1 00:00:10,960 --> 00:00:15,040 Speaker 1: Hello, and welcome to another episode of the Odd Lots podcast. 2 00:00:15,080 --> 00:00:19,439 Speaker 1: I'm Joe Wisn't All and I'm Tracy Halloway. Tracy, you 3 00:00:19,480 --> 00:00:21,440 Speaker 1: know what the funny thing is is that even though 4 00:00:21,480 --> 00:00:24,880 Speaker 1: it's been an incredible year in the stock market, I 5 00:00:24,880 --> 00:00:28,000 Speaker 1: mean just extraordinary biole accounts as everyone knows, I feel 6 00:00:28,000 --> 00:00:30,040 Speaker 1: like it's also probably been a frustrating one for a 7 00:00:30,080 --> 00:00:33,360 Speaker 1: lot of investors. Oh yeah, for sure. I mean, first 8 00:00:33,360 --> 00:00:36,520 Speaker 1: of all, markets didn't really do what a lot of people, 9 00:00:36,600 --> 00:00:40,080 Speaker 1: I guess would would say they should do rationally in 10 00:00:40,159 --> 00:00:43,760 Speaker 1: the face of the biggest economic crisis in decades. But 11 00:00:43,920 --> 00:00:46,000 Speaker 1: I feel like a lot of people just sort of 12 00:00:46,479 --> 00:00:49,000 Speaker 1: missed various turning points in the market as well, and 13 00:00:49,000 --> 00:00:53,800 Speaker 1: are very very frustrated. Absolutely, I mean just super super high, 14 00:00:53,840 --> 00:00:56,760 Speaker 1: super high levels of frustration. Also, even if you were 15 00:00:56,800 --> 00:00:59,760 Speaker 1: along this market and sort of like generally bullish, the 16 00:01:00,040 --> 00:01:02,400 Speaker 1: only way to have really won this year would be 17 00:01:02,480 --> 00:01:05,520 Speaker 1: super concentration in tech stocks. And I feel like if 18 00:01:05,560 --> 00:01:08,800 Speaker 1: you were under exposed to like a handful of tech stocks, 19 00:01:08,840 --> 00:01:11,759 Speaker 1: which we could count down about two hands, then you're 20 00:01:11,800 --> 00:01:15,160 Speaker 1: almost guaranteed to be sort of underperforming your benchmark this year, 21 00:01:15,240 --> 00:01:18,880 Speaker 1: whatever it is Yeah, I think that's absolutely true. And 22 00:01:18,920 --> 00:01:21,640 Speaker 1: of course we've been talking about for years and years 23 00:01:21,640 --> 00:01:24,120 Speaker 1: and years that the big Tex stocks, saying whatever you 24 00:01:24,120 --> 00:01:27,839 Speaker 1: want to call it, are potentially overvalued. So it's it's 25 00:01:28,000 --> 00:01:31,480 Speaker 1: doubly ironic that this year you would have underperformed have 26 00:01:31,680 --> 00:01:34,440 Speaker 1: you not invested in the stocks that people say it 27 00:01:34,520 --> 00:01:37,920 Speaker 1: might be the most overvalued? Right, And of course that 28 00:01:38,120 --> 00:01:41,240 Speaker 1: is a big frustration to investors who have been waiting 29 00:01:41,319 --> 00:01:44,400 Speaker 1: a long time for other sort of factors to do well. 30 00:01:44,480 --> 00:01:47,560 Speaker 1: So investors like to talk into factors and this sort 31 00:01:47,560 --> 00:01:51,360 Speaker 1: of the growth factor has done phenomenally well, but historically 32 00:01:52,280 --> 00:01:56,120 Speaker 1: the value factor, so called cheaper stocks, those have done well, 33 00:01:56,160 --> 00:01:58,559 Speaker 1: and everyone keeps waiting for this turn or for other 34 00:01:58,720 --> 00:02:03,760 Speaker 1: factors to immerge, whether it's value or low beta or 35 00:02:03,800 --> 00:02:07,720 Speaker 1: something else. Uh, never seems to happen. And if anything, 36 00:02:07,720 --> 00:02:09,960 Speaker 1: this year did not prove to be a turning point 37 00:02:10,040 --> 00:02:12,359 Speaker 1: in the market, but really just sort of an accelerant 38 00:02:12,400 --> 00:02:14,959 Speaker 1: of it. Yeah, I think that's right. I'm actually looking 39 00:02:15,000 --> 00:02:18,519 Speaker 1: at a chart from Big of America Merrill Lynch right now, 40 00:02:18,600 --> 00:02:23,480 Speaker 1: and they point out that values relative performance to growth 41 00:02:24,280 --> 00:02:27,680 Speaker 1: was the worst this year since the dot com bubble, 42 00:02:27,919 --> 00:02:32,400 Speaker 1: So um something to remember, But we're not. This isn't 43 00:02:32,440 --> 00:02:35,960 Speaker 1: this podcast isn't about value versus growth? Is it? No, 44 00:02:36,080 --> 00:02:38,600 Speaker 1: it's not. But I think that the frustration that people 45 00:02:39,160 --> 00:02:41,960 Speaker 1: probably have this year does lead to um, you know, 46 00:02:42,040 --> 00:02:46,160 Speaker 1: people looking for other approaches to investing, and of course 47 00:02:46,200 --> 00:02:49,280 Speaker 1: in times like this, people wonder if, like maybe other 48 00:02:49,400 --> 00:02:53,600 Speaker 1: sort of quantitative or algorithmic strategies more money should be 49 00:02:53,680 --> 00:02:56,880 Speaker 1: poured into them as an alternative to this ride where 50 00:02:56,880 --> 00:02:58,920 Speaker 1: you just sort of by the big tech s docks 51 00:02:58,919 --> 00:03:02,239 Speaker 1: and I hope that you you know, avoid the turning point. Well, 52 00:03:02,280 --> 00:03:03,919 Speaker 1: I guess another way of putting it is a lot 53 00:03:03,919 --> 00:03:07,119 Speaker 1: of the a lot of the quant strategies are sort 54 00:03:07,160 --> 00:03:09,839 Speaker 1: of momentum based, right, So if you can figure out 55 00:03:10,200 --> 00:03:14,040 Speaker 1: where the money is flowing to, even if it's tex stocks, Uh, 56 00:03:14,080 --> 00:03:15,840 Speaker 1: that might be a good way of investing in the 57 00:03:15,840 --> 00:03:19,880 Speaker 1: current environment. If everything's about liquidity and following the flows, 58 00:03:20,160 --> 00:03:23,600 Speaker 1: then quant investing or algorithmic trading, whatever you want to 59 00:03:23,600 --> 00:03:26,640 Speaker 1: call it, might be a good way forward. Yeah. But 60 00:03:26,960 --> 00:03:29,400 Speaker 1: you know, backing up, it's like we talk about quant 61 00:03:29,520 --> 00:03:32,760 Speaker 1: investing and the word quant gets used all the time, 62 00:03:32,800 --> 00:03:36,480 Speaker 1: and sometimes uh, it's used to describe these super technical funds, 63 00:03:36,480 --> 00:03:39,320 Speaker 1: and sometimes it gets used to just describe sort of 64 00:03:39,640 --> 00:03:43,800 Speaker 1: anything that has some statistical analysis of it. That that 65 00:03:44,080 --> 00:03:49,240 Speaker 1: term feels extremely vague. Yeah, and potentially overused as well. Right, 66 00:03:49,280 --> 00:03:53,200 Speaker 1: Like everyone wants to seem like they are quantitative in 67 00:03:53,360 --> 00:03:55,720 Speaker 1: some way or another. No one wants to say that 68 00:03:55,760 --> 00:03:59,360 Speaker 1: they're investing purely on emotion and gut feeling and that 69 00:03:59,440 --> 00:04:02,720 Speaker 1: kind of stuff. So quant gets bandied about quite a bit. 70 00:04:03,640 --> 00:04:06,400 Speaker 1: So today we are going to talk with an expert 71 00:04:06,440 --> 00:04:10,040 Speaker 1: who is ah, knows a lot about quant investing studies 72 00:04:10,080 --> 00:04:13,760 Speaker 1: that can help us define it and uh also hopefully 73 00:04:13,840 --> 00:04:16,240 Speaker 1: sort of explain to us what it takes to win 74 00:04:16,320 --> 00:04:18,640 Speaker 1: in this space, because again, everyone sort of wants to 75 00:04:18,720 --> 00:04:22,599 Speaker 1: be in the space, even you know, traditional hedge funds 76 00:04:23,040 --> 00:04:26,040 Speaker 1: over the years have allocated more and more money to quant, 77 00:04:26,120 --> 00:04:29,560 Speaker 1: to hiring PhD s, to building up their computer systems, 78 00:04:30,240 --> 00:04:33,160 Speaker 1: But what it really takes to win and can lots 79 00:04:33,160 --> 00:04:36,960 Speaker 1: of players succeed is still uh kind of an open question. Yeah, 80 00:04:37,040 --> 00:04:39,480 Speaker 1: I think that's exactly right. And as we're going to discuss, 81 00:04:39,640 --> 00:04:43,600 Speaker 1: quant investing is probably one of the most expensive ventures 82 00:04:43,640 --> 00:04:46,599 Speaker 1: that you can sort of embark on. Yes, Okay, So 83 00:04:46,800 --> 00:04:48,680 Speaker 1: without further Ado. Let's bring in our guest. He is 84 00:04:48,680 --> 00:04:51,960 Speaker 1: an expert in the field. He is ciamac Millmy. He 85 00:04:52,080 --> 00:04:55,320 Speaker 1: is a professor of business. He's a professor at the 86 00:04:55,360 --> 00:04:58,599 Speaker 1: Columbia Business School, done a lot of research in the 87 00:04:58,680 --> 00:05:02,880 Speaker 1: area of quant investor. He's also a part time partner 88 00:05:03,400 --> 00:05:06,760 Speaker 1: at a fund himself. Thank you very much for joining us. 89 00:05:07,040 --> 00:05:10,440 Speaker 1: Thanks for having me. I'm delighted to be here. When 90 00:05:10,480 --> 00:05:13,800 Speaker 1: I say quant investing or when people say quant investing, 91 00:05:14,360 --> 00:05:16,200 Speaker 1: what does that mean to you? Like, how would you 92 00:05:16,240 --> 00:05:18,960 Speaker 1: just define that term and so that it's a useful 93 00:05:19,160 --> 00:05:23,120 Speaker 1: so that it's a useful term. Well, people have different definitions. 94 00:05:23,160 --> 00:05:27,520 Speaker 1: I personally define it as having two key characteristics. Um. 95 00:05:27,560 --> 00:05:32,120 Speaker 1: The first characteristic is that the investment process is entirely systematic. 96 00:05:32,680 --> 00:05:36,080 Speaker 1: So there's many different times of types of investment strategies 97 00:05:36,520 --> 00:05:40,920 Speaker 1: that the people implement that employ at some level quantitative methods. 98 00:05:40,960 --> 00:05:43,800 Speaker 1: But I think the key to the quantitative methods that 99 00:05:44,040 --> 00:05:46,440 Speaker 1: we're going to speak about today is that at the 100 00:05:46,520 --> 00:05:49,760 Speaker 1: trade by trade level, there is no discretion. Right. Um, 101 00:05:49,839 --> 00:05:54,240 Speaker 1: You've you've set up an algorithm, a particular system on 102 00:05:54,279 --> 00:05:56,520 Speaker 1: a you know, second by second trade by trade basis, 103 00:05:56,800 --> 00:05:59,599 Speaker 1: the everything is being automatically done. You know. That isn't 104 00:05:59,600 --> 00:06:02,359 Speaker 1: to say that there isn't like a portfolio manager involved. 