1 00:00:01,440 --> 00:00:12,321 Welcome to Top Traders Unplugged. In markets success doesn’t 2 00:00:12,321 --> 00:00:24,411 come from predicting what happens next, it comes from being 3 00:00:24,411 --> 00:00:31,665 prepared for what you can’t predict. In each episode we go deep 4 00:00:04,620 --> 00:00:10,450 with some of the world’s most thoughtful minds in investing, economics, 5 00:00:10,450 --> 00:00:16,280 and beyond to understand how they think, how they prepare, and 6 00:00:16,280 --> 00:00:21,580 how they decide, and the experiences that shaped how they 7 00:00:21,580 --> 00:00:26,880 see the world. No noise, no short-cuts, just real conversations 8 00:00:26,880 --> 00:00:31,650 to help you think better and invest with confidence. 9 00:00:35,980 --> 00:00:39,742 Welcome and welcome back to this week's edition of the Systematic 10 00:00:39,742 --> 00:00:42,820 Investor series with Harry Moore and I, Niels Kaastrup-Larsen, 11 00:00:42,820 --> 00:00:46,924 where each week we take the pulse of the global markets through 12 00:00:46,924 --> 00:00:51,370 the lens of a rules-based investor. I also want to say a warm 13 00:00:51,370 --> 00:00:55,816 welcome if today is the first time you're joining us, and if someone 14 00:00:55,816 --> 00:00:59,578 who cares about you and your portfolio recommended that you tune 15 00:00:59,578 --> 00:01:04,366 into the podcast, I would like to say a big thank you for sharing 16 00:01:04,366 --> 00:01:08,128 this episode with your friends and your colleagues. It really does 17 00:01:08,128 --> 00:01:08,812 help us. Anyways, Harry, it's wonderful to see you and have you 18 00:00:47,330 --> 00:01:02,085 back on the show this week. How are you doing? What's going on 19 00:01:02,085 --> 00:01:08,895 in your part of the world? 20 00:01:09,620 --> 00:01:14,744 Yeah, great to see you, Niels. All good over here in New York. I 21 00:01:14,744 --> 00:01:18,770 think we've hit 100% humidity today, so a bit uncomfortable, a 22 00:01:18,770 --> 00:01:22,430 bit of an uncomfortable journey in. But other than that, 23 00:01:22,430 --> 00:01:27,188 all good over here. And yeah, thanks for having me back. I really 24 00:01:27,188 --> 00:01:27,920 appreciate it. 25 00:01:28,540 --> 00:01:32,544 Yeah, absolutely. We have a great lineup, not least thanks to 26 00:01:32,544 --> 00:01:37,640 you. And so, I'm excited about it. Before we dive into all of that, 27 00:01:37,640 --> 00:01:43,100 I love to just get a sense of kind of what's come across your desk 28 00:01:43,100 --> 00:01:47,104 that you found interesting. It doesn't have to be anything related 29 00:01:47,104 --> 00:01:51,836 to what we do. But is there anything that you have come across 30 00:01:51,836 --> 00:01:53,656 that you have enjoyed recently? 31 00:01:55,360 --> 00:02:06,241 Yeah, I think… So, as I mentioned, I moved to New York kind 32 00:02:06,241 --> 00:02:17,122 of two and a half years ago and obviously you look around, you 33 00:02:17,122 --> 00:02:27,166 see the bridges, the highways, you're like, oh, where did this kind 34 00:02:27,166 --> 00:02:38,047 of all come from? So, the book that is supposed to explain that 35 00:02:38,047 --> 00:02:47,254 best is the biography on Robert Moses. So, it's called The 36 00:02:47,254 --> 00:02:48,928 Power Broker. So, I ordered this book, it's like 1,300 pages. 37 00:01:58,798 --> 00:02:03,764 So, I started kind of lugging that around on the subway. It's a 38 00:02:03,764 --> 00:02:09,494 bit of a brick, to be honest. So, I've kind of given up. I don't 39 00:02:09,494 --> 00:02:14,460 know, it's a failure on my part, but I've given up and I've 40 00:02:14,460 --> 00:02:19,426 gone for the Kindle. So, I've been reading about it. But it's just 41 00:02:19,426 --> 00:02:24,392 very interesting just to see like why is the west side a huge 42 00:02:24,392 --> 00:02:28,976 road when we could have had some coastline here. But you just 43 00:02:28,976 --> 00:02:34,706 get a feel for a bit of a biography of New York, in a way, 44 00:02:34,706 --> 00:02:39,672 I feel, for how the city's been built. So, been making my way 45 00:02:39,672 --> 00:02:43,874 through that. Haven't quite finished it yet, but yeah, did give 46 00:02:43,874 --> 00:02:48,840 in to the paperback and had to go to Kindle on that, sadly. 