:39 And then you create your model.
31:41 Towards the same question,
31:43 how to create a successful model?
31:45 I would define it like this.
31:46 First create your top down rule or investment philosophy,
31:50 then have the grip to dissect it into one or two
31:53 or three factors which are going to be defining
31:56 that particular strategy,
31:57 and then go on and build it up very easily.
32:00 Very simple.
32:01 - Thanks, I think that will be very instructive
32:05 for I think everybody who is looking to,
32:07 foray into quant investing,
32:10 because it's always, I think, I'm sure you will agree,
32:13 had some sort of an appeal, right?
32:15 That if you can break it down into factors and numbers,
32:18 it seems so clean cut from that perspective, right?
32:22 And then there are rules.
32:23 So as long as you follow the rules,
32:25 the assumption is that you will make it to the finish line.
32:28 Great.
32:31 - So on the lighter side,
32:32 I hope that our, your, this show, Deepika,
32:34 will not help only people who are interested
32:37 in quantitative investing or building a career
32:39 in quant investing, but also, as I mentioned,
32:42 free advice for Mumbai marathon running also.
32:44 - Right, absolutely.
32:47 Absolutely.
32:48 Everybody likes that.
32:49 (speaking in foreign language)
32:51 It's great, right?
32:53 - Great, so I think maybe next up,
32:55 we can just discuss your perspective or your opinion
32:58 in terms of where quant funds,
33:01 and maybe more from the perspective of India,
33:03 where quant funds are headed.
33:05 And is the future going to be more of a machine
33:07 versus humans kind of scenario,
33:09 or is it going to be machine plus humans?
33:12 - Oh, well, well, for that, I think it's very clear cut.
33:16 In my mind, I am very clear, Deepika,
33:19 whenever I'm making any decisions,
33:23 whether it's about any personal choice
33:25 or about any business decisions,
33:27 unless it's selecting some particular stock,
33:30 but otherwise, I always prefer to use
33:34 the power of and versus or.
33:36 Okay, so if you go with and,
33:41 the results are often very powerful.
33:44 So I believe strongly it will be humans and machines.
33:48 It's not going to be machine versus human, I mean.
33:51 So it's because one needs the other,
33:53 each needs the other, otherwise it will be a disaster.
33:56 - Correct.
33:56 - So it certainly has to be a combo.
33:59 I mean, of the two.
34:00 Why disaster we have seen in some global markets
34:02 that if there was only machine driven thing,
34:04 which was running some high frequency trading
34:06 and something went wrong, suddenly it corrected,
34:09 and it could result in some very awkward situation
34:12 in the capital markets.
34:13 - Right.
34:14 - And why, I mean, what is the reason for that?
34:16 Likewise, if you are doing only human,
34:19 I mean, as an individual, you are a,
34:22 let's say, best stock picker,
34:24 but you refuse to realign your old school thinking
34:29 and adopt to certain inputs from machine
34:31 or enhance your data inputs,
34:34 like collecting this type of emotional data also,
34:38 emotional performance data,
34:40 then it is going to be more and more challenging for you,
34:43 and you might be a little bit marginalized
34:44 because that's how things are changing.
34:47 Also, I strongly feel that it has to be
34:49 a combination of both.
34:52 So it will be humans and machines.
34:55 Humans will bring their thinking intellect
34:58 and keep realigning that on a periodic basis
35:00 into the model, right?
35:02 When one model is made, it's not, it can be,
35:04 it's like a holy grail of investing.
35:06 (speaking in foreign language)
35:08 One would create the model,
35:09 and then I would just sit back and relax
35:10 and go to the Mauritius and set up my thing over there
35:13 and be cool, right?
35:15 Any model that is built also needs
35:17 a constant ongoing review.
35:19 I'm not saying it should be tweaked and changed every day,
35:21 but it needs a human touch again to keep refining it,
35:26 keep reviewing it.
35:29 Likewise, on the other side,
35:31 like you will need the machines
35:33 to help the fundamental analysis also,
35:36 like which will give good factor inputs,
35:39 which will, I'm saying if nothing,
35:41 even just some good technology
35:44 to capture your old history of thinking
35:48 or your old history of data.
35:50 Today when analyzing a company,
35:51 you might want to see historical data,
35:53 even if you want to just do a cyclically adjusted
35:55 P/E ratio of the company
35:57 or earnings cycle of some company, how it is seen.
