WEBVTT - What It Takes To Win At Quant Investing 0:00:10.960 --> 0:00:15.040 Hello, and welcome to another episode of the Odd Lots podcast. 0:00:15.080 --> 0:00:19.439 I'm Joe Wisn't All and I'm Tracy Halloway. Tracy, you 0:00:19.480 --> 0:00:21.440 know what the funny thing is is that even though 0:00:21.480 --> 0:00:24.880 it's been an incredible year in the stock market, I 0:00:24.880 --> 0:00:28.000 mean just extraordinary biole accounts as everyone knows, I feel 0:00:28.000 --> 0:00:30.040 like it's also probably been a frustrating one for a 0:00:30.080 --> 0:00:33.360 lot of investors. Oh yeah, for sure. I mean, first 0:00:33.360 --> 0:00:36.520 of all, markets didn't really do what a lot of people, 0:00:36.600 --> 0:00:40.080 I guess would would say they should do rationally in 0:00:40.159 --> 0:00:43.760 the face of the biggest economic crisis in decades. But 0:00:43.920 --> 0:00:46.000 I feel like a lot of people just sort of 0:00:46.479 --> 0:00:49.000 missed various turning points in the market as well, and 0:00:49.000 --> 0:00:53.800 are very very frustrated. Absolutely, I mean just super super high, 0:00:53.840 --> 0:00:56.760 super high levels of frustration. Also, even if you were 0:00:56.800 --> 0:00:59.760 along this market and sort of like generally bullish, the 0:01:00.040 --> 0:01:02.400 only way to have really won this year would be 0:01:02.480 --> 0:01:05.520 super concentration in tech stocks. And I feel like if 0:01:05.560 --> 0:01:08.800 you were under exposed to like a handful of tech stocks, 0:01:08.840 --> 0:01:11.759 which we could count down about two hands, then you're 0:01:11.800 --> 0:01:15.160 almost guaranteed to be sort of underperforming your benchmark this year, 0:01:15.240 --> 0:01:18.880 whatever it is Yeah, I think that's absolutely true. And 0:01:18.920 --> 0:01:21.640 of course we've been talking about for years and years 0:01:21.640 --> 0:01:24.120 and years that the big Tex stocks, saying whatever you 0:01:24.120 --> 0:01:27.839 want to call it, are potentially overvalued. So it's it's 0:01:28.000 --> 0:01:31.480 doubly ironic that this year you would have underperformed have 0:01:31.680 --> 0:01:34.440 you not invested in the stocks that people say it 0:01:34.520 --> 0:01:37.920 might be the most overvalued? Right, And of course that 0:01:38.120 --> 0:01:41.240 is a big frustration to investors who have been waiting 0:01:41.319 --> 0:01:44.400 a long time for other sort of factors to do well. 0:01:44.480 --> 0:01:47.560 So investors like to talk into factors and this sort 0:01:47.560 --> 0:01:51.360 of the growth factor has done phenomenally well, but historically 0:01:52.280 --> 0:01:56.120 the value factor, so called cheaper stocks, those have done well, 0:01:56.160 --> 0:01:58.559 and everyone keeps waiting for this turn or for other 0:01:58.720 --> 0:02:03.760 factors to immerge, whether it's value or low beta or 0:02:03.800 --> 0:02:07.720 something else. Uh, never seems to happen. And if anything, 0:02:07.720 --> 0:02:09.960 this year did not prove to be a turning point 0:02:10.040 --> 0:02:12.359 in the market, but really just sort of an accelerant 0:02:12.400 --> 0:02:14.959 of it. Yeah, I think that's right. I'm actually looking 0:02:15.000 --> 0:02:18.519 at a chart from Big of America Merrill Lynch right now, 0:02:18.600 --> 0:02:23.480 and they point out that values relative performance to growth 0:02:24.280 --> 0:02:27.680 was the worst this year since the dot com bubble, 0:02:27.919 --> 0:02:32.400 So um something to remember, But we're not. This isn't 0:02:32.440 --> 0:02:35.960 this podcast isn't about value versus growth? Is it? No, 0:02:36.080 --> 0:02:38.600 it's not. But I think that the frustration that people 0:02:39.160 --> 0:02:41.960 probably have this year does lead to um, you know, 0:02:42.040 --> 0:02:46.160 people looking for other approaches to investing, and of course 0:02:46.200 --> 0:02:49.280 in times like this, people wonder if, like maybe other 0:02:49.400 --> 0:02:53.600 sort of quantitative or algorithmic strategies more money should be 0:02:53.680 --> 0:02:56.880 poured into them as an alternative to this ride where 0:02:56.880 --> 