105 00:06:02,520 --> 00:06:04,479 Speaker 1: But the job of the portfolio manager is not so 106 00:06:04,560 --> 00:06:06,720 Speaker 1: much deciding on trades and sizing them and so on, 107 00:06:06,960 --> 00:06:10,400 Speaker 1: but more setting up the computer algorithms in advance and 108 00:06:10,440 --> 00:06:13,120 Speaker 1: tweaking them and improving them over time. So so that's 109 00:06:13,120 --> 00:06:18,279 Speaker 1: really the first big component to be entirely systematic nondiscretionary. 110 00:06:18,680 --> 00:06:21,280 Speaker 1: The second component of the ones that I focus on 111 00:06:21,400 --> 00:06:24,640 Speaker 1: is that they're really active investment strategies in the sense 112 00:06:24,680 --> 00:06:27,360 Speaker 1: that you're buying now because you think the asset will 113 00:06:27,400 --> 00:06:29,920 Speaker 1: be worth more later it's it's mispriced in some level, 114 00:06:30,200 --> 00:06:32,720 Speaker 1: or alternatively, you're selling short now because you think the 115 00:06:33,360 --> 00:06:36,080 Speaker 1: value later will be will be lower. There are other 116 00:06:36,200 --> 00:06:40,080 Speaker 1: flavors of quantitative strategies that are somewhat more passive, things 117 00:06:40,160 --> 00:06:44,200 Speaker 1: like uh, you know, exotic beta, investing in factors and 118 00:06:44,240 --> 00:06:47,160 Speaker 1: so on. Um, those are not so much a little 119 00:06:47,160 --> 00:06:49,279 Speaker 1: bit less my area, and I have my own views, 120 00:06:49,279 --> 00:06:51,240 Speaker 1: and then we can get into later perhaps, But the 121 00:06:51,320 --> 00:06:55,080 Speaker 1: key things I'm thinking about here, you're using algorithms and 122 00:06:55,160 --> 00:06:57,280 Speaker 1: data and machine learning and so on. You're taking an 123 00:06:57,320 --> 00:07:00,920 Speaker 1: active view on what the current prices are relative to 124 00:07:01,120 --> 00:07:04,280 Speaker 1: what you know the value might be later. So is 125 00:07:04,360 --> 00:07:08,120 Speaker 1: quant investing proof that markets aren't efficient? I feel like 126 00:07:08,160 --> 00:07:10,640 Speaker 1: this comes up a lot, but maybe it's worth asking 127 00:07:10,680 --> 00:07:13,720 Speaker 1: this question early on. If the whole strategy is to 128 00:07:13,920 --> 00:07:18,640 Speaker 1: automatically arbitrage price discrepancies in the short term versus the 129 00:07:18,640 --> 00:07:23,480 Speaker 1: long term, does that mean that markets aren't doing their job? Well? 130 00:07:23,520 --> 00:07:24,920 Speaker 1: I mean, I think if you want to sort of 131 00:07:24,960 --> 00:07:27,520 Speaker 1: take the straw man that the markets are, you know, 132 00:07:27,600 --> 00:07:30,840 Speaker 1: sort of a hundred percent efficient and prices are incorporating 133 00:07:30,840 --> 00:07:33,800 Speaker 1: all potential information, I think that's clearly not true. And 134 00:07:33,840 --> 00:07:37,200 Speaker 1: I think, um, the long term success uh and incredible 135 00:07:37,240 --> 00:07:40,920 Speaker 1: performance of you know, quant investors like Renaissance is is 136 00:07:40,960 --> 00:07:43,800 Speaker 1: sort of one piece of that um. But that doesn't 137 00:07:43,840 --> 00:07:48,239 Speaker 1: mean that the markets are completely uh inefficient either. LASSA 138 00:07:48,320 --> 00:07:50,320 Speaker 1: Peterson who's from from n y U and a q R. 139 00:07:50,400 --> 00:07:53,080 Speaker 1: He has that he has a nice phrase called inefficiently efficient, 140 00:07:53,400 --> 00:07:57,720 Speaker 1: or I should say efficiently inefficient, meaning that there are inefficiencies, 141 00:07:57,760 --> 00:07:59,840 Speaker 1: but it's a competitive game and there are lots of 142 00:08:00,000 --> 00:08:02,800 Speaker 1: smart people with you know, a lot of resources going 143 00:08:02,840 --> 00:08:06,480 Speaker 1: after these inefficiencies, and when you identify them and trade 144 00:08:06,480 --> 00:08:09,880 Speaker 1: on them, Um, they disappear, they're armed away. So you know, 145 00:08:09,960 --> 00:08:13,880 Speaker 1: these these inefficiencies typically lie around the frontier of the 146 00:08:13,920 --> 00:08:17,320 Speaker 1: transaction costs, of what it costs to trade. So, um, yes, 147 00:08:17,400 --> 00:08:20,400 Speaker 1: there are inefficiencies, but they're they're hard to find, and 148 00:08:20,480 --> 00:08:23,880 Speaker 1: you know, they disappear over time. So one common concept 149 00:08:23,960 --> 00:08:27,000 Speaker 1: that quants talk about is is alpha decay. Like you 150 00:08:27,000 --> 00:08:30,680 Speaker 1: you identify some some signal or some inefficiency and uh, 151 00:08:30,760 --> 00:08:32,600 Speaker 1: you know, generates a certain amount of P and L 152 00:08:33,120 --> 00:08:36,160 Speaker 1: and literally year over year you can see that decay away. 153 00:08:36,240 --> 00:08:38,839 Speaker 1: And you know that's that's because that inefficiency eventually is 154 00:08:38,880 --> 00:08:41,440 Speaker 1: identified by other people and as more and more people 155 00:08:41,480 --> 00:08:44,080 Speaker 1: trade on it, you know again it disappears. So it's 156 00:08:44,120 --> 00:08:46,360 Speaker 1: not that um, you set up an algorithm and it 157 00:08:46,559 --> 00:08:49,560 Speaker 1: just you know, sort of prints money. Um uh you know, 158 00:08:49,720 --> 00:08:52,240 Speaker 1: some sort of gross violation of the the efficient markets 159 00:08:52,280 --> 00:08:54,480 Speaker 1: hypoth is That's that's not how it works. The people 160 00:08:54,520 --> 00:08:58,680 Speaker 1: who are successful at this are constantly investing and deploying 161 00:08:58,960 --> 00:09:03,040 Speaker 1: enormous resources, hiring large numbers of PhD s, and uhum, 162 00:09:03,120 --> 00:09:06,320 Speaker 1: progressively innovating in order to have new models because because 163 00:09:06,320 --> 00:09:09,840 Speaker 1: the old stuff will simply stop working. So it sounds 164 00:09:09,920 --> 00:09:11,560 Speaker 1: like I mean I guess you just said it, but 165 00:09:11,720 --> 00:09:13,959 Speaker 1: it sounds like the key to winning. And we'll get 166 00:09:14,000 --> 00:09:17,720 Speaker 1: more granular in a second. Is that continuous process. It's 167 00:09:17,760 --> 00:09:20,920 Speaker 1: not about identifying some flaw in the market or some 168 00:09:21,080 --> 00:09:25,000 Speaker 1: inefficiency or some opportunity to make money. It's about having 169 00:09:25,120 --> 00:09:29,080 Speaker 1: a team and a process to keep finding those over 170 00:09:29,120 --> 00:09:33,400 Speaker 1: and over again. That's right again, because all the inefficiencies 171 00:09:33,440 --> 00:09:52,160 Speaker 1: that I've ever seen are are short lived. So can 172 00:09:52,160 --> 00:09:54,160 Speaker 1: you maybe um talk to us a little bit more 173 00:09:54,160 --> 00:09:58,200 Speaker 1: than about how a quant strategy might be developed. So 174 00:09:58,280 --> 00:10:02,240 Speaker 1: obviously you have the techno logical aspect of it, the 175 00:10:02,400 --> 00:10:07,199 Speaker 1: need for computers that are able to trade very very quickly. 176 00:10:07,840 --> 00:10:10,200 Speaker 1: You have the need for servers, many of them co 177 00:10:10,360 --> 00:10:14,440 Speaker 1: located close to the exchanges. But then you also have 178 00:10:14,920 --> 00:10:19,800 Speaker 1: proprietary data sets sometimes and then you have proprietary algorithms. 179 00:10:19,800 --> 00:10:23,600 Speaker 1: So how does that all come together into one quant strategy? 180 00:10:23,720 --> 00:10:26,600 Speaker 1: And which one of those is sort of the most 181 00:10:27,400 --> 00:10:30,160 Speaker 1: or the biggest investment for a quant firm? Got it? 182 00:10:30,240 --> 00:10:34,240 Speaker 1: So I think there's definitely a technological investment, may or 183 00:10:34,280 --> 00:10:37,480 Speaker 1: may not involve things like co location near the exchanges, 184 00:10:38,000 --> 00:10:40,960 Speaker 1: so at least anecdotally. For example, Renaissance, which is the 185 00:10:41,000 --> 00:10:45,120 Speaker 1: most successful quantitative firm does not co locate. You know, again, 186 00:10:45,360 --> 00:10:47,120 Speaker 1: I don't know, but that's that's that's what I've heard. 187 00:10:47,240 --> 00:10:50,880 Speaker 1: Co location is quite important when you're trading UH and 188 00:10:51,120 --> 00:10:54,439 Speaker 1: you require very low latency and and that's typically the 189 00:10:54,520 --> 00:10:57,240 Speaker 1: high frequency trading domain and which again intersects with with 190 00:10:57,400 --> 00:11:00,240 Speaker 1: quant in many ways. But if you're looking UM a 191 00:11:00,320 --> 00:11:02,560 Speaker 1: little bit longer, if your horizons are a little bit longer, 192 00:11:02,920 --> 00:11:05,599 Speaker 1: it becomes a little bit a little bit less important. 193 00:11:05,600 --> 00:11:08,080 Speaker 1: Your broader point, I think is correct. Technology is important. 194 00:11:08,360 --> 00:11:11,840 Speaker 1: I think I'm more important. It's kind of a research process. 195 00:11:12,080 --> 00:11:14,160 Speaker 1: There there's a number of kind of high level pieces 196 00:11:14,200 --> 00:11:17,600 Speaker 1: to a successful quantitative strategy. It's not like uh, UM 197 00:11:17,600 --> 00:11:20,240 Speaker 1: there's just a black box and UM in gooes data 198 00:11:20,320 --> 00:11:23,480 Speaker 1: outgoes trades. There's there's a number of pieces in there 199 00:11:23,800 --> 00:11:26,480 Speaker 1: that UM sort of split the problem into to kind 200 00:11:26,520 --> 00:11:29,160 Speaker 1: of make it manageable. UM at the front end UM. 201 00:11:29,200 --> 00:11:31,559 Speaker 1: You know, going back to the heart of active investing, 202 00:11:31,840 --> 00:11:33,840 Speaker 1: you've got to have a view on asset prices. Right, 203 00:11:33,880 --> 00:11:36,320 Speaker 1: so you're trading some universes I don't know, US equities 204 00:11:36,400 --> 00:11:38,360 Speaker 1: something like that, you've got to have a view stock 205 00:11:38,400 --> 00:11:39,880 Speaker 1: by stock what's the price is going to be in 206 00:11:39,920 --> 00:11:42,760 Speaker 1: a day? Um, uh, two weeks, a month, so on 207 00:11:42,840 --> 00:11:45,800 Speaker 1: and so forth, right and so UM that front end 208 00:11:45,960 --> 00:11:51,320 Speaker 1: is called signal generation or generating alpha's right, using data 209 00:11:51,400 --> 00:11:55,080 Speaker 1: and machine learning techniques to come up with anomalies to 210 00:11:55,480 --> 00:11:57,680 Speaker 1: that that you identify, and then you build models upon 211 00:11:57,960 --> 00:12:01,080 Speaker 1: to sort of make a prediction of, um, what the 212 00:12:01,280 --> 00:12:02,760 Speaker 1: what the what the price is going to be. So 213 00:12:03,000 --> 00:12:07,520 Speaker 1: there's all sorts of types of data and uh algorithms 214 00:12:07,520 --> 00:12:11,000 Speaker 1: that people use. Historically, much of quant investment has been 215 00:12:11,280 --> 00:12:14,720 Speaker 1: building um what are called um quote unquote technical models, 216 00:12:15,000 --> 00:12:18,960 Speaker 1: wherein basically you're using historical price and trade data to 217 00:12:19,040 --> 00:12:22,440 Speaker 1: forecast future price movements. Right, so you might think of 218 00:12:22,800 --> 00:12:26,840 Speaker 1: things like momentum or reversals or so on and so forth. 