47 00:02:50,060 --> 00:02:54,130 Yeah, super interesting, super interesting. Now, for my part, by 48 00:02:54,130 --> 00:02:59,421 the way, speaking of bricks, I don't know if you remember but back 49 00:02:59,421 --> 00:03:03,898 in the very early days of mobile phones, actually my brother-in-law 50 00:03:03,898 --> 00:03:07,968 still keeps his first-ever mobile phone and literally it was 51 00:03:07,968 --> 00:03:13,666 like a brick with an antenna on. I mean it looks so weird now, 52 00:03:13,666 --> 00:03:18,143 but it's, that's how people were carrying them around back then. 53 00:03:19,630 --> 00:03:23,309 Amazing. So, my first was a 3210. I don't know if you remember 54 00:03:23,309 --> 00:03:27,271 the 3210, but it had Snake. I think it did actually have a tiny 55 00:03:27,271 --> 00:03:30,950 little antenna, but it was one of the first ones that you could 56 00:03:30,950 --> 00:03:34,346 like easily kind of carry around in your pocket. So not quite 57 00:03:34,346 --> 00:03:36,610 as far back as those huge brick phones. 58 00:03:37,130 --> 00:03:38,570 Yeah, exactly. 59 00:03:39,850 --> 00:03:42,658 This book would have put me in good stead for lugging around one 60 00:03:42,658 --> 00:03:44,170 of those old phones, I think, Niels. 61 00:03:45,530 --> 00:04:06,606 Yeah. Now, so on my radar, nothing really surprising I guess, 62 00:04:06,606 --> 00:04:27,682 but it's just some of the headlines that I followed. Nothing 63 00:04:27,682 --> 00:04:50,674 really that is, well, somewhat linked to what we do, of course. 64 00:04:50,674 --> 00:05:13,666 But the first one is this story about this hedge fund, young 65 00:05:13,666 --> 00:05:27,078 guy, Leopold Aschenbrenner, and the situational awareness. And 66 00:03:46,815 --> 00:04:00,950 of course, you read the headlines and the news stories, and 67 00:04:00,950 --> 00:04:09,945 you shouldn't, probably shouldn't believe everything that 68 00:04:09,945 --> 00:04:27,935 it says. But in any event, I mean some of the things that you 69 00:04:27,935 --> 00:04:42,070 do read, the level of experience, the level of assets that 70 00:04:42,070 --> 00:04:56,205 he suddenly commanded, the level of leverage he was using that 71 00:04:56,205 --> 00:05:10,340 people probably should have been aware of, so to speak. There 72 00:05:10,340 --> 00:05:27,045 are just so many red flags, in my opinion, when you see that. And 73 00:03:46,461 --> 00:03:55,771 yet for many firms with decades of experience, and having 74 00:03:55,771 --> 00:04:06,943 demonstrated their ability to manage risk, and so on, and so forth, 75 00:04:06,943 --> 00:04:19,046 you see how difficult it is to raise money. And here someone with 76 00:04:19,046 --> 00:04:28,356 very little experience, very little track record ends up having 77 00:04:28,356 --> 00:04:37,666 $45 billion or so under management before losing two thirds 78 00:04:37,666 --> 00:04:50,700 of that, it seems, in a matter of weeks. So, I mean, it's close 79 00:04:50,700 --> 00:05:01,872 up probably from location wise to where you are. But, yeah, it's 80 00:05:01,872 --> 00:05:14,906 just a lot of red flags when you see something like that. Is that 81 00:05:14,906 --> 00:05:27,009 anything you talk about in New York at the moment? I imagine so. 82 00:05:27,160 --> 00:05:35,924 Yeah, I mean I think it's a good example of where you can be 83 00:05:35,924 --> 00:05:44,062 right and still hit issues. So, I mean it called the trade, he 84 00:05:44,062 --> 00:05:51,574 obviously had a great story from writing the essay. He was very 85 00:05:51,574 --> 00:05:59,086 well known in those circles. And while it was, I guess, known 86 00:05:59,086 --> 00:06:07,224 as a hedge fund, the hedge was essentially the other side of the 87 00:06:07,224 --> 00:06:15,362 AI trade. So, even if on paper you look somewhat like dollar neutral 88 00:06:15,362 --> 00:06:22,874 or market neutral, your theme has kind of caught you out there, 89 00:06:22,874 --> 00:06:24,126 I think. I chatted through this a little bit with like Yoav 90 00:05:37,150 --> 00:05:50,137 and Nick a few weeks ago on sort of capturing those themes, and 91 00:05:50,137 --> 00:06:03,124 it was a big bet on a theme. And that theme, obviously, reversed 92 00:06:03,124 --> 00:06:17,110 in July and that can show if you kind of compound on one idea 93 00:06:17,110 --> 00:06:24,103 what can happen if it goes wrong. 