0:02:58.920 you just sort of by the big tech s docks 0:02:58.919 --> 0:03:02.239 and I hope that you you know, avoid the turning point. Well, 0:03:02.280 --> 0:03:03.919 I guess another way of putting it is a lot 0:03:03.919 --> 0:03:07.119 of the a lot of the quant strategies are sort 0:03:07.160 --> 0:03:09.839 of momentum based, right, So if you can figure out 0:03:10.200 --> 0:03:14.040 where the money is flowing to, even if it's tex stocks, Uh, 0:03:14.080 --> 0:03:15.840 that might be a good way of investing in the 0:03:15.840 --> 0:03:19.880 current environment. If everything's about liquidity and following the flows, 0:03:20.160 --> 0:03:23.600 then quant investing or algorithmic trading, whatever you want to 0:03:23.600 --> 0:03:26.640 call it, might be a good way forward. Yeah. But 0:03:26.960 --> 0:03:29.400 you know, backing up, it's like we talk about quant 0:03:29.520 --> 0:03:32.760 investing and the word quant gets used all the time, 0:03:32.800 --> 0:03:36.480 and sometimes uh, it's used to describe these super technical funds, 0:03:36.480 --> 0:03:39.320 and sometimes it gets used to just describe sort of 0:03:39.640 --> 0:03:43.800 anything that has some statistical analysis of it. That that 0:03:44.080 --> 0:03:49.240 term feels extremely vague. Yeah, and potentially overused as well. Right, 0:03:49.280 --> 0:03:53.200 Like everyone wants to seem like they are quantitative in 0:03:53.360 --> 0:03:55.720 some way or another. No one wants to say that 0:03:55.760 --> 0:03:59.360 they're investing purely on emotion and gut feeling and that 0:03:59.440 --> 0:04:02.720 kind of stuff. So quant gets bandied about quite a bit. 0:04:03.640 --> 0:04:06.400 So today we are going to talk with an expert 0:04:06.440 --> 0:04:10.040 who is ah, knows a lot about quant investing studies 0:04:10.080 --> 0:04:13.760 that can help us define it and uh also hopefully 0:04:13.840 --> 0:04:16.240 sort of explain to us what it takes to win 0:04:16.320 --> 0:04:18.640 in this space, because again, everyone sort of wants to 0:04:18.720 --> 0:04:22.599 be in the space, even you know, traditional hedge funds 0:04:23.040 --> 0:04:26.040 over the years have allocated more and more money to quant, 0:04:26.120 --> 0:04:29.560 to hiring PhD s, to building up their computer systems, 0:04:30.240 --> 0:04:33.160 But what it really takes to win and can lots 0:04:33.160 --> 0:04:36.960 of players succeed is still uh kind of an open question. Yeah, 0:04:37.040 --> 0:04:39.480 I think that's exactly right. And as we're going to discuss, 0:04:39.640 --> 0:04:43.600 quant investing is probably one of the most expensive ventures 0:04:43.640 --> 0:04:46.599 that you can sort of embark on. Yes, Okay, So 0:04:46.800 --> 0:04:48.680 without further Ado. Let's bring in our guest. He is 0:04:48.680 --> 0:04:51.960 an expert in the field. He is ciamac Millmy. He 0:04:52.080 --> 0:04:55.320 is a professor of business. He's a professor at the 0:04:55.360 --> 0:04:58.599 Columbia Business School, done a lot of research in the 0:04:58.680 --> 0:05:02.880 area of quant investor. He's also a part time partner 0:05:03.400 --> 0:05:06.760 at a fund himself. Thank you very much for joining us. 0:05:07.040 --> 0:05:10.440 Thanks for having me. I'm delighted to be here. When 0:05:10.480 --> 0:05:13.800 I say quant investing or when people say quant investing, 0:05:14.360 --> 0:05:16.200 what does that mean to you? Like, how would you 0:05:16.240 --> 0:05:18.960 just define that term and so that it's a useful 0:05:19.160 --> 0:05:23.120 so that it's a useful term. Well, people have different definitions. 0:05:23.160 --> 0:05:27.520 I personally define it as having two key characteristics. Um. 0:05:27.560 --> 0:05:32.120 The first characteristic is that the investment process is entirely systematic. 0:05:32.680 --> 0:05:36.080 So there's many different times of types of investment strategies 0:05:36.520 --> 0:05:40.920 that the people implement that employ at some level quantitative methods. 