219 00:12:26,880 --> 00:12:29,720 Speaker 1: That's you know, leveraging you know, kind of purely um 220 00:12:29,720 --> 00:12:33,160 Speaker 1: technical data from the markets. UM. What we've seen emerged 221 00:12:33,160 --> 00:12:35,679 Speaker 1: really over the past ten years is there's also been 222 00:12:35,720 --> 00:12:38,880 Speaker 1: a shift to sort of quote unquote alternative data. Right, 223 00:12:38,920 --> 00:12:41,760 Speaker 1: so you might look at things like you know, everybody's 224 00:12:41,760 --> 00:12:44,200 Speaker 1: heard the famous story of satellite images of parking lots, 225 00:12:44,320 --> 00:12:47,079 Speaker 1: right to try and assess you know, um, is you know, 226 00:12:47,120 --> 00:12:49,560 Speaker 1: what's the occupancy at Walmart. This you're gonna they're gonna 227 00:12:49,600 --> 00:12:52,040 Speaker 1: make their earnings. You know. Quantitative investment would take that 228 00:12:52,080 --> 00:12:55,080 Speaker 1: kind of data and leverage it to a model which forecasts, Okay, um, 229 00:12:55,120 --> 00:12:56,680 Speaker 1: what's the return going to be up for Walmart over 230 00:12:56,720 --> 00:12:58,520 Speaker 1: the next week, the next month, next two months, and 231 00:12:58,520 --> 00:13:02,000 Speaker 1: so on. Right, So the front end you have um 232 00:13:01,800 --> 00:13:05,360 Speaker 1: uh this this identifying the data combined with the machine 233 00:13:05,440 --> 00:13:09,679 Speaker 1: learning technology which is going to build build predictions. Now, 234 00:13:10,000 --> 00:13:12,480 Speaker 1: oftentimes you're looking at um or I should say really 235 00:13:12,480 --> 00:13:16,480 Speaker 1: always these days, you're looking at having many, many um anomalies. 236 00:13:16,520 --> 00:13:19,240 Speaker 1: So you may have a technical model based on momentum 237 00:13:19,240 --> 00:13:21,280 Speaker 1: and reversals. You may have bought a whole bunch of 238 00:13:21,360 --> 00:13:23,320 Speaker 1: parking lot data, you have some some model for the 239 00:13:23,360 --> 00:13:25,680 Speaker 1: retail sector based on that. You have some some credit 240 00:13:25,679 --> 00:13:28,640 Speaker 1: card data, some social media data, maybe some some news data. 241 00:13:28,880 --> 00:13:31,160 Speaker 1: You have all of these. And so the second part 242 00:13:31,160 --> 00:13:34,040 Speaker 1: of the process is to kind of uh combine these 243 00:13:34,080 --> 00:13:37,040 Speaker 1: different types of signals or views into sort of one 244 00:13:37,080 --> 00:13:39,280 Speaker 1: compositive view because at the end of the day, um, 245 00:13:39,400 --> 00:13:41,400 Speaker 1: all you care about is is net and that is 246 00:13:41,440 --> 00:13:43,240 Speaker 1: this asset price is going to go up or go down, 247 00:13:43,480 --> 00:13:46,199 Speaker 1: and and and that part is called alpha mixing or 248 00:13:46,240 --> 00:13:49,040 Speaker 1: signal mixing. Right, Um, you have these these separate models 249 00:13:49,080 --> 00:13:50,920 Speaker 1: that that that you've built, and you want to combine 250 00:13:50,960 --> 00:13:54,920 Speaker 1: them to one kind of uh composite view. So um, 251 00:13:54,960 --> 00:13:57,720 Speaker 1: that's that's kind of the front end again, having a 252 00:13:57,800 --> 00:13:59,680 Speaker 1: view on what prices are going to be over the 253 00:13:59,720 --> 00:14:04,280 Speaker 1: other relevant timeframes. Historically that is where the vast majority 254 00:14:04,320 --> 00:14:07,360 Speaker 1: of U the energy was spent. The idea was that 255 00:14:07,440 --> 00:14:09,960 Speaker 1: if you have a good signals, if you have good predictions, 256 00:14:10,240 --> 00:14:12,680 Speaker 1: you can make money. If you don't have good signals, 257 00:14:12,720 --> 00:14:14,280 Speaker 1: you're not going to make money, and the rest of 258 00:14:14,320 --> 00:14:16,400 Speaker 1: it doesn't matter so much. So I believe if you 259 00:14:16,400 --> 00:14:18,040 Speaker 1: don't have signals, you're not going to make money. That's 260 00:14:18,040 --> 00:14:21,520 Speaker 1: certainly true. But these days the market has gotten competitive 261 00:14:21,640 --> 00:14:24,520 Speaker 1: enough and there are enough kind of quant players that, um, 262 00:14:24,560 --> 00:14:26,840 Speaker 1: what you do with the signals also matters how you 263 00:14:26,880 --> 00:14:29,840 Speaker 1: try to to monetize them. So here the kind of 264 00:14:29,840 --> 00:14:32,840 Speaker 1: the next step is that you have Now you know, 265 00:14:32,960 --> 00:14:34,960 Speaker 1: you're waking up to open to the market. It's nine 266 00:14:35,000 --> 00:14:37,800 Speaker 1: thirty in the morning, right, you have a prediction for 267 00:14:37,960 --> 00:14:40,640 Speaker 1: you know, a universe of three thousand US equities. Now 268 00:14:40,680 --> 00:14:42,800 Speaker 1: you have to kind of decide, um, what's the target 269 00:14:42,800 --> 00:14:44,920 Speaker 1: portfolio you want to form? So that's kind of a 270 00:14:45,160 --> 00:14:48,080 Speaker 1: called a portfolio construction case, right, And so the kind 271 00:14:48,120 --> 00:14:51,840 Speaker 1: of things you're thinking about are balancing sort of risk 272 00:14:52,000 --> 00:14:54,280 Speaker 1: versus return. You know, you don't want to be um 273 00:14:54,400 --> 00:14:57,520 Speaker 1: um uh longer short, maybe you want to be market neutral. 274 00:14:57,680 --> 00:15:00,320 Speaker 1: You don't want too much exposure in the individual set ters, 275 00:15:00,400 --> 00:15:02,720 Speaker 1: you know, UM, so on and so forth. Right, UM, 276 00:15:02,720 --> 00:15:05,880 Speaker 1: you're balancing that also with with with with transaction costs 277 00:15:05,840 --> 00:15:08,320 Speaker 1: and so on, and you kind of decide like, um, 278 00:15:08,520 --> 00:15:10,520 Speaker 1: you know again based on what my current view is 279 00:15:10,560 --> 00:15:13,200 Speaker 1: of the world, UM, what's the target portfolio I want 280 00:15:13,200 --> 00:15:16,000 Speaker 1: to hold? And this is something you periodically revisit. It 281 00:15:16,120 --> 00:15:18,400 Speaker 1: used to be sort of um quants sort of you know, 282 00:15:18,400 --> 00:15:20,880 Speaker 1: traded once a day and had a trade list the 283 00:15:20,920 --> 00:15:23,240 Speaker 1: beginning of the day and you know, um generated trades 284 00:15:23,280 --> 00:15:25,240 Speaker 1: and revisited the next day. Now it's much more of 285 00:15:25,280 --> 00:15:28,280 Speaker 1: a continuous procedure because you know, as as the market 286 00:15:28,280 --> 00:15:30,320 Speaker 1: evolves and as you get more data and news comes 287 00:15:30,320 --> 00:15:33,040 Speaker 1: out and so on, those underlying views which you're driving 288 00:15:33,040 --> 00:15:36,280 Speaker 1: the trades are changing. So so that's the kind of 289 00:15:36,280 --> 00:15:39,240 Speaker 1: the middle piece, UM, figuring out what what portfolio to hold, 290 00:15:39,440 --> 00:15:41,600 Speaker 1: and and then the final piece is actually um, sort 291 00:15:41,600 --> 00:15:44,520 Speaker 1: of generating the trades. Sometimes quants do this themselves. I 292 00:15:44,520 --> 00:15:46,800 Speaker 1: think more and more quants of doing this themselves. Um, 293 00:15:46,840 --> 00:15:50,320 Speaker 1: you can farm this also to UH. Basically every major 294 00:15:50,560 --> 00:15:53,840 Speaker 1: bank or prime broker that services UH quants has an 295 00:15:53,920 --> 00:15:56,360 Speaker 1: agency algorithms desk that will do this for you. But 296 00:15:56,440 --> 00:15:58,680 Speaker 1: here here the idea is, Okay, I decided I need 297 00:15:58,720 --> 00:16:01,560 Speaker 1: to buy two million dollars of Google stock over the 298 00:16:01,240 --> 00:16:04,520 Speaker 1: uh the next fifteen minutes. Um, how can I do that? Um? 299 00:16:04,600 --> 00:16:07,120 Speaker 1: You know? Should I use exchanges? Should I use dark pools? Um? 300 00:16:07,280 --> 00:16:09,960 Speaker 1: How should I uh spread that out over time? Um? 301 00:16:09,960 --> 00:16:12,200 Speaker 1: You know, should I use limit orders? Market orders? Um? 302 00:16:12,240 --> 00:16:15,960 Speaker 1: This kind of thing? And uh again, Historically, UM, you know, 303 00:16:16,120 --> 00:16:18,360 Speaker 1: people focus a little bit less on that, but now 304 00:16:18,400 --> 00:16:20,720 Speaker 1: as the market has gotten more more competitive, it's also 305 00:16:21,120 --> 00:16:23,720 Speaker 1: being important. If if you're not doing those latter two phases, 306 00:16:23,720 --> 00:16:27,840 Speaker 1: the portfolio construction and the trade optimization, well you're you're 307 00:16:27,960 --> 00:16:29,880 Speaker 1: leaving money on the table in a way that almost 308 00:16:29,920 --> 00:16:32,360 Speaker 1: may may not be may not be profitable. I think 309 00:16:32,400 --> 00:16:34,520 Speaker 1: one thing that's that's not obvious that or I should say, 310 00:16:34,560 --> 00:16:37,240 Speaker 1: it's quite different about plant trading versus uh, other types 311 00:16:37,280 --> 00:16:39,360 Speaker 1: of hedge fund trading. If you look at a guy 312 00:16:39,520 --> 00:16:41,560 Speaker 1: like um, you know, I don't know, just to sort 313 00:16:41,560 --> 00:16:44,040 Speaker 1: of pick someone random like Bill Ackman, right, um, when 314 00:16:44,040 --> 00:16:46,720 Speaker 1: when he goes in and buys a stock, he has like, 315 00:16:46,800 --> 00:16:50,440 Speaker 1: you know, really kind of strong um conviction. He takes 316 00:16:50,440 --> 00:16:52,800 Speaker 1: some massive positions, and he also he probably expects to 317 00:16:52,800 --> 00:16:55,320 Speaker 1: make or you know, something like