36:00 You'll need that type of a robust technology
36:02 to support that.
36:03 So definitely for that, my response is very clear.
36:07 Use the power of and versus or.
36:10 I feel very much it's going to be
36:12 both combination of machine and human.
36:14 Quant funds, the second question about
36:17 how I see the way ahead for quant funds,
36:20 most certainly they will be gradually increasing
36:24 in popularity.
36:25 See, there is no magic formula or secret sauce
36:30 which works all the time, like I said in the beginning.
36:32 Like it reminds me of a very famous movie,
36:39 little bit on the digressing away,
36:41 but like it is called Kung Fu Panda.
36:43 Where the panda's father,
36:46 I mean, he's very famous in so,
36:48 but like there is a secret.
36:49 So then it comes to the light that there is no secret,
36:51 so this is your own effort and your own thing.
36:54 So likewise, I feel very strongly that for even quant funds,
36:59 it can never be any one single secret sauce
37:02 or magic formula that you work at all times.
37:05 Quant funds also will have some situations
37:07 where they will underperform in certain cycles,
37:09 certain periods of time.
37:11 So more than the success of the model
37:14 or that number or count,
37:18 it is important to enhance the awareness.
37:21 Like for example, today this session which we are doing,
37:24 one of the main purposes is to increase the awareness
37:27 for people to understand
37:28 what exactly we think of quant investing.
37:30 It's not just high frequency investing
37:31 or it's not something randomly just going on and off,
37:33 we are not even understanding
37:34 or it's not some black box.
37:36 - Sure.
37:37 - Like we explained in simple term,
37:38 today the marathon runners of the city
37:40 also will be able to relate to what quant investing is.
37:44 Just example I'm saying.
37:45 So as the awareness increases,
37:50 automatically investors' mind will get clear
37:53 that in what circumstances or in what market scenarios,
37:56 this one XYZ quant fund or ABC quant strategy
38:00 will work well for my portfolio.
38:02 And they're definitely going to go for that
38:04 and add them in their portfolio.
38:06 - Right.
38:07 - So as awareness increases,
38:10 proper understanding enhances,
38:12 it will increase the acceptance
38:14 and growth of the quant funds or quant strategies,
38:16 you know, in whichever format you think of it.
38:20 - Sure, sure, absolutely.
38:22 So, you know, if you're talking about the power of and,
38:25 I think let's also extend that to partnerships, right?
38:28 So when you consider a strategic partner for quant funds,
38:33 what are the attributes that you would generally look for?
38:36 - Sure, so see, structurally, when we look at a partner,
38:41 I mean, it could be in the form of like a partner
38:43 for multiple things, like, you know,
38:45 one of the first thing, what I mean,
38:47 you could think of is a partner for the data
38:49 and gathering the, running the quant model,
38:54 doing the back testing, you know,
38:55 my go-to partner for a technology platform
38:58 or how I would do my coding.
39:00 So two or three key things that one would look for
39:03 is that the particular platform or partner
39:06 should have a very robust platform
39:09 for absorbing the data, large quantum of data.
39:13 So, you know, making sure
39:15 there are long historical time series of data available,
39:17 which are good quality data.
39:19 So that is crucial,
39:21 because if I don't have good quality data,
39:24 you know, just because of some number going haywire,
39:27 my end result can be completely different.
39:29 Maybe I'm looking at some company's historical
39:32 cash flow moments,
39:33 or I'm looking at how the top line has increased
39:36 and connected to that,
39:37 how the cash flows have been changing.
39:39 But if some of those two data points
39:41 out of this 20 years history I've seen
39:42 are incorrect for whatever reason,
39:45 I will never come to know
39:46 because I'm just relying on the data.
39:49 So data quality and quantity, I mean,
39:53 both are very crucial.
39:54 We need that, what we call is point in time content,
39:58 and we need to ensure that
40:00 the way we say data can be fed to us
40:05 as the user also is very seamless.
40:07 So I'm going to very technical aspect,
40:10 but like, you know,
40:11 there are these file transfer protocol,
40:11 I've got more than FTP,
40:12 FTP and APIs like it.
40:14 So let's say there is some particular technology partner
40:16 with whom, whose services or whose platform you're using.