0:05:40.960 --> 0:05:43.800 But I think the key to the quantitative methods that 0:05:44.040 --> 0:05:46.440 we're going to speak about today is that at the 0:05:46.520 --> 0:05:49.760 trade by trade level, there is no discretion. Right. Um, 0:05:49.839 --> 0:05:54.240 You've you've set up an algorithm, a particular system on 0:05:54.279 --> 0:05:56.520 a you know, second by second trade by trade basis, 0:05:56.800 --> 0:05:59.599 the everything is being automatically done. You know. That isn't 0:05:59.600 --> 0:06:02.359 to say that there isn't like a portfolio manager involved. 0:06:02.520 --> 0:06:04.479 But the job of the portfolio manager is not so 0:06:04.560 --> 0:06:06.720 much deciding on trades and sizing them and so on, 0:06:06.960 --> 0:06:10.400 but more setting up the computer algorithms in advance and 0:06:10.440 --> 0:06:13.120 tweaking them and improving them over time. So so that's 0:06:13.120 --> 0:06:18.279 really the first big component to be entirely systematic nondiscretionary. 0:06:18.680 --> 0:06:21.280 The second component of the ones that I focus on 0:06:21.400 --> 0:06:24.640 is that they're really active investment strategies in the sense 0:06:24.680 --> 0:06:27.360 that you're buying now because you think the asset will 0:06:27.400 --> 0:06:29.920 be worth more later it's it's mispriced in some level, 0:06:30.200 --> 0:06:32.720 or alternatively, you're selling short now because you think the 0:06:33.360 --> 0:06:36.080 value later will be will be lower. There are other 0:06:36.200 --> 0:06:40.080 flavors of quantitative strategies that are somewhat more passive, things 0:06:40.160 --> 0:06:44.200 like uh, you know, exotic beta, investing in factors and 0:06:44.240 --> 0:06:47.160 so on. Um, those are not so much a little 0:06:47.160 --> 0:06:49.279 bit less my area, and I have my own views, 0:06:49.279 --> 0:06:51.240 and then we can get into later perhaps, But the 0:06:51.320 --> 0:06:55.080 key things I'm thinking about here, you're using algorithms and 0:06:55.160 --> 0:06:57.280 data and machine learning and so on. You're taking an 0:06:57.320 --> 0:07:00.920 active view on what the current prices are relative to 0:07:01.120 --> 0:07:04.280 what you know the value might be later. So is 0:07:04.360 --> 0:07:08.120 quant investing proof that markets aren't efficient? I feel like 0:07:08.160 --> 0:07:10.640 this comes up a lot, but maybe it's worth asking 0:07:10.680 --> 0:07:13.720 this question early on. If the whole strategy is to 0:07:13.920 --> 0:07:18.640 automatically arbitrage price discrepancies in the short term versus the 0:07:18.640 --> 0:07:23.480 long term, does that mean that markets aren't doing their job? Well? 0:07:23.520 --> 0:07:24.920 I mean, I think if you want to sort of 0:07:24.960 --> 0:07:27.520 take the straw man that the markets are, you know, 0:07:27.600 --> 0:07:30.840 sort of a hundred percent efficient and prices are incorporating 0:07:30.840 --> 0:07:33.800 all potential information, I think that's clearly not true. And 0:07:33.840 --> 0:07:37.200 I think, um, the long term success uh and incredible 0:07:37.240 --> 0:07:40.920 performance of you know, quant investors like Renaissance is is 0:07:40.960 --> 0:07:43.800 sort of one piece of that um. But that doesn't 0:07:43.840 --> 0:07:48.239 mean that the markets are completely uh inefficient either. LASSA 0:07:48.320 --> 0:07:50.320 Peterson who's from from n y U and a q R. 0:07:50.400 --> 0:07:53.080 He has that he has a nice phrase called inefficiently efficient, 0:07:53.400 --> 0:07:57.720 or I should say efficiently inefficient, meaning that there are inefficiencies, 0:07:57.760 --> 0:07:59.840 but it's a competitive game and there are lots of 0:08:00.000 --> 0:08:02.800 smart people with you know, a lot of resources going 0:08:02.840 --> 0:08:06.480 after these inefficiencies, and when you identify them and trade 0:08:06.480 --> 0:08:09.880 on them, Um, they disappear, they're armed away. So you know, 0:08:09.960 --> 0:08:13.880 these these inefficiencies typically lie around the frontier of the 0:08:13.920 --> 0:08:17.320 transaction costs, of what it costs to trade. So, um, yes, 0:08:17.400 --> 0:08:20.400 there are inefficiencies, but they're they're hard to find, and 0:08:20.480 --> 0:08:23.880 you know, they disappear over time. So one common concept 0:08:23.960 --> 