that. Again, I don't 318 00:16:55,320 --> 00:16:56,800 Speaker 1: do that type of trading. I don't know, but he 319 00:16:56,840 --> 00:16:59,360 Speaker 1: expects to make tens of percents, right, a quant in 320 00:16:59,400 --> 00:17:02,280 Speaker 1: any individual position, you probably measure your expected profit in 321 00:17:02,320 --> 00:17:04,879 Speaker 1: basis points, right, And it's all this and you know, 322 00:17:04,960 --> 00:17:06,800 Speaker 1: you might expect to make three basis points on the 323 00:17:06,840 --> 00:17:09,600 Speaker 1: transaction cost of two basis points, right, So you really 324 00:17:09,680 --> 00:17:13,000 Speaker 1: like carefully controlling your costs and managing execution and so 325 00:17:13,080 --> 00:17:15,840 Speaker 1: on is extremely important. Like you know, Bill Ackman, if 326 00:17:15,840 --> 00:17:18,800 Speaker 1: he thinks he's going to make on a particular trade, 327 00:17:18,960 --> 00:17:21,480 Speaker 1: it doesn't matter if he's paying uh, you know, two 328 00:17:21,520 --> 00:17:23,800 Speaker 1: basis points for twenty basis points or even a hundred 329 00:17:23,800 --> 00:17:25,840 Speaker 1: basis points, right, he's going to make so much more 330 00:17:25,840 --> 00:17:28,960 Speaker 1: in his mind, it's irrelevant. Whereas for for for quants, 331 00:17:28,960 --> 00:17:31,760 Speaker 1: you're really operating on a very thin margin. First of all, 332 00:17:31,760 --> 00:17:35,400 Speaker 1: that was a sort of great explanation of the whole process, 333 00:17:35,400 --> 00:17:37,560 Speaker 1: really nice overview, But I want to go back to 334 00:17:37,720 --> 00:17:41,280 Speaker 1: just these sort of search for the original signals or 335 00:17:41,440 --> 00:17:44,919 Speaker 1: search for this sort of the the initial inputs. And 336 00:17:44,920 --> 00:17:49,280 Speaker 1: I'm thinking about large tech companies like Microsoft and Google 337 00:17:49,560 --> 00:17:52,600 Speaker 1: and Facebook and how they have a lot of like 338 00:17:52,800 --> 00:17:56,400 Speaker 1: researchers who are engaged in sort of pure tech research 339 00:17:57,000 --> 00:17:59,600 Speaker 1: and you know, always out there filing patents, and there's 340 00:17:59,600 --> 00:18:03,280 Speaker 1: probably a long sort of distance between anything that they 341 00:18:03,320 --> 00:18:07,280 Speaker 1: discover and their own research budget um and then what 342 00:18:07,680 --> 00:18:10,800 Speaker 1: ultimately might show up in a consumer product or a 343 00:18:10,840 --> 00:18:13,600 Speaker 1: business product. And I'm curious if there is sort of 344 00:18:13,880 --> 00:18:17,720 Speaker 1: an analogy in quant land where you have people who 345 00:18:17,760 --> 00:18:21,120 Speaker 1: really are sort of at the frontier without a sort 346 00:18:21,160 --> 00:18:24,000 Speaker 1: of crystal clear idea of okay, this is going to 347 00:18:24,080 --> 00:18:27,199 Speaker 1: lead to something that will turn into a trade. But 348 00:18:27,320 --> 00:18:30,800 Speaker 1: it's that process of sort of really exploring that frontier 349 00:18:30,840 --> 00:18:35,520 Speaker 1: which eventually leads to concrete ideas that do lead to trades. 350 00:18:35,840 --> 00:18:38,720 Speaker 1: And I'm curious if that's sort of like the analogy 351 00:18:38,760 --> 00:18:43,600 Speaker 1: and how investors and how the portfolio managers think about 352 00:18:44,280 --> 00:18:47,480 Speaker 1: where to explore and where those frontiers are, and where 353 00:18:47,520 --> 00:18:51,639 Speaker 1: to invest expensive sort of time, energy and computing power 354 00:18:52,119 --> 00:18:56,320 Speaker 1: in discovering these alpha generating signals. So I think quantity 355 00:18:56,359 --> 00:18:59,439 Speaker 1: of investors operate quite differently than some of the research 356 00:18:59,480 --> 00:19:01,480 Speaker 1: groups and in big tech places, like if you go 357 00:19:01,560 --> 00:19:04,480 Speaker 1: to a place like Google Research or Microsoft Research, it's 358 00:19:04,600 --> 00:19:08,040 Speaker 1: really not that different than an academic institution. Um, their 359 00:19:08,080 --> 00:19:11,800 Speaker 1: their main output is really papers, right in journal papers, 360 00:19:11,800 --> 00:19:14,760 Speaker 1: conference papers, so on and so forth. Uh, And it's 361 00:19:14,800 --> 00:19:18,320 Speaker 1: really just a different way to do uh, almost academic research, 362 00:19:18,400 --> 00:19:21,560 Speaker 1: kind of the classical Bell Labs model. And and maybe 363 00:19:22,080 --> 00:19:24,320 Speaker 1: I mean they they do consult on internal projects and 364 00:19:24,359 --> 00:19:26,359 Speaker 1: so forth. But I think in the in the in 365 00:19:26,400 --> 00:19:28,760 Speaker 1: the quant world, it is much much much more applied. 366 00:19:29,280 --> 00:19:31,840 Speaker 1: So I think typically the kind of thing would be like, um, 367 00:19:32,160 --> 00:19:35,240 Speaker 1: you think, you know, maybe someone comes to you a vendor, 368 00:19:35,160 --> 00:19:37,960 Speaker 1: or you identify a data set that you might that 369 00:19:38,000 --> 00:19:41,000 Speaker 1: you think might have some relevance. You start looking at 370 00:19:41,400 --> 00:19:44,199 Speaker 1: building various models of trying to predict prices or you know, 371 00:19:44,240 --> 00:19:46,560 Speaker 1: things that are relatively of relevant to prices. You try 372 00:19:46,560 --> 00:19:50,439 Speaker 1: and parent some different machine learning kind of techniques. But 373 00:19:50,520 --> 00:19:54,840 Speaker 1: I think from the beginning it's really oriented around concrete 374 00:19:54,880 --> 00:19:57,439 Speaker 1: things like let me build a price for let me 375 00:19:57,440 --> 00:19:59,240 Speaker 1: build a model, sorry for for what the return of 376 00:19:59,240 --> 00:20:01,280 Speaker 1: this asset is going to be over the next month, right, 377 00:20:01,359 --> 00:20:04,240 Speaker 1: Or let me build a model for how I should 378 00:20:04,440 --> 00:20:08,760 Speaker 1: efficiently trade large box of stock over the next fifteen minutes. 379 00:20:09,000 --> 00:20:13,080 Speaker 1: Broadly speaking, it's much less of them sort of blue 380 00:20:13,119 --> 00:20:15,560 Speaker 1: sky research that that isn't to say that some people 381 00:20:15,560 --> 00:20:17,560 Speaker 1: don't do that. I think, uh, I think people do. 382 00:20:17,880 --> 00:20:21,159 Speaker 1: But um, the the incentives aren't there because you know, 383 00:20:21,200 --> 00:20:25,760 Speaker 1: for the most part speaking practitioners, um, there's no publishing, right, 384 00:20:25,840 --> 00:20:30,080 Speaker 1: and um, I think people are extremely paranoid and sensitive 385 00:20:30,359 --> 00:20:33,760 Speaker 1: because if if your IP leaks and other people do 386 00:20:33,880 --> 00:20:37,160 Speaker 1: similar things, maybe what you do will stop working as well. 387 00:20:37,400 --> 00:20:40,320 Speaker 1: And so there's, uh, there's not that much of an 388 00:20:40,359 --> 00:20:43,080 Speaker 1: incentive to have to do that versus the very kind 389 00:20:43,080 --> 00:20:47,600 Speaker 1: of visceral incentive of you know, making money, having you know, 390 00:20:47,600 --> 00:20:49,840 Speaker 1: outperforming in the market in the in the short term. 391 00:20:50,160 --> 00:20:53,000 Speaker 1: So so research in the quant world, for the most part, 392 00:20:53,080 --> 00:20:56,200 Speaker 1: tends to be much more implied. I have a sort 393 00:20:56,240 --> 00:21:00,160 Speaker 1: of related question, but why is why is quant investing 394 00:21:00,240 --> 00:21:04,480 Speaker 1: or why are quants so um secretive about everything? Or 395 00:21:04,880 --> 00:21:06,439 Speaker 1: I mean I don't want to call them weird, but 396 00:21:06,520 --> 00:21:11,160 Speaker 1: there is this sort of like odd culture around quant investing. 397 00:21:11,400 --> 00:21:14,560 Speaker 1: And you think of places like Renaissance and Citadel, they're 398 00:21:14,600 --> 00:21:18,480 Speaker 1: all sort of shrouded in mystique. I once heard that 399 00:21:18,520 --> 00:21:22,520 Speaker 1: Citadel had an original Enigma machine from World War Two 400 00:21:22,600 --> 00:21:24,800 Speaker 1: in one of its offices. I don't know if that's true, 401 00:21:24,840 --> 00:21:26,840 Speaker 1: but just the fact that people are saying this kind 402 00:21:26,840 --> 00:21:31,359 Speaker 1: of thing tells you something about how they regard these big, 403 00:21:31,400 --> 00:21:37,280 Speaker 1: storied quant companies. Why is there this very specific culture, 404 00:21:37,440 --> 00:21:44,040 Speaker 1: mysterious secretive culture. So I think, broadly speaking, um, people 405 00:21:44,280 --> 00:21:47,800 Speaker 1: in the by side teople in the Hedgeman industry are 406 00:21:47,640 --> 00:21:51,640 Speaker 1: are generally secretive, but I think the with with regards 407 00:21:51,680 --> 00:21:54,320 Speaker 1: to sort of their internal I P and and processes. 408 00:21:54,600 --> 00:21:56,760 Speaker 1: But I think the nature of I P in the 409 00:21:56,880 --> 00:22:00,800 Speaker 1: quant space creates incentives for people to be more secretive. Right. 