40:20 But if that cannot talk to my technology platform,
40:24 my system, so my CTO will say,
40:26 "You can't use this, for example."
40:29 So it needs a very robust connectivity
40:33 and customizability for this kind of thing.
40:38 And it should be very easy to use and very flexible.
40:42 See, today I'll be very honest,
40:45 we don't want only people with technology background
40:47 to come and run that.
40:49 So you don't need just a coder,
40:50 you need to, your partner platform that is being used
40:55 should be such that it can absorb,
40:59 simplicity, like a good user interface.
41:02 In the day when you were using computer,
41:04 you need to have that MS-DOS command language,
41:07 you need to remember, right?
41:09 DIR/P, I mean, I don't even know,
41:10 but and then graphical user interface,
41:12 I just press the button, I want this.
41:14 Like that, even this,
41:16 some syntax and some coding, Python,
41:18 some areas are important.
41:20 I'm just saying all this also is history.
41:22 People, young professionals today,
41:25 like for me also,
41:27 I sometimes feel that I need to learn more,
41:29 languages.
41:30 So like I'm already starting to try
41:32 and a little bit get my hands around newer languages
41:34 called Julia, for example.
41:35 Python is very important right now,
41:37 but might be five years later in the future,
41:39 maybe these are even more powerful and faster
41:42 and better languages one has to start learning.
41:44 So that has to be there,
41:45 but still the platform has to have the ease of use
41:49 and flexibility.
41:50 Let's say, suppose suddenly we just decided
41:53 that we need to, while running a model
41:55 or while building a model,
41:56 we want to build in a cost of frictional trades,
41:59 a trading cost or frictional cost.
42:04 Because when I actually implemented,
42:06 I might pay some brokerage,
42:08 some slippage in the execution.
42:10 Suppose we decide,
42:11 "Hey, no, there is a request from top down
42:14 during the investment committee
42:15 that you've taken a frictional cost of 1%,
42:18 but that's not seeming enough.
42:20 You need to take the frictional cost to 1.5%.
42:24 Now, what is my team going to do?
42:27 Will they go back to all of the analytics
42:29 and change everywhere 1%
42:30 and run it back again with 2.5%?"
42:33 Not feasible.
42:34 So the platform should be powerful enough to,
42:36 "Okay, (speaking in foreign language)
42:41 Just put that number over there
42:44 and the entire workings, the model will,
42:47 I mean, ease of implementing,
42:50 accounting for this kind of changes."
42:54 That's crucial when selecting this kind of a platform.
42:58 Partner, yeah.
42:59 - Perfect.
43:00 Superb, Anil.
43:01 I think my takeaways today have been immense,
43:03 but I'll just sum this up for our viewers,
43:07 and then you can definitely interject
43:09 and tell me whatever I might have missed.
43:12 So I think we started off with the fact
43:14 that there are several types of quant strategies.
43:16 So we should not put quant in a certain box.
43:19 A momentum is one type of strategy that's followed,
43:21 but there are many other strategies as well.
43:23 And one needs to be aware what these strategies are
43:27 and then invest accordingly, right?
43:29 Then we also touched upon the aspect of risk,
43:33 black swan events and gray rhinos.
43:36 And I think the conclusion from that point was that
43:39 it's very difficult, one, of course,
43:42 to envisage the magnitude of big risk events,
43:45 but it's also very expensive.
43:47 So then it's better to be a little proactive
43:50 in spotting risk events,
43:52 and in some way or the other,
43:53 incorporating it into our models.
43:55 And like you rightly said,
43:57 a lot of what we discussed is not applicable
43:59 only to quant models,
44:01 but I think to any fund manager,
44:03 any investor who is trying to create a robust portfolio.
44:08 Then on, I think the next point,
44:10 which is essential to our conversation
44:13 is the importance of data.
44:15 There are many, many, many now reams of data,
44:18 which is available.
44:19 There is a traditional data metrics,
44:21 and now there are also new data points
44:24 on structured data, which is coming up,
44:26 which also means that technologies like AI and ML
44:29 are going to gain more traction in the investment world.
44:32 So it's no longer just for the coders and the technologists,
44:35 but also for fund managers.
44:37 And the most important part, I think, at least for me,
44:40 was how to create a robust quant model,
44:43 or like you said, an investment model at large.