0:08:27.000 that quants talk about is is alpha decay. Like you 0:08:27.000 --> 0:08:30.680 you identify some some signal or some inefficiency and uh, 0:08:30.760 --> 0:08:32.600 you know, generates a certain amount of P and L 0:08:33.120 --> 0:08:36.160 and literally year over year you can see that decay away. 0:08:36.240 --> 0:08:38.839 And you know that's that's because that inefficiency eventually is 0:08:38.880 --> 0:08:41.440 identified by other people and as more and more people 0:08:41.480 --> 0:08:44.080 trade on it, you know again it disappears. So it's 0:08:44.120 --> 0:08:46.360 not that um, you set up an algorithm and it 0:08:46.559 --> 0:08:49.560 just you know, sort of prints money. Um uh you know, 0:08:49.720 --> 0:08:52.240 some sort of gross violation of the the efficient markets 0:08:52.280 --> 0:08:54.480 hypoth is That's that's not how it works. The people 0:08:54.520 --> 0:08:58.680 who are successful at this are constantly investing and deploying 0:08:58.960 --> 0:09:03.040 enormous resources, hiring large numbers of PhD s, and uhum, 0:09:03.120 --> 0:09:06.320 progressively innovating in order to have new models because because 0:09:06.320 --> 0:09:09.840 the old stuff will simply stop working. So it sounds 0:09:09.920 --> 0:09:11.560 like I mean I guess you just said it, but 0:09:11.720 --> 0:09:13.959 it sounds like the key to winning. And we'll get 0:09:14.000 --> 0:09:17.720 more granular in a second. Is that continuous process. It's 0:09:17.760 --> 0:09:20.920 not about identifying some flaw in the market or some 0:09:21.080 --> 0:09:25.000 inefficiency or some opportunity to make money. It's about having 0:09:25.120 --> 0:09:29.080 a team and a process to keep finding those over 0:09:29.120 --> 0:09:33.400 and over again. That's right again, because all the inefficiencies 0:09:33.440 --> 0:09:52.160 that I've ever seen are are short lived. So can 0:09:52.160 --> 0:09:54.160 you maybe um talk to us a little bit more 0:09:54.160 --> 0:09:58.200 than about how a quant strategy might be developed. So 0:09:58.280 --> 0:10:02.240 obviously you have the techno logical aspect of it, the 0:10:02.400 --> 0:10:07.199 need for computers that are able to trade very very quickly. 0:10:07.840 --> 0:10:10.200 You have the need for servers, many of them co 0:10:10.360 --> 0:10:14.440 located close to the exchanges. But then you also have 0:10:14.920 --> 0:10:19.800 proprietary data sets sometimes and then you have proprietary algorithms. 0:10:19.800 --> 0:10:23.600 So how does that all come together into one quant strategy? 0:10:23.720 --> 0:10:26.600 And which one of those is sort of the most 0:10:27.400 --> 0:10:30.160 or the biggest investment for a quant firm? Got it? 0:10:30.240 --> 0:10:34.240 So I think there's definitely a technological investment, may or 0:10:34.280 --> 0:10:37.480 may not involve things like co location near the exchanges, 0:10:38.000 --> 0:10:40.960 so at least anecdotally. For example, Renaissance, which is the 0:10:41.000 --> 0:10:45.120 most successful quantitative firm does not co locate. You know, again, 0:10:45.360 --> 0:10:47.120 I don't know, but that's that's that's what I've heard. 0:10:47.240 --> 0:10:50.880 Co location is quite important when you're trading UH and 0:10:51.120 --> 0:10:54.439 you require very low latency and and that's typically the 0:10:54.520 --> 0:10:57.240 high frequency trading domain and which again intersects with with 0:10:57.400 --> 0:11:00.240 quant in many ways. But if you're looking UM a 0:11:00.320 --> 0:11:02.560 little bit longer, if your horizons are a little bit longer, 0:11:02.920 --> 0:11:05.599 it becomes a little bit a little bit less important. 0:11:05.600 --> 0:11:08.080 Your broader point, I think is correct. Technology is important. 0:11:08.360 --> 0:11:11.840 I think I'm more important. It's kind of a research process. 