410 00:22:00,880 --> 00:22:03,919 Speaker 1: So again, just you know, pulling our hypothetical kind of 411 00:22:03,960 --> 00:22:09,520 Speaker 1: Bill Ackman example, if he identifies some asset that's undervalued, UM, 412 00:22:09,800 --> 00:22:12,560 Speaker 1: he's going to be sort of very quiet about it 413 00:22:12,680 --> 00:22:16,840 Speaker 1: until he goes in and accumulates the position you watch, 414 00:22:16,920 --> 00:22:18,640 Speaker 1: because he doesn't want other people to know and other 415 00:22:18,680 --> 00:22:20,840 Speaker 1: people to front run him and to sort of take 416 00:22:20,880 --> 00:22:23,520 Speaker 1: that opportunity away. Now, once he's a mass, that position 417 00:22:23,560 --> 00:22:26,760 Speaker 1: perhaps will actually start even advertising it, right because now 418 00:22:26,800 --> 00:22:29,320 Speaker 1: if sort of people sort of follow him, works to 419 00:22:29,400 --> 00:22:32,040 Speaker 1: his benefit and it will push prices in the way 420 00:22:32,040 --> 00:22:34,600 Speaker 1: that he wants. The quant space doesn't quite work like that. 421 00:22:34,640 --> 00:22:37,160 Speaker 1: Like again, any individual trade is a very short rise 422 00:22:37,200 --> 00:22:39,840 Speaker 1: and maybe a couple of weeks. Right. Trades are sort 423 00:22:39,840 --> 00:22:44,359 Speaker 1: of very small and diffused across many many assets. But 424 00:22:44,520 --> 00:22:47,760 Speaker 1: the idea of the trade, the data source coupled with 425 00:22:48,240 --> 00:22:50,840 Speaker 1: whatever is generating the signal and the uilarity methodology and 426 00:22:50,880 --> 00:22:53,480 Speaker 1: so on, that has lasting value that might you know, 427 00:22:53,560 --> 00:22:56,400 Speaker 1: work for for the next six years. Again, year on year. 428 00:22:56,520 --> 00:23:00,840 Speaker 1: It will the performance goes down as anomalies disappear, but uh, 429 00:23:01,080 --> 00:23:03,680 Speaker 1: you know, it has multiple years of value. So uh. 430 00:23:04,000 --> 00:23:06,680 Speaker 1: The general feeling is if people sort of figure out 431 00:23:06,720 --> 00:23:09,280 Speaker 1: what you're doing, and um, where the opportunities are, and 432 00:23:09,320 --> 00:23:11,600 Speaker 1: what data sets you're doing and so on, they will 433 00:23:11,640 --> 00:23:13,520 Speaker 1: also do a similar kind of thing. That will they 434 00:23:13,520 --> 00:23:17,119 Speaker 1: will copy you and then those anomalies will disappear faster, 435 00:23:17,600 --> 00:23:19,520 Speaker 1: you know, at least in my experience, because of the 436 00:23:20,359 --> 00:23:24,400 Speaker 1: longer time horizons over which this um, this IP decays 437 00:23:24,720 --> 00:23:29,760 Speaker 1: people are more paranoid about being extremely secretive. And that's 438 00:23:29,800 --> 00:23:33,680 Speaker 1: not only for for outsiders, but that's even within firms. 439 00:23:33,720 --> 00:23:37,000 Speaker 1: So so many firms are siloed down to the level 440 00:23:37,119 --> 00:23:41,240 Speaker 1: of individual quant researchers, where um you maybe um uh 441 00:23:41,280 --> 00:23:43,320 Speaker 1: you know, you may have a team of a couple 442 00:23:43,400 --> 00:23:46,600 Speaker 1: dozen people all um uh let's say under a single PM, 443 00:23:46,640 --> 00:23:50,960 Speaker 1: all working on the same overall strategy, but you won't 444 00:23:50,960 --> 00:23:53,080 Speaker 1: know what the guy next to you is working on, right, 445 00:23:53,080 --> 00:23:55,800 Speaker 1: And if you pass data sets across maybe you um 446 00:23:56,200 --> 00:23:58,199 Speaker 1: label them in sort of random ways and so on. 447 00:23:58,240 --> 00:24:01,320 Speaker 1: So no, nobody, nobody, sort of maybe has the full 448 00:24:01,359 --> 00:24:05,280 Speaker 1: picture except a handful of people um on the top. 449 00:24:05,480 --> 00:24:07,720 Speaker 1: And again the idea there is that you know, over 450 00:24:07,800 --> 00:24:11,359 Speaker 1: time people quit or leave or whatever, um you want, 451 00:24:11,800 --> 00:24:13,639 Speaker 1: the firms would like them to have as little of 452 00:24:13,680 --> 00:24:16,679 Speaker 1: the I P as possible in terms of uh, you know, 453 00:24:16,720 --> 00:24:18,639 Speaker 1: not decaying the value of their own IP. Now, I 454 00:24:18,640 --> 00:24:21,760 Speaker 1: think famously renaissance does not operate this way, So ret 455 00:24:21,760 --> 00:24:25,600 Speaker 1: renaissance is um. One example I've heard where a firm 456 00:24:25,640 --> 00:24:27,760 Speaker 1: which is uh I think, very very difficult to get 457 00:24:27,760 --> 00:24:30,000 Speaker 1: into in terms of being hired. But but once you're 458 00:24:30,000 --> 00:24:33,040 Speaker 1: in there. They're quite open in terms of what are 459 00:24:33,040 --> 00:24:34,600 Speaker 1: the different things we've tried, Where are the things that 460 00:24:34,640 --> 00:24:36,520 Speaker 1: are working now, where the things that haven't worked before, 461 00:24:36,840 --> 00:24:38,119 Speaker 1: and you know, so on and so forth. And I 462 00:24:38,119 --> 00:24:40,440 Speaker 1: think actually from the perspective of research, that works much 463 00:24:40,480 --> 00:24:43,840 Speaker 1: better um quant researchers tend to, uh believe it or not, 464 00:24:43,920 --> 00:24:46,240 Speaker 1: tend to be kind of social animals, and it's it's 465 00:24:46,240 --> 00:24:48,160 Speaker 1: always more fun to work on things with other people 466 00:24:48,240 --> 00:24:50,200 Speaker 1: rather than just sort of sit at your desk with 467 00:24:50,480 --> 00:24:52,600 Speaker 1: with the with the blinders on and so on. You know. 468 00:24:52,920 --> 00:24:56,120 Speaker 1: Interesting about Renaissance is how they've been able to manage 469 00:24:56,119 --> 00:24:58,800 Speaker 1: it so that very very few people have have have 470 00:24:58,960 --> 00:25:01,919 Speaker 1: left and it seems like, you know, they have not 471 00:25:02,000 --> 00:25:04,920 Speaker 1: had the kind of ip loss that other people worry about. 472 00:25:06,240 --> 00:25:11,800 Speaker 1: So Renaissance famously just puts up extraordinary numbers year after 473 00:25:11,960 --> 00:25:15,679 Speaker 1: year after year. And the sort of the trick or 474 00:25:15,720 --> 00:25:19,119 Speaker 1: one trick besides there being a bunch of mathematical geniuses, 475 00:25:19,359 --> 00:25:24,000 Speaker 1: is a having this sort of open culture of collaboration 476 00:25:24,680 --> 00:25:29,000 Speaker 1: and research and be somehow preventing a lot of exodus 477 00:25:29,040 --> 00:25:30,679 Speaker 1: so that no one else has really been able to 478 00:25:31,000 --> 00:25:35,879 Speaker 1: replicate their approaches. Uh In any way, how hard is this? 479 00:25:35,920 --> 00:25:38,159 Speaker 1: So you think about like someone like I don't know, 480 00:25:38,200 --> 00:25:41,439 Speaker 1: like you hear about other other managers, like you know, 481 00:25:41,480 --> 00:25:44,280 Speaker 1: Steve Cohen is like, oh, I wanna allocate money to 482 00:25:44,480 --> 00:25:47,119 Speaker 1: quant How hard is it? And this is sort of 483 00:25:47,200 --> 00:25:49,280 Speaker 1: something I want to explore more now, is like, how 484 00:25:49,320 --> 00:25:52,520 Speaker 1: hard is it to sort of anti up into that 485 00:25:52,600 --> 00:25:55,520 Speaker 1: game and to sort of start being competitive if this 486 00:25:55,640 --> 00:25:59,399 Speaker 1: if you're sort of starting from zero right now, Um, 487 00:25:59,440 --> 00:26:01,919 Speaker 1: I think it's uh, it's a tough place. It's a 488 00:26:01,920 --> 00:26:05,120 Speaker 1: it's a it's a competitive game. Maybe not so much anymore. 489 00:26:05,160 --> 00:26:08,840 Speaker 1: But over the past five seven years, Um, my general 490 00:26:08,880 --> 00:26:13,760 Speaker 1: perspective is that the buy side active managing sort of 491 00:26:13,760 --> 00:26:16,280 Speaker 1: hedge funds have been shrinking overall. The one sector that 492 00:26:16,280 --> 00:26:19,160 Speaker 1: has not been shrinking his quant and so I think 493 00:26:19,200 --> 00:26:22,280 Speaker 1: there has been an entrance of uh kind of new 494 00:26:22,320 --> 00:26:25,280 Speaker 1: players there. Um. Now, I'm Steve Cohen you specifically mentioned. 495 00:26:25,320 --> 00:26:27,080 Speaker 1: He's actually been at it for a while. He's been 496 00:26:27,160 --> 00:26:31,720 Speaker 1: in the quant space for the since the early two thousand's. 497 00:26:31,720 --> 00:26:34,119 Speaker 1: He on the order of twenty to thirty percent of 498 00:26:34,200 --> 00:26:36,639 Speaker 1: his assets are actually quant some some something like that, 499 00:26:36,680 --> 00:26:38,719 Speaker 1: like a non you know, people mainly think of him 500 00:26:38,720 --> 00:26:41,600 Speaker 1: as a long short uh kind of guy, and that's 501 00:26:41,600 --> 00:26:43,720 Speaker 1: probably mainly what he is. But again, um, you know, 502 00:26:43,760 --> 00:26:45,920 Speaker 1: maybe a third of his assets are in a quant 503 00:26:45,920 --> 00:26:48,680 Speaker 1: space through after Cubist and so on. Now he operates 504 00:26:48,760 --> 00:26:52,280 Speaker 1: very differently, um, he operates. His quant funds operate in 505 00:26:52,600 --> 00:26:55,919 Speaker 1: uh kind of like traditional UM long short guys operate, 506 00:26:56,119 --> 00:27:00,000 Speaker 1: wherein you hire individual pms, you watch them a very 507 00:27:00,200 --> 00:27:02,320 Speaker 1: kind of carefully. They make money or they lose money 508 00:27:02,359 --> 00:27:04,520 Speaker 1: if if they if they're not making money quickly enough, 509 00:27:04,560 --> 00:27:07,640 Speaker 1: you fire them. And uh you sort of you kind 510 00:27:07,640 --> 00:27:10,600 Speaker 1: of have a portfolio of these these individual managers who 511 00:27:10,640 --> 00:27:13,040 Speaker 1: are who are doing their own things, who are tightly siloed, 512 00:27:13,440 --> 00:27:16,760 Speaker 1: and uh you know, uh you try to manage that. 513 00:27:17,119 --> 00:27:20,200 Speaker 1: And that's the way his quant operation manages. So there's 514 00:27:20,400 --> 00:27:23,120 Speaker 1: you know, again, um, a whole bunch of small um 515 00:27:23,280 --> 00:27:25,800 Speaker 1: let's call them pods or whatever, of you know, two 516 00:27:25,880 --> 00:27:27,920 Speaker 1: or three people each kind of doing their own thing 517 00:27:28,200 --> 00:27:31,240 Speaker 1: in an uncoordinated way. You know. That's again quite a 518 00:27:31,280 --> 00:27:34,120 Speaker 1: different model than let's say, the Renaissance, which is uh 519 00:27:34,560 --> 00:27:37,320 Speaker 1: um uh you know, one kind of open strategy. And 520 00:27:37,359 --> 00:27:39,800 Speaker 1: I think the the advantage of the Steve Cohen model 521 00:27:40,000 --> 00:27:43,160 Speaker 1: is that uh Um, you know, uh, it's it's easy 522 00:27:43,240 --> 00:27:46,000 Speaker 1: to to hire people from the HR. Process is very easy. 523 00:27:46,040 --> 00:27:47,560 Speaker 1: You don't have to care when people come and go 524 00:27:47,640 --> 00:27:49,520 Speaker 1: on so on, because you're not really investing in any 525 00:27:49,600 --> 00:27:51,920 Speaker 1: of their their individual I P. Right when when someone 526 00:27:52,000 --> 00:27:54,280 Speaker 1: leaves or like let's you fire someone, it's because they 527 00:27:54,320 --> 00:27:56,679 Speaker 1: didn't do well and whatever they have is maybe not 528 00:27:56,760 --> 00:27:59,120 Speaker 1: worth um that much and they don't know anything else 529 00:27:59,160 --> 00:28:01,239 Speaker 1: about what your other p are doing and so so 530 00:28:01,280 --> 00:28:04,080 Speaker 1: that process is very easy. But I think the downside 531 00:28:04,359 --> 00:28:07,639 Speaker 1: is that what we're sort of starting to see is 532 00:28:07,680 --> 00:28:11,520 Speaker 1: throughout the quant space, like you know, the broader technology industry, 533 00:28:11,560 --> 00:28:12,840 Speaker 1: we're starting to see that there are a lot of 534 00:28:13,080 --> 00:28:16,679 Speaker 1: increasing returns to scale that as you get bigger and 535 00:28:16,760 --> 00:28:20,760 Speaker 1: bigger firms are able to build advantages. And one kind 536 00:28:20,800 --> 00:28:24,880 Speaker 1: of concrete source of this is around trading costs. Right, Um, 537 00:28:25,119 --> 00:28:27,400 Speaker 1: when you're thinking about, like let's say, on an individual 538 00:28:27,400 --> 00:28:29,400 Speaker 1: trade by trade basis, do I want to get into 539 00:28:29,480 --> 00:28:32,199 Speaker 1: this trade? You have a prediction of how much you're 540 00:28:32,200 --> 00:28:34,359 Speaker 1: going to make if your if your models are correct, 541 00:28:34,560 --> 00:28:36,600 Speaker 1: but also there are these costs that you're paying, these 542 00:28:37,000 --> 00:28:40,400 Speaker 1: transaction costs, and if your prediction doesn't exceed your costs. 