44:46 And here, I think what you said in terms of
44:50 the first and foremost point of
44:51 aligning your own philosophy with the model
44:55 is extremely important, right?
44:56 Because I think that's where you kind of
44:58 get rid of all of the biases,
44:59 because you're well aligned with the investment model.
45:02 And then from there on, I think the whole idea
45:05 about breaking your investments into factors, right?
45:09 Instead of just talking about stock A, stock B, stock C,
45:12 but the different factors which contribute to the movement
45:15 in the stock prices and the companies,
45:17 we need to look at that.
45:19 And then of course, at the end of the day,
45:20 we need to be aware of what the models are all about.
45:24 And the future is definitely going to,
45:25 I like the power of and actually.
45:28 Like you said, a lot of what we've discussed today
45:31 is not just about investing,
45:33 but just generally in life as well.
45:35 Instead of having the dichotomy of either or,
45:38 if you have an and, I think you can definitely
45:41 check out more avenues and do well.
45:46 - Yeah.
45:47 That's been, and yes, of course,
45:49 your main view in terms of,
45:51 as awareness of quant funds increases,
45:54 we will be seeing more investors
45:56 probably getting interested in quant funds.
45:58 But that's my summary.
46:00 I would still like you to maybe just give us
46:02 a few closing words.
46:03 - Sure. I think, no, you summed up very well.
46:05 I think it's very important that as an investor,
46:09 once you are aligned with your style of investing
46:13 or what you want out of this particular piece
46:15 of your portfolio.
46:16 - Sure.
46:17 - You could have certain advisors guiding you
46:19 that is always best that, okay,
46:21 some percent of your portfolio has to go into fixed income,
46:24 some into gold, some into real assets
46:26 and some into equity within that.
46:27 Some part, let's say, take it to this kind of a
46:30 quant strategy or quant funds or quant models.
46:33 And then within that, once you are thinking right,
46:36 or you are aligned to how you want to build it,
46:39 that, okay, right now I feel in the market right now,
46:41 there is going to be a shift.
46:42 And now instead of like X factor,
46:45 let's say value is going to work, for example,
46:47 or in Indian markets, structurally,
46:48 we have seen that always momentum and low volatility
46:53 means min-vol and value.
46:57 So you then decide and select that particular fund manager's
47:01 strategy or that particular quant strategy,
47:04 which gives you the best combination of which factor
47:06 or which one or two factors you want.
47:08 - Right.
47:09 - And then take the comfort and go ahead with that.
47:13 And yes, I certainly believe this power of and,
47:17 and in fact, I often conduct short polls
47:20 on my social media platform.
47:22 So I think this is, you have given me, in fact,
47:24 a good question today in the evening.
47:25 I will put up that, I'll try to put up some similar type
47:29 of a message into a poll and let's see how the outcome comes.
47:31 - You know what my answer is going to be.
47:33 - Yes, absolutely.
47:35 - Perfect, perfect, lovely, lovely.
47:36 So with this, we come to the end of today's episode
47:39 of MasterTalk presented by Refinitiv,
47:42 providers of financial technology data and expertise
47:44 to keep your head.
47:46 Thank you Anil, thank you very much for joining us today
47:50 and sharing your thoughts.
47:52 It's been very value-critical for me personally,
47:56 and I think our viewers and all those present
47:59 will definitely agree.
48:00 - Sure, I'm sure.
48:01 Yeah, it was a superb set of questions to be very honest.
48:04 Not very often I've come across such depth of questioning.
48:07 So I was just a little bit, I mean,
48:09 very much alert in terms of now what's the next question,
48:11 when to come.
48:12 So I think it was a very engaging chat,
48:13 I really appreciate that.
48:14 - I'm glad you enjoyed it.
48:15 And viewers, I hope you enjoyed our conversation today
48:20 as well, and thank you for sharing all of your feedbacks
48:23 and comments.
48:25 If you have anything further to share with us,
48:26 please do not hesitate and write to us
48:28 with your suggestions or questions
48:30 at editor@outlookmoney.com.
48:33 And we will be back soon with the fourth episode.
48:36 Until next time, stay safe, stay strong
48:38 and keep following us.
48:39 Thank you.
48:41 (upbeat music)
48:43 (dramatic music)
48:46 [BLANK_AUDIO]