0:11:12.080 --> 0:11:14.160 There there's a number of kind of high level pieces 0:11:14.200 --> 0:11:17.600 to a successful quantitative strategy. It's not like uh, UM 0:11:17.600 --> 0:11:20.240 there's just a black box and UM in gooes data 0:11:20.320 --> 0:11:23.480 outgoes trades. There's there's a number of pieces in there 0:11:23.800 --> 0:11:26.480 that UM sort of split the problem into to kind 0:11:26.520 --> 0:11:29.160 of make it manageable. UM at the front end UM. 0:11:29.200 --> 0:11:31.559 You know, going back to the heart of active investing, 0:11:31.840 --> 0:11:33.840 you've got to have a view on asset prices. Right, 0:11:33.880 --> 0:11:36.320 so you're trading some universes I don't know, US equities 0:11:36.400 --> 0:11:38.360 something like that, you've got to have a view stock 0:11:38.400 --> 0:11:39.880 by stock what's the price is going to be in 0:11:39.920 --> 0:11:42.760 a day? Um, uh, two weeks, a month, so on 0:11:42.840 --> 0:11:45.800 and so forth, right and so UM that front end 0:11:45.960 --> 0:11:51.320 is called signal generation or generating alpha's right, using data 0:11:51.400 --> 0:11:55.080 and machine learning techniques to come up with anomalies to 0:11:55.480 --> 0:11:57.680 that that you identify, and then you build models upon 0:11:57.960 --> 0:12:01.080 to sort of make a prediction of, um, what the 0:12:01.280 --> 0:12:02.760 what the what the price is going to be. So 0:12:03.000 --> 0:12:07.520 there's all sorts of types of data and uh algorithms 0:12:07.520 --> 0:12:11.000 that people use. Historically, much of quant investment has been 0:12:11.280 --> 0:12:14.720 building um what are called um quote unquote technical models, 0:12:15.000 --> 0:12:18.960 wherein basically you're using historical price and trade data to 0:12:19.040 --> 0:12:22.440 forecast future price movements. Right, so you might think of 0:12:22.800 --> 0:12:26.840 things like momentum or reversals or so on and so forth. 0:12:26.880 --> 0:12:29.720 That's you know, leveraging you know, kind of purely um 0:12:29.720 --> 0:12:33.160 technical data from the markets. UM. What we've seen emerged 0:12:33.160 --> 0:12:35.679 really over the past ten years is there's also been 0:12:35.720 --> 0:12:38.880 a shift to sort of quote unquote alternative data. Right, 0:12:38.920 --> 0:12:41.760 so you might look at things like you know, everybody's 0:12:41.760 --> 0:12:44.200 heard the famous story of satellite images of parking lots, 0:12:44.320 --> 0:12:47.079 right to try and assess you know, um, is you know, 0:12:47.120 --> 0:12:49.560 what's the occupancy at Walmart. This you're gonna they're gonna 0:12:49.600 --> 0:12:52.040 make their earnings. You know. Quantitative investment would take that 0:12:52.080 --> 0:12:55.080 kind of data and leverage it to a model which forecasts, Okay, um, 0:12:55.120 --> 0:12:56.680 what's the return going to be up for Walmart over 0:12:56.720 --> 0:12:58.520 the next week, the next month, next two months, and 0:12:58.520 --> 0:13:02.000 so on. Right, So the front end you have um 0:13:01.800 --> 0:13:05.360 uh this this identifying the data combined with the machine 0:13:05.440 --> 0:13:09.679 learning technology which is going to build build predictions. Now, 0:13:10.000 --> 0:13:12.480 oftentimes you're looking at um or I should say really 0:13:12.480 --> 0:13:16.480 always these days, you're looking at having many, many um anomalies. 0:13:16.520 --> 0:13:19.240 So you may have a technical model based on momentum 0:13:19.240 --> 0:13:21.280 and reversals. You may have bought a whole bunch of 0:13:21.360 --> 0:13:23.320 parking lot data, you have some some model for the 0:13:23.360 --> 0:13:25.680 retail sector based on that. You have some some credit 0:13:25.679 --> 0:13:28.640 card data, some social media data, maybe some some news data. 0:13:28.880 --> 0:13:31.160 You have all of these. And so the second part 0:13:31.160 --> 0:13:34.040 of the process is to kind of uh combine these 0:13:34.080 --> 0:13:37.040 different types of signals or views into sort of one 0:13:37.080 --> 0:13:39.280 compositive view because at the end of the day, um, 0:13:39.400 --> 0:13:41.400 all you care about is is net and that is 0:13:41.440 --> 0:13:43.240 this asset price is going to go up or go down, 0:13:43.480 --> 0:13:46.199 and and and that part is called alpha mixing or 0:13:46.240 --> 0:13:49.040 signal mixing. Right, Um, you have these these separate models 0:13:49.080 --> 0:13:50.920 that that