543 00:28:40,400 --> 00:28:42,160 Speaker 1: You shouldn't put on that trade because even in the 544 00:28:42,160 --> 00:28:45,600 Speaker 1: best case, you're you're not going to make your money, right. So, um, 545 00:28:45,920 --> 00:28:49,480 Speaker 1: what what's happened is that as more and more people 546 00:28:49,480 --> 00:28:51,880 Speaker 1: have gotten into the quant space and more and more 547 00:28:51,880 --> 00:28:55,520 Speaker 1: of these identically anomaly sorry are identified and markets get 548 00:28:55,560 --> 00:28:59,040 Speaker 1: the more efficient, the signals have gotten weaker, right, and 549 00:28:59,160 --> 00:29:01,280 Speaker 1: so um, just to sort of give a give a 550 00:29:01,320 --> 00:29:04,000 Speaker 1: maybe a concrete example. One signal that's sort of um, 551 00:29:04,240 --> 00:29:08,240 Speaker 1: quite well known throughout the klant industry and un academics 552 00:29:08,280 --> 00:29:10,480 Speaker 1: of published papers and so on, its order book of balance. Right, 553 00:29:10,800 --> 00:29:12,640 Speaker 1: if you go out and you look at an electronic 554 00:29:12,720 --> 00:29:15,240 Speaker 1: order book and they're more buyers than there are sellers 555 00:29:15,240 --> 00:29:17,960 Speaker 1: in terms of the resting limit orders. Um, it's it's 556 00:29:17,960 --> 00:29:20,239 Speaker 1: more likely that the press will go up and go down. Right. 557 00:29:20,240 --> 00:29:21,480 Speaker 1: You can. You can go out and try that that 558 00:29:21,520 --> 00:29:24,760 Speaker 1: has a predictive value. Now, however, if that's all you know, 559 00:29:25,080 --> 00:29:27,480 Speaker 1: you won't make money because you might think the price 560 00:29:27,520 --> 00:29:29,080 Speaker 1: is going to go up you know, a tenth of 561 00:29:29,080 --> 00:29:30,959 Speaker 1: a basis point just to throughout a number, but your 562 00:29:30,960 --> 00:29:33,720 Speaker 1: transaction costs or two basis points, and you know you're 563 00:29:33,800 --> 00:29:37,560 Speaker 1: just not. Um, you can't exceed your your costs. So, um, 564 00:29:37,600 --> 00:29:40,560 Speaker 1: the transaction costs to a first approximation, they're they're kind 565 00:29:40,600 --> 00:29:42,760 Speaker 1: of like on a trade by trade basis, a fixed 566 00:29:42,800 --> 00:29:45,320 Speaker 1: costs that you have to exceed. Now, if you're in 567 00:29:45,320 --> 00:29:48,040 Speaker 1: a world where you have many, many signals, maybe tens, 568 00:29:48,040 --> 00:29:50,880 Speaker 1: maybe hundreds, maybe thousands, and you're adding them up and 569 00:29:50,920 --> 00:29:54,640 Speaker 1: they're independent and you trade when they're all aligned, now 570 00:29:54,720 --> 00:29:57,120 Speaker 1: you can have sort of, um, you know, signals that 571 00:29:57,160 --> 00:30:00,600 Speaker 1: are weak individually, and nevertheless, when you you combine them, 572 00:30:00,640 --> 00:30:03,480 Speaker 1: when you aggregate them, you are able to exceed transaction 573 00:30:03,520 --> 00:30:06,760 Speaker 1: costs and monetize them. So that that order and balanced 574 00:30:06,760 --> 00:30:09,720 Speaker 1: signal that I just sort of talked about. If you're 575 00:30:09,800 --> 00:30:11,440 Speaker 1: sort of one guy in your basement and that's all 576 00:30:11,480 --> 00:30:13,200 Speaker 1: you knew, you can't make money off that. But if 577 00:30:13,200 --> 00:30:15,480 Speaker 1: you have twenty other signals and you're you know, you're 578 00:30:15,560 --> 00:30:18,040 Speaker 1: going to put on a trade anyway, in some sense, 579 00:30:18,080 --> 00:30:21,400 Speaker 1: the transaction costs become a sunk cost, and and that 580 00:30:21,600 --> 00:30:23,400 Speaker 1: you know, point one basis point that you're going to 581 00:30:23,480 --> 00:30:26,120 Speaker 1: get because of this well known signal, that becomes free money. 582 00:30:26,400 --> 00:30:28,640 Speaker 1: So so as you get that kind of economies of 583 00:30:28,640 --> 00:30:31,680 Speaker 1: scale because of fixed costs. I think it becomes harder 584 00:30:31,720 --> 00:30:35,280 Speaker 1: and harder to have UM quant strategies where um, you 585 00:30:35,320 --> 00:30:37,240 Speaker 1: don't have a lot of people UM you know, in 586 00:30:37,280 --> 00:30:39,600 Speaker 1: a very kind of coordinated research process where you have 587 00:30:39,720 --> 00:30:44,080 Speaker 1: people working essentially independently UM the kinds of UM you 588 00:30:44,120 --> 00:30:46,960 Speaker 1: know places that are structured like like like let's say 589 00:30:47,040 --> 00:30:51,200 Speaker 1: Renaissance again, where you might have like two plant researchers 590 00:30:51,400 --> 00:30:54,480 Speaker 1: all working on different aspects of the thing, and then 591 00:30:54,560 --> 00:30:57,000 Speaker 1: you know, these things combined to one sort of overall 592 00:30:57,080 --> 00:30:59,640 Speaker 1: view of the market. I think that is able to 593 00:30:59,720 --> 00:31:03,000 Speaker 1: sort of better monetize a lot of these signals in 594 00:31:03,000 --> 00:31:06,680 Speaker 1: this kind of more competitive world. So on that note, 595 00:31:06,800 --> 00:31:09,160 Speaker 1: if if you are running a lot of these strategies, 596 00:31:09,200 --> 00:31:11,800 Speaker 1: getting a lot of these signals, and you're able to 597 00:31:11,960 --> 00:31:15,400 Speaker 1: lower your transaction costs because of that scale, and at 598 00:31:15,440 --> 00:31:19,160 Speaker 1: the same time, quant investing has these big barriers to 599 00:31:19,320 --> 00:31:22,440 Speaker 1: entry because you have to have these technological outlays, you 600 00:31:22,480 --> 00:31:25,240 Speaker 1: have to hire a bunch of PhDs and things like that. 601 00:31:25,880 --> 00:31:29,880 Speaker 1: Does that mean that the industry is inevitably sort of 602 00:31:29,920 --> 00:31:33,600 Speaker 1: trending towards a monopoly? Are are we going to get 603 00:31:33,600 --> 00:31:37,240 Speaker 1: a situation where there is just one or maybe two 604 00:31:37,440 --> 00:31:40,600 Speaker 1: or three really big quant investors because no one else 605 00:31:40,640 --> 00:31:44,160 Speaker 1: can compete with them effectively. I think we're kind of there. 606 00:31:44,200 --> 00:31:46,160 Speaker 1: I mean, I think there are only a handful of 607 00:31:46,240 --> 00:31:48,320 Speaker 1: large quants. Most of them have been doing it for 608 00:31:48,360 --> 00:31:52,360 Speaker 1: a long time. I mean, Renaissance, d SHAW, PDT, two Sigma. 609 00:31:52,640 --> 00:31:55,000 Speaker 1: You know, there's there, there's a there's a handful of others. 610 00:31:55,360 --> 00:31:58,040 Speaker 1: I think it's it's harder to see, you know, maybe 611 00:31:58,080 --> 00:32:00,800 Speaker 1: there's some exceptions, you know, in terms of funds that 612 00:32:00,840 --> 00:32:04,400 Speaker 1: have launched UHM more more recently, but it's difficult to 613 00:32:04,440 --> 00:32:08,080 Speaker 1: see people of of that scale with with the similar 614 00:32:08,120 --> 00:32:11,360 Speaker 1: track records. So I think we are seeing some degree 615 00:32:11,440 --> 00:32:14,479 Speaker 1: of consolidation. I don't know what the altar I mean, 616 00:32:14,520 --> 00:32:16,640 Speaker 1: I don't know if it's gonna um come down to 617 00:32:16,720 --> 00:32:19,200 Speaker 1: one firm. I think, you know, probably not, just probably 618 00:32:19,680 --> 00:32:21,720 Speaker 1: kind of more competition, but I think it will be 619 00:32:21,800 --> 00:32:25,560 Speaker 1: harder to have uh sort of either either more independent 620 00:32:25,600 --> 00:32:28,800 Speaker 1: managers or like UM kind of the siloed model of 621 00:32:28,880 --> 00:32:47,680 Speaker 1: places like uh you know, UM S, A C and Millennium. 