that you've built, and you want to combine 0:13:50.960 --> 0:13:54.920 them to one kind of uh composite view. So um, 0:13:54.960 --> 0:13:57.720 that's that's kind of the front end again, having a 0:13:57.800 --> 0:13:59.680 view on what prices are going to be over the 0:13:59.720 --> 0:14:04.280 other relevant timeframes. Historically that is where the vast majority 0:14:04.320 --> 0:14:07.360 of U the energy was spent. The idea was that 0:14:07.440 --> 0:14:09.960 if you have a good signals, if you have good predictions, 0:14:10.240 --> 0:14:12.680 you can make money. If you don't have good signals, 0:14:12.720 --> 0:14:14.280 you're not going to make money, and the rest of 0:14:14.320 --> 0:14:16.400 it doesn't matter so much. So I believe if you 0:14:16.400 --> 0:14:18.040 don't have signals, you're not going to make money. That's 0:14:18.040 --> 0:14:21.520 certainly true. But these days the market has gotten competitive 0:14:21.640 --> 0:14:24.520 enough and there are enough kind of quant players that, um, 0:14:24.560 --> 0:14:26.840 what you do with the signals also matters how you 0:14:26.880 --> 0:14:29.840 try to to monetize them. So here the kind of 0:14:29.840 --> 0:14:32.840 the next step is that you have Now you know, 0:14:32.960 --> 0:14:34.960 you're waking up to open to the market. It's nine 0:14:35.000 --> 0:14:37.800 thirty in the morning, right, you have a prediction for 0:14:37.960 --> 0:14:40.640 you know, a universe of three thousand US equities. Now 0:14:40.680 --> 0:14:42.800 you have to kind of decide, um, what's the target 0:14:42.800 --> 0:14:44.920 portfolio you want to form? So that's kind of a 0:14:45.160 --> 0:14:48.080 called a portfolio construction case, right, And so the kind 0:14:48.120 --> 0:14:51.840 of things you're thinking about are balancing sort of risk 0:14:52.000 --> 0:14:54.280 versus return. You know, you don't want to be um 0:14:54.400 --> 0:14:57.520 um uh longer short, maybe you want to be market neutral. 0:14:57.680 --> 0:15:00.320 You don't want too much exposure in the individual set ters, 0:15:00.400 --> 0:15:02.720 you know, UM, so on and so forth. Right, UM, 0:15:02.720 --> 0:15:05.880 you're balancing that also with with with with transaction costs 0:15:05.840 --> 0:15:08.320 and so on, and you kind of decide like, um, 0:15:08.520 --> 0:15:10.520 you know again based on what my current view is 0:15:10.560 --> 0:15:13.200 of the world, UM, what's the target portfolio I want 0:15:13.200 --> 0:15:16.000 to hold? And this is something you periodically revisit. It 0:15:16.120 --> 0:15:18.400 used to be sort of um quants sort of you know, 0:15:18.400 --> 0:15:20.880 traded once a day and had a trade list the 0:15:20.920 --> 0:15:23.240 beginning of the day and you know, um generated trades 0:15:23.280 --> 0:15:25.240 and revisited the next day. Now it's much more of 0:15:25.280 --> 0:15:28.280 a continuous procedure because you know, as as the market 0:15:28.280 --> 0:15:30.320 evolves and as you get more data and news comes 0:15:30.320 --> 0:15:33.040 out and so on, those underlying views which you're driving 0:15:33.040 --> 0:15:36.280 the trades are changing. So so that's the kind of 0:15:36.280 --> 0:15:39.240 the middle piece, UM, figuring out what what portfolio to hold, 0:15:39.440 --> 0:15:41.600 and and then the final piece is actually um, sort 0:15:41.600 --> 0:15:44.520 of generating the trades. Sometimes quants do this themselves. I 0:15:44.520 --> 0:15:46.800 think more and more quants of doing this themselves. Um, 0:15:46.840 --> 0:15:50.320 you can farm this also to UH. Basically every major 0:15:50.560 --> 0:15:53.840 bank or prime broker that services UH quants has an 0:15:53.920 --> 0:15:56.360 agency algorithms desk that will do this for you. But 0:15:56.440 --> 0:15:58.680 here here the idea is, Okay, I decided I need 0:15:58.720 --> 0:16:01.560 to buy two million dollars of Google stock over the 0:16:01.240 --> 0:16:04.520 uh the next fifteen minutes. Um, how can I do that? Um? 0:16:04.600 --> 0:16:07.120 You know? Should I use exchanges? Should I use dark pools? Um? 0:16:07.280 --> 0:16:09.960 How should I uh spread that out over time? Um? 