622 00:32:47,720 --> 00:32:50,320 Speaker 1: If I want to start a quant fund, what are 623 00:32:50,360 --> 00:32:53,400 Speaker 1: we talking about in terms of how much it's just 624 00:32:53,440 --> 00:32:56,560 Speaker 1: gonna cost for computers and data just to give even 625 00:32:56,600 --> 00:32:58,560 Speaker 1: get in the game. Don't do it, Joe. I feel 626 00:32:58,600 --> 00:33:01,680 Speaker 1: like this whole conversation is how you shouldn't be doing that. No, 627 00:33:01,760 --> 00:33:04,320 Speaker 1: I realized, I realized that it's a bad idea. But 628 00:33:04,400 --> 00:33:06,360 Speaker 1: let's say I'm an idiot and I try anyway, Like, 629 00:33:06,400 --> 00:33:10,120 Speaker 1: what are we talking about? So? Um, I think things 630 00:33:10,120 --> 00:33:14,120 Speaker 1: have gotten over time much more expensive, things like data 631 00:33:14,160 --> 00:33:17,520 Speaker 1: feeds and uh you know, so on the exchanges have 632 00:33:17,600 --> 00:33:21,320 Speaker 1: constantly been ramping the prices on on these things. But 633 00:33:21,520 --> 00:33:24,640 Speaker 1: um you know, these days what's become one of the 634 00:33:24,640 --> 00:33:28,120 Speaker 1: biggest costs is actually just pure computation and and this 635 00:33:28,240 --> 00:33:30,280 Speaker 1: is also a trend we see um uh, you know, 636 00:33:30,320 --> 00:33:33,640 Speaker 1: more broadly in a technology, you know, if you look 637 00:33:33,680 --> 00:33:36,720 Speaker 1: at kind of the state of the art models for 638 00:33:37,040 --> 00:33:42,360 Speaker 1: things like um uh, computer vision, object recognition for um uh, 639 00:33:42,520 --> 00:33:44,800 Speaker 1: you know, playing games like a chess and go and 640 00:33:44,840 --> 00:33:48,840 Speaker 1: so on, these types of models leverage approaches and machine 641 00:33:48,960 --> 00:33:51,720 Speaker 1: learning that are really based on having a lot of 642 00:33:51,800 --> 00:33:54,239 Speaker 1: data and doing even more than that, doing a lot 643 00:33:54,280 --> 00:33:57,400 Speaker 1: of computation and and and so the spirit there, you know, 644 00:33:57,440 --> 00:34:00,080 Speaker 1: coming out of places like deep minded Google or be 645 00:34:00,200 --> 00:34:03,680 Speaker 1: ai and stuff, um open ai, um of you know 646 00:34:04,200 --> 00:34:07,800 Speaker 1: are artificial intelligence UM company that their main model is 647 00:34:07,800 --> 00:34:09,920 Speaker 1: is literally like we're going to do simple things, but 648 00:34:09,960 --> 00:34:12,920 Speaker 1: we're going to leverage it to massive scale computation, right, 649 00:34:13,040 --> 00:34:15,080 Speaker 1: and so so I think you're starting to see that 650 00:34:15,120 --> 00:34:19,160 Speaker 1: in finance as well, where you need to do things 651 00:34:19,200 --> 00:34:22,160 Speaker 1: like let's say you need to um UM back test 652 00:34:22,560 --> 00:34:25,480 Speaker 1: a trading strategy. UM, but you have some parameters, and 653 00:34:25,520 --> 00:34:27,560 Speaker 1: you want to try tens of thousands of combinations of 654 00:34:27,600 --> 00:34:30,719 Speaker 1: those trading parameters, and each one involves a simulation over 655 00:34:30,840 --> 00:34:33,000 Speaker 1: you know, twenty years and so on and so forth. 656 00:34:33,360 --> 00:34:36,600 Speaker 1: You need a lot of computers. So UM. Someone told 657 00:34:36,600 --> 00:34:41,040 Speaker 1: me anecdotally that at a major quant shop, each quantitative 658 00:34:41,280 --> 00:34:44,719 Speaker 1: researcher has given kind of a quote unquote budget of 659 00:34:44,520 --> 00:34:47,759 Speaker 1: of of ten thousand CPUs, right, so I ain't given time, 660 00:34:47,800 --> 00:34:50,319 Speaker 1: they can use up to ten thousand individual kind of 661 00:34:50,880 --> 00:34:53,120 Speaker 1: processing units. And just to give you a sense of 662 00:34:53,160 --> 00:34:55,000 Speaker 1: what that costs, UM, you know, if you're to go, 663 00:34:55,400 --> 00:34:58,759 Speaker 1: you know, buy that on Amazon at AWS, that would 664 00:34:58,760 --> 00:35:02,239 Speaker 1: be the order of magnitude maybe a million dollars a year. Right, 665 00:35:02,280 --> 00:35:04,240 Speaker 1: And this is just for this is just for research. 666 00:35:04,280 --> 00:35:06,120 Speaker 1: This is not to actually generate the trades or whatever. 667 00:35:06,160 --> 00:35:08,440 Speaker 1: This is just a tune all the parameters and and 668 00:35:08,640 --> 00:35:12,280 Speaker 1: and sort of really optimize your performance. That's really interesting. 669 00:35:12,320 --> 00:35:16,120 Speaker 1: It kind of makes me wonder how how good I 670 00:35:16,160 --> 00:35:20,279 Speaker 1: guess academic research is at gauging quant strategies if the 671 00:35:20,320 --> 00:35:24,160 Speaker 1: outlays just to run a few experiments are so massive. 672 00:35:24,400 --> 00:35:26,880 Speaker 1: But on a slightly different topic, I wanted to ask you, 673 00:35:26,920 --> 00:35:30,239 Speaker 1: I guess this question is kind of inevitable. Whenever you 674 00:35:30,320 --> 00:35:35,640 Speaker 1: talk about algorithmic trading or systematic trading, what value do 675 00:35:35,680 --> 00:35:41,040 Speaker 1: you think quant investing actually creates for society? So, for instance, 676 00:35:41,080 --> 00:35:45,000 Speaker 1: when we talk about traditional investing, that's supposed to channel 677 00:35:45,040 --> 00:35:48,839 Speaker 1: capital in the most efficient way possible to good companies, 678 00:35:48,880 --> 00:35:52,080 Speaker 1: and that should in theory benefit the entire economy. But 679 00:35:52,200 --> 00:35:55,239 Speaker 1: quant investing, as we've discussed, isn't really about that. It's 680 00:35:55,239 --> 00:35:59,520 Speaker 1: about arbitraging these small differences. So maybe it makes prices 681 00:35:59,560 --> 00:36:05,000 Speaker 1: slightly more efficient, but is that worth the enormous infrastructure 682 00:36:05,120 --> 00:36:10,040 Speaker 1: investment that we've been discussing being spent on it? So 683 00:36:10,239 --> 00:36:13,840 Speaker 1: I think there is, um there are some benefits. You know, 684 00:36:14,080 --> 00:36:17,439 Speaker 1: it varies based on the strategy and based on really 685 00:36:17,480 --> 00:36:20,200 Speaker 1: the incident time. But I think a lot of uh, 686 00:36:20,480 --> 00:36:22,759 Speaker 1: you know, to to a first approximation. If you see 687 00:36:22,800 --> 00:36:25,719 Speaker 1: a price move in a direction that's unusual, UM, it 688 00:36:25,760 --> 00:36:28,840 Speaker 1: could continue or it could revert. Right to the extent 689 00:36:28,880 --> 00:36:30,560 Speaker 1: that you think it's going to revert, you're going to 690 00:36:30,640 --> 00:36:33,400 Speaker 1: sort of bet against it. And what what that amounts 691 00:36:33,440 --> 00:36:37,400 Speaker 1: to is basically supplying temporary liquidity to the market. Right, So, 692 00:36:37,440 --> 00:36:42,400 Speaker 1: I think the positive aspect to UM quantitative investing is 693 00:36:42,480 --> 00:36:44,800 Speaker 1: that UM I think a lot of it is supplying 694 00:36:44,800 --> 00:36:48,480 Speaker 1: liquidity to the market on a horizon of uh, let's 695 00:36:48,520 --> 00:36:51,880 Speaker 1: say days to two weeks right now. The flip side 696 00:36:51,960 --> 00:36:54,960 Speaker 1: is if you're if you're really it's more of a 697 00:36:55,000 --> 00:36:59,160 Speaker 1: momentum that you might be accelerating the trends UM you're 698 00:36:59,280 --> 00:37:01,880 Speaker 1: taking away like wuity, you're competing for that liquidity, but 699 00:37:02,000 --> 00:37:05,640 Speaker 1: as you said, maybe you're making prices more more efficient. 700 00:37:06,000 --> 00:37:08,760 Speaker 1: So I think, on balance, I think that that probably 701 00:37:08,760 --> 00:37:12,759 Speaker 1: there is some benefit. I think it's probably small. Admittedly, 702 00:37:13,360 --> 00:37:16,040 Speaker 1: is it worth all these uh, you know, very smart 703 00:37:16,080 --> 00:37:18,160 Speaker 1: people being drawn away from other fields and so on, 704 00:37:18,800 --> 00:37:21,400 Speaker 1: I'm not sure, But you know, probably as much or 705 00:37:21,440 --> 00:37:24,160 Speaker 1: more resources or spent at places like Facebook and Google 706 00:37:24,160 --> 00:37:26,640 Speaker 1: getting people to click on ads. Right, I'm not sure 707 00:37:26,640 --> 00:37:30,359 Speaker 1: that that's uh as positive the pressing. Think about all 708 00:37:30,360 --> 00:37:34,960 Speaker 1: these people, um, you know, looking for signals to squeeze 709 00:37:35,000 --> 00:37:37,480 Speaker 1: out three basis points in the market, because there could 710 00:37:37,520 --> 00:37:40,560 Speaker 1: be some great innovations in squeezing more ads onto a 711 00:37:40,600 --> 00:37:42,719 Speaker 1: mobile phone that they'd be working on. And there you go, 712 00:37:43,520 --> 00:37:46,600 Speaker 1: kind of a sad allocation of resources. See you think 713 00:37:46,640 --> 00:37:52,360 Speaker 1: Joe's joking, but he's he probably doesn't. So uh, here's 714 00:37:52,360 --> 00:37:55,040 Speaker 1: one thing that also always tends to come up. It's 715 00:37:55,040 --> 00:37:59,279 Speaker 1: this idea of um, this type of trading reaching the 716 00:37:59,440 --> 00:38:04,359 Speaker 1: limits of available technology and pushing the strategies to sort 717 00:38:04,400 --> 00:38:09,960 Speaker 1: of greater extremes. But those extremes eventually have limits. And 718 00:38:10,280 --> 00:38:13,440 Speaker 1: so I guess I'm just wondering, is there a limit 719 00:38:13,920 --> 00:38:17,880 Speaker 1: to quant investing? Is there a point at which quants 720 00:38:17,920 --> 00:38:21,960 Speaker 1: sort of arbitrage everything out of the market and the 721 00:38:22,040 --> 00:38:25,960 Speaker 1: signals are no longer useful or the algorithms themselves are 722 00:38:26,000 --> 00:38:29,479 Speaker 1: impacting the market in some way? And on that note, 723 00:38:29,600 --> 00:38:33,960 Speaker 1: what's what's the next big thing in quant investing? I 724 00:38:34,000 --> 00:38:38,719 Speaker 1: guess yeah, So, I mean I think there's a constant balance. 725 00:38:39,000 --> 00:38:42,400 Speaker 1: These are finishing of inefficiencies are being identified in arbitrage 726 00:38:42,440 --> 00:38:45,920 Speaker 1: the way because there's money in it, right, and so 727 00:38:45,960 --> 00:38:49,279 Speaker 1: as arbitraged in it, Um, the money sort of disappears 728 00:38:49,680 --> 00:38:52,279 Speaker 1: and then you get sort of a fewer people kind 729 00:38:52,280 --> 00:38:56,279 Speaker 1: of doing it. But um, so long as there's uh, 730 00:38:56,760 --> 00:39:00,560 Speaker 1: you know, kind of traders out there, we're not paying 731 00:39:00,560 --> 00:39:03,759 Speaker 1: attention to this stuff and uh you know, the Robin 732 00:39:03,800 --> 00:39:06,760 Speaker 1: Hood traders or whatever and are kind of leaving money 733 00:39:06,760 --> 00:39:09,640 Speaker 1: on the table, Um, there will be people um there 734 00:39:09,640 --> 00:39:12,880 Speaker 1: who are trying to uh sweep up the crumbs in 735 00:39:12,960 --> 00:39:14,960 Speaker 1: terms of where it's going. What the what the next 736 00:39:15,640 --> 00:39:18,200 Speaker 1: big thing is. I think it's it's it's pretty hard 737 00:39:18,239 --> 00:39:21,360 Speaker 1: to predict, but I think um, uh you know, broadly 738 00:39:21,400 --> 00:39:25,280 Speaker 1: a shift towards uh things that are even more black box, 739 00:39:25,440 --> 00:39:29,719 Speaker 1: even more computationally driven and uh um not so much. 740 00:39:29,880 --> 00:39:32,759 Speaker 1: Uh you know, have like kind of nice structural explanations. 741 00:39:33,200 --> 00:39:35,759 Speaker 1: Um again sort of following a lot of what's going 742 00:39:35,800 --> 00:39:38,360 Speaker 1: on in in the tech world as we shift to 743 00:39:38,480 --> 00:39:42,200 Speaker 1: ideas like deep neural networks and reinforcement learning and so 744 00:39:42,239 --> 00:39:44,280 Speaker 1: on and so forth. Um. You know, you know, again, 745 00:39:44,320 --> 00:39:46,680 Speaker 1: you have these these systems that work worked great for 746 00:39:46,840 --> 00:39:49,319 Speaker 1: Let's say I'm playing go, but it's really hard to 747 00:39:49,320 --> 00:39:52,080 Speaker 1: explain what's going on and I think we're starting to 748 00:39:52,120 --> 00:39:55,400 Speaker 1: see that in the quant world as well, again leveraging 749 00:39:55,440 --> 00:39:58,439 Speaker 1: a computation but really really ending up with with things 750 00:39:58,480 --> 00:40:01,160 Speaker 1: that are you know, um, you know black boxes that 751 00:40:01,640 --> 00:40:05,360 Speaker 1: you know just are completely not transparent. So in other words, 752 00:40:05,920 --> 00:40:08,280 Speaker 1: you know, like you could look at something like satellite 753 00:40:08,320 --> 00:40:10,600 Speaker 1: images and say, oh, there's a lot of cars parked 754 00:40:10,600 --> 00:40:13,439 Speaker 1: of Walmart and then predict the Walmart stock is going 755 00:40:13,440 --> 00:40:16,920 Speaker 1: to be up. But the next, um, the next generation 756 00:40:16,960 --> 00:40:20,160 Speaker 1: of things to watch out for is this works, and 757 00:40:20,200 --> 00:40:23,520 Speaker 1: it works consistently, but we as humans can't really articulate 758 00:40:23,560 --> 00:40:29,680 Speaker 1: why exactly. That's super interesting. Well, on that note of 759 00:40:29,960 --> 00:40:33,680 Speaker 1: humans not really even being being able to explain what 760 00:40:33,680 --> 00:40:36,359 Speaker 1: they're doing, um, it seems like a perfect place to stop. 761 00:40:36,560 --> 00:41:03,200 Speaker 1: Thank you so much for joining us. Thank you so much, Tracy. 762 00:41:03,320 --> 00:41:06,160 Speaker 1: You know, uh, as a as a media person, I 763 00:41:06,200 --> 00:41:10,240 Speaker 1: have my own experience with the sort of alpha decay 764 00:41:10,440 --> 00:41:15,280 Speaker 1: that CMX was talking about. Do you know what it is? Um, 765 00:41:15,400 --> 00:41:18,080 Speaker 1: did you build some sort of algorithm to take advantage 766 00:41:18,080 --> 00:41:21,000 Speaker 1: of like Google Ads or something and then it stopped working? No, 767 00:41:21,000 --> 00:41:24,360 Speaker 1: there's nothing so sophisticated. But back in the early days 768 00:41:24,400 --> 00:41:27,640 Speaker 1: of like blogging and stuff, I remember this phenomenon where 769 00:41:27,680 --> 00:41:30,200 Speaker 1: you would come up with some like headline construction. You'd 770 00:41:30,320 --> 00:41:34,640 Speaker 1: like five things you need to know today. Remember, like 771 00:41:34,680 --> 00:41:37,799 Speaker 1: the old upworthy headlines, they were like and you could 772 00:41:37,920 --> 00:41:40,160 Speaker 1: and you can't guess what you know. And then those 773 00:41:40,200 --> 00:41:42,879 Speaker 1: work and those generate like excess traffic, and they get 774 00:41:42,880 --> 00:41:46,520 Speaker 1: shared on Facebook, and then everybody discovers that these headlines 775 00:41:46,560 --> 00:41:49,919 Speaker 1: cliches work, and then everyone does them, and then people 776 00:41:49,960 --> 00:41:51,880 Speaker 1: stopped clicking on them, and you need to like find 777 00:41:52,000 --> 00:41:54,480 Speaker 1: I don't do clickbait anymore. But I always thought at 778 00:41:54,480 --> 00:41:56,760 Speaker 1: the time like that was like a very similar process 779 00:41:57,280 --> 00:42:00,640 Speaker 1: to uh, to this sort of quant approach to investing, 780 00:42:00,719 --> 00:42:03,600 Speaker 1: the sort of search for alpha and alpha decay of 781 00:42:03,680 --> 00:42:07,600 Speaker 1: a blog headline anymore was the key word in that 782 00:42:07,680 --> 00:42:11,360 Speaker 1: sentence about clickbait. But I think it's a really good analogy. 783 00:42:11,840 --> 00:42:14,400 Speaker 1: It is a good analogy because like the usefulness of 784 00:42:14,440 --> 00:42:17,640 Speaker 1: those headline constructions decays over time, as you point out, 785 00:42:17,640 --> 00:42:20,359 Speaker 1: because more people are copying them. But it also kind 786 00:42:20,400 --> 00:42:23,879 Speaker 1: of gets to that point about the limits of this 787 00:42:23,960 --> 00:42:26,440 Speaker 1: type of investing. There are only so many ways that 788 00:42:26,520 --> 00:42:29,960 Speaker 1: you can construct a headline, and eventually people kind of 789 00:42:30,000 --> 00:42:33,840 Speaker 1: catch on two different ones and they become not so enticing, 790 00:42:34,280 --> 00:42:36,840 Speaker 1: and I kind of wonder if the same thing could 791 00:42:36,840 --> 00:42:40,919 Speaker 1: eventually happen to quant investing. So obviously there are many 792 00:42:41,000 --> 00:42:44,799 Speaker 1: many more possibilities in quant investing, and it's possible that 793 00:42:45,080 --> 00:42:48,680 Speaker 1: markets are always changing and so opportunities for arbitrage and 794 00:42:48,760 --> 00:42:52,000 Speaker 1: identifying these signals are always coming up. But it does 795 00:42:52,080 --> 00:42:55,879 Speaker 1: make you wonder. It certainly does. And what he's talking 796 00:42:55,920 --> 00:42:57,920 Speaker 1: about at the end, where maybe the signals of the 797 00:42:57,920 --> 00:43:01,399 Speaker 1: future are just things that work it can't be articulated, 798 00:43:01,560 --> 00:43:04,640 Speaker 1: is just like a super kind of fascinating phenomenon to 799 00:43:04,760 --> 00:43:06,960 Speaker 1: just like wrap your head around. Yeah, I feel like 800 00:43:07,000 --> 00:43:09,960 Speaker 1: that's a good microcosm for maybe the human experience in 801 00:43:10,000 --> 00:43:13,120 Speaker 1: the future. Like we have the technology, we're not entirely 802 00:43:13,120 --> 00:43:15,399 Speaker 1: sure how it works, but we're just going to sort 803 00:43:15,400 --> 00:43:18,759 Speaker 1: of let it run and hope for the best. One 804 00:43:18,800 --> 00:43:21,239 Speaker 1: of the things that sort of interested me is like 805 00:43:21,280 --> 00:43:23,920 Speaker 1: a sort of thing to watch going forward, is okay, 806 00:43:23,960 --> 00:43:26,120 Speaker 1: So we talked about a huge aspect of that was 807 00:43:26,160 --> 00:43:28,319 Speaker 1: just the costs and how like you might be able 808 00:43:28,360 --> 00:43:32,239 Speaker 1: to identify a profitable anomaly. But unless the cost of 809 00:43:32,280 --> 00:43:35,600 Speaker 1: getting the data and executing the trade is lower than that, 810 00:43:35,719 --> 00:43:38,160 Speaker 1: it's um, it's useless, but you know, you also have 811 00:43:38,160 --> 00:43:41,600 Speaker 1: to wonder, like, okay, right now, like a certain handful 812 00:43:41,640 --> 00:43:45,600 Speaker 1: of exchanges, say, control a lot of the trade data costs. 813 00:43:46,120 --> 00:43:48,960 Speaker 1: In theory, that seems like an area where maybe new 814 00:43:49,040 --> 00:43:51,680 Speaker 1: entities will come and find a way to provide data 815 00:43:51,760 --> 00:43:57,200 Speaker 1: cheaper Amazon Web services. You know, presumably computation costs are 816 00:43:57,239 --> 00:44:00,040 Speaker 1: going to keep coming down, and obviously that was a 817 00:44:00,040 --> 00:44:02,200 Speaker 1: big breakthrough from probably the old days where you had 818 00:44:02,239 --> 00:44:05,759 Speaker 1: some sort of main frame on premise services. You know, 819 00:44:05,880 --> 00:44:09,000 Speaker 1: computation has gotten cheaper, so there's probably always going to 820 00:44:09,080 --> 00:44:13,480 Speaker 1: be new opportunities to squeeze out even smaller profits because 821 00:44:13,520 --> 00:44:16,399 Speaker 1: there are ways to shave costs in sort of your 822 00:44:16,800 --> 00:44:20,719 Speaker 1: in your research, your work. Yeah. Maybe the other thing 823 00:44:20,760 --> 00:44:24,400 Speaker 1: that was really interesting was the idea that quants um. 824 00:44:24,440 --> 00:44:28,200 Speaker 1: I think CMAC described them as actually social animals, which 825 00:44:28,280 --> 00:44:29,880 Speaker 1: kind of flies in the face and I think of 826 00:44:29,920 --> 00:44:33,400 Speaker 1: a lot of stereotypes. But I'm also I'm really curious. 827 00:44:33,440 --> 00:44:36,960 Speaker 1: I would love to be embedded in a firm like 828 00:44:37,000 --> 00:44:41,360 Speaker 1: Citadel and just observe how they work together and what's 829 00:44:41,400 --> 00:44:44,880 Speaker 1: considered a good out go, a good systematic strategy versus 830 00:44:44,880 --> 00:44:48,080 Speaker 1: a bad systematic strategy. Obviously you wanted to make money, 831 00:44:48,120 --> 00:44:51,560 Speaker 1: but are there certain things that are more valued over others? 832 00:44:51,680 --> 00:44:56,040 Speaker 1: Maybe cheapness to execute or um, I don't know, risk management, 833 00:44:56,120 --> 00:44:58,640 Speaker 1: something like that. I'd be so curious to see how 834 00:44:58,680 --> 00:45:02,640 Speaker 1: that all works. Um, I'm sure if we just walked 835 00:45:02,640 --> 00:45:04,200 Speaker 1: in there, just let us in the door and let 836 00:45:04,400 --> 00:45:06,240 Speaker 1: we could just hang out there for a while. Yeah, 837 00:45:06,520 --> 00:45:09,520 Speaker 1: I'm sure they wouldn't mind at all. No, let us 838 00:45:09,520 --> 00:45:13,680 Speaker 1: see their white boards stuff like that. Citadel, if you're listening, 839 00:45:14,120 --> 00:45:16,840 Speaker 1: we would like to come sit you. Okay, should we 840 00:45:16,920 --> 00:45:19,560 Speaker 1: leave it there? Let's leave it there. This has been 841 00:45:19,600 --> 00:45:23,160 Speaker 1: another episode of the ad Thoughts podcast. I'm Tracy Alloway. 842 00:45:23,239 --> 00:45:26,200 Speaker 1: You can follow me on Twitter at Tracy Alloway and 843 00:45:26,239 --> 00:45:29,239 Speaker 1: I'm Joe Wisntal. You can follow me on Twitter at 844 00:45:29,280 --> 00:45:31,960 Speaker 1: a Stalwart. And you should follow our guest on Twitter 845 00:45:32,120 --> 00:45:36,520 Speaker 1: cmx Millenmy he's at Cmax. Follow our producer on Twitter, 846 00:45:36,600 --> 00:45:40,279 Speaker 1: Laura Carlson at Laura M. Carlton. Follow the Bloomberg head 847 00:45:40,320 --> 00:45:44,359 Speaker 1: of podcast, Francesca Levi at Francesca Today, and check out 848 00:45:44,360 --> 00:45:47,880 Speaker 1: all of our podcasts under the handle AD Podcasts. Thanks 849 00:45:47,920 --> 00:46:13,439 Speaker 1: for listening to the year