0:16:09.960 --> 0:16:12.200 You know, should I use limit orders? Market orders? Um? 0:16:12.240 --> 0:16:15.960 This kind of thing? And uh again, Historically, UM, you know, 0:16:16.120 --> 0:16:18.360 people focus a little bit less on that, but now 0:16:18.400 --> 0:16:20.720 as the market has gotten more more competitive, it's also 0:16:21.120 --> 0:16:23.720 being important. If if you're not doing those latter two phases, 0:16:23.720 --> 0:16:27.840 the portfolio construction and the trade optimization, well you're you're 0:16:27.960 --> 0:16:29.880 leaving money on the table in a way that almost 0:16:29.920 --> 0:16:32.360 may may not be may not be profitable. I think 0:16:32.400 --> 0:16:34.520 one thing that's that's not obvious that or I should say, 0:16:34.560 --> 0:16:37.240 it's quite different about plant trading versus uh, other types 0:16:37.280 --> 0:16:39.360 of hedge fund trading. If you look at a guy 0:16:39.520 --> 0:16:41.560 like um, you know, I don't know, just to sort 0:16:41.560 --> 0:16:44.040 of pick someone random like Bill Ackman, right, um, when 0:16:44.040 --> 0:16:46.720 when he goes in and buys a stock, he has like, 0:16:46.800 --> 0:16:50.440 you know, really kind of strong um conviction. He takes 0:16:50.440 --> 0:16:52.800 some massive positions, and he also he probably expects to 0:16:52.800 --> 0:16:55.320 make or you know, something like that. Again, I don't 0:16:55.320 --> 0:16:56.800 do that type of trading. I don't know, but he 0:16:56.840 --> 0:16:59.360 expects to make tens of percents, right, a quant in 0:16:59.400 --> 0:17:02.280 any individual position, you probably measure your expected profit in 0:17:02.320 --> 0:17:04.879 basis points, right, And it's all this and you know, 0:17:04.960 --> 0:17:06.800 you might expect to make three basis points on the 0:17:06.840 --> 0:17:09.600 transaction cost of two basis points, right, So you really 0:17:09.680 --> 0:17:13.000 like carefully controlling your costs and managing execution and so 0:17:13.080 --> 0:17:15.840 on is extremely important. Like you know, Bill Ackman, if 0:17:15.840 --> 0:17:18.800 he thinks he's going to make on a particular trade, 0:17:18.960 --> 0:17:21.480 it doesn't matter if he's paying uh, you know, two 0:17:21.520 --> 0:17:23.800 basis points for twenty basis points or even a hundred 0:17:23.800 --> 0:17:25.840 basis points, right, he's going to make so much more 0:17:25.840 --> 0:17:28.960 in his mind, it's irrelevant. Whereas for for for quants, 0:17:28.960 --> 0:17:31.760 you're really operating on a very thin margin. First of all, 0:17:31.760 --> 0:17:35.400 that was a sort of great explanation of the whole process, 0:17:35.400 --> 0:17:37.560 really nice overview, But I want to go back to 0:17:37.720 --> 0:17:41.280 just these sort of search for the original signals or 0:17:41.440 --> 0:17:44.919 search for this sort of the the initial inputs. And 0:17:44.920 --> 0:17:49.280 I'm thinking about large tech companies like Microsoft and Google 0:17:49.560 --> 0:17:52.600 and Facebook and how they have a lot of like 0:17:52.800 --> 0:17:56.400 researchers who are engaged in sort of pure tech research 0:17:57.000 --> 0:17:59.600 and you know, always out there filing patents, and there's 0:17:59.600 --> 0:18:03.280 probably a long sort of distance between anything that they 0:18:03.320 --> 0:18:07.280 discover and their own research budget um and then what 0:18:07.680 --> 0:18:10.800 ultimately might show up in a consumer product or a 0:18:10.840 --> 0:18:13.600 business product. And I'm curious if there is sort of 0:18:13.880 --> 0:18:17.720 an analogy in quant land where you have people who 0:18:17.760 --> 0:18:21.120 really are sort of at the frontier without a sort 0:18:21.160 --> 0:18:24.000 of crystal clear idea of okay, this is going to 0:18:24.080 --> 0:18:27.199 lead to something that will turn into a trade. But 0:18:27.320 --> 0:18:30.800 it's that process of sort of really exploring that frontier 0:18:30.840 --> 0:18:35.520 which eventually leads to concrete ideas that do lead to trades. 0:18:35.840 --> 0:18:38.720 And I'm curious if that's sort of like the analogy 0:18:38.760 --> 0:18:43.600 and how investors and how the portfolio managers think about 0:18:44.280 --> 0:18:47.480 where to explore and where those frontiers are, and where 0:18:47.520 --> 0:18:51.639 to invest expensive sort of time, energy and computing power 0:18:52.119 --> 0:18:56.320 in discovering these alpha generating signals. So I think quantity 0:18:56.359 --> 0:18:59.439 of investors operate quite differently than some of the research 0:18:59.480 --> 0:19:01.480 groups and in big tech places, like if you go 0:19:01.560 --> 0:19:04.480 to a place like Google Research or Microsoft Research, it's 0:19:04.600 --> 0:19:08.040 really not that different than an academic institution. Um, their 0:19:08.080 --> 0:19:11.800 their main output is really papers, right in journal papers, 0:19:11.800 --> 0:19:14.760 conference papers, so on and so forth. Uh, And it's 0:19:14.800 --> 0:19:18.320 really just a different way to do uh, almost academic research, 0:19:18.400 --> 0:19:21.560 kind of the classical Bell Labs model. And and maybe 0:19:22.080 --> 0:19:24.320 I mean they they do consult on internal projects and 0:19:24.359 --> 0:19:26.359 so forth. But I think in the in the in 0:19:26.400 --> 0:19:28.760 the quant world, it is much much much more applied. 0:19:29.280 --> 0:19:31.840 So I think typically the kind of thing would be like, um, 0:19:32.160 --> 0:19:35.240 you think, you know, maybe someone comes to you a vendor, 0:19:35.160 --> 0:19:37.960 or you identify a data set that you might that 0:19:38.000 --> 0:19:41.000 you think might have some relevance. You start looking at 0:19:41.400 --> 0:19:44.199 building various models of trying to predict prices or you know, 0:19:44.240 --> 0:19:46.560 things that are relatively of relevant to prices. You try 0:19:46.560 --> 0:19:50.439 and parent some different machine learning kind of techniques. But 0:19:50.520 --> 0:19:54.840 I think from the beginning it's really oriented around concrete 0:19:54.880 --> 0:19:57.439 things like let me build a price for let me 0:19:57.440 --> 0:19:59.240 build a model, sorry for for what the return of 0:19:59.240 --> 0:20:01.280 this asset is going to be over the next month, right, 0:20:01.359 --> 0:20:04.240 Or let me build a model for how I should 0:20:04.440 --> 0:20:08.760 efficiently trade large box of stock over the next fifteen minutes. 0:20:09.000 --> 0:20:13.080 Broadly speaking, it's much less of them sort of blue 0:20:13.119 --> 0:20:15.560 sky research that that isn't to say that some people 0:20:15.560 --> 0:20:17.560 don't do that. I think, uh, I think people do. 0:20:17.880 --> 0:20:21.159 But um, the the incentives aren't there because you know, 0:20:21.200 --> 0:20:25.760 for the most part speaking practitioners, um, there's no publishing, right, 0:20:25.840 --> 0:20:30.080 and um, I think people are extremely paranoid and sensitive 0:20:30.359 --> 0:20:33.760 because if if your IP leaks and other people do 0:20:33.880 --> 0:20:37.160 similar things, maybe what you do will stop working as well. 0:20:37.400 --> 0:20:40.320 And so there's, uh, there's not that much of an 0:20:40.359 --> 0:20:43.080 incentive to have to do that versus the very kind 0:20:43.080 --> 0:20:47.600 of visceral incentive of you know, making money, having you know, 0:20:47.600 --> 0:20:49.840 outperforming in the market in the in the short term. 0:20:50.160 --> 0:20:53.000 So so research in the quant world, for the most part, 0:20:53.080 --> 0:20:56.200 tends to be much more implied. I have a sort 0:20:56.240 --> 0:21:00.160 of related question, but why is why is quant investing 0:21:00.240 --> 0:21:04.480 or why are quants so um secretive about everything? Or 0:21:04.880 --> 0:21:06.439 I mean I don't want to call them weird, but 0:21:06.520 --> 0:21:11.160 there is this sort of like odd culture around quant investing. 0:21:11.400 --> 0:21:14.560 And you think of places like Renaissance and Citadel, they're 0:21:14.600 --> 0:21:18.480 all sort of shrouded in mystique. I once heard that 0:21:18.520 --> 0:21:22.520 Citadel had an original Enigma machine from World War Two 0:21:22.600 --> 0:21:24.800 in one of its offices. I don't know if that's true, 0:21:24.840 --> 0:21:26.840