94 00:06:24,450 --> 00:06:31,950 Yeah, I think that's a good point, actually, that diversification 95 00:06:31,950 --> 00:06:40,950 really does matter, so to speak. The other thing where I guess 96 00:06:40,950 --> 00:06:49,200 you could say it's not necessary that the people who act 97 00:06:49,200 --> 00:06:58,950 are always right, but when they do act it does have an effect 98 00:06:58,950 --> 00:07:07,950 in the markets, and that's the central banks of Japan and possibly, 99 00:07:07,950 --> 00:07:16,950 I guess, the US as well intervening in the Japanese yen only 100 00:07:16,950 --> 00:07:26,700 a couple of days before month end, which it did have an effect. 101 00:07:26,700 --> 00:07:36,450 It did show up in, I think, performance of our industry as well. 102 00:06:24,450 --> 00:06:34,304 But, of course, on the back of 30 years of zero interest rates, 103 00:06:34,304 --> 00:06:44,158 only a couple of years ago, the BOJ started to raise rates for 104 00:06:44,158 --> 00:06:53,254 the first time in ages. And, of course, interest rates have continued 105 00:06:53,254 --> 00:07:00,834 to move higher, and the yen has (despite) actually continued 106 00:07:00,834 --> 00:07:11,446 to move lower. But then they made a real effort at least to change 107 00:07:11,446 --> 00:07:20,542 that. They've done it, I think, a couple of times before in 108 00:07:20,542 --> 00:07:28,880 the last two, three years. It's never really lasted very long, 109 00:07:28,880 --> 00:07:36,460 but it was interesting that they decided to come out. 110 00:07:37,530 --> 00:07:41,694 Yeah, it's one of those that, from a trend perspective, is tough 111 00:07:41,694 --> 00:07:45,511 as well. Right? Because sometimes I think it kind of works 112 00:07:45,511 --> 00:07:48,287 because you take that uncomfortable door position, and 113 00:07:48,287 --> 00:07:52,798 you hold it, and you write it down, and there's rumors is there 114 00:07:52,798 --> 00:07:56,268 going to be intervention? Humans start to panic. The rules-based 115 00:07:56,268 --> 00:07:59,738 systems don't. But of course, on this occasion, the intervention 116 00:07:59,738 --> 00:08:04,943 you did see a bump up. I mean, I did enjoy the photo. (I wasn't 117 00:08:04,943 --> 00:08:09,107 sure if it was a meme.) Apparently, it was real, of Scott 118 00:08:09,107 --> 00:08:13,618 Bessent's To-Do list for the day and it just was a blank sheet 119 00:08:13,618 --> 00:08:18,476 of paper with one bullet point that said ‘buy yen’. So, I mean I 120 00:08:18,476 --> 00:08:22,987 might need to fact that if that was an AI generated image. But 121 00:08:22,987 --> 00:08:27,151 it did make me chuckle. But yeah, it's the importance of sizing 122 00:08:27,151 --> 00:08:30,968 and making sure that one position can't hit you too much. 123 00:08:30,968 --> 00:08:34,438 But yeah, a bit painful when an intervention like that… 124 00:08:35,430 --> 00:08:35,990 Yeah. 125 00:08:35,990 --> 00:08:37,190 …makes something turn around. 126 00:08:38,150 --> 00:08:42,854 Sure. I mean you probably know this much better than I do, but if 127 00:08:42,854 --> 00:08:46,886 I'm not mistaken, I think it was like a 10 standard deviation 128 00:08:46,886 --> 00:08:51,254 move, which is rare. I know that when the Swiss franc got de-pegged 129 00:08:51,254 --> 00:08:55,622 from the euro, I think it was like 20 or 30 standard deviations 130 00:08:55,622 --> 00:08:59,318 or something like that. But still, 10 standard deviation is not 131 00:08:59,318 --> 00:09:01,670 something we should expect really to see. 132 00:09:02,250 --> 00:09:07,388 Not at all. Not at all. Right? That does not appear, and if you're 133 00:09:07,388 --> 00:09:11,058 building market data, it's very hard to actually simulate something 134 00:09:11,058 --> 00:09:15,829 like that. But we know they exist from time to time. I think 135 00:09:15,829 --> 00:09:20,967 when you've had a move go so far for so long, natural market forces 136 00:09:20,967 --> 00:09:25,371 weren't bringing it back and I think that was where they decided 137 00:09:25,371 --> 00:09:29,041 to intervene and that human intervention's caused the, as you 138 00:09:29,041 --> 00:09:31,977 say, very, very large move in the end. 139 00:09:32,150 --> 00:09:41,630 Yeah. All right, well, let's do a little bit of trend following 140 00:09:41,630 --> 00:09:51,110 update and then we'll get into some of these wonderful topics that 141 00:09:51,110 --> 00:09:59,800 we're going to be talking about. So, my trend barometer finished 142 00:09:59,800 --> 00:10:09,280 yesterday at 27. That's a very low reading actually. It's a pretty 143 00:10:09,280 --> 00:10:17,970 weak signal. So, it suggests some softness to the beginning of 144 00:10:17,970 --> 00:10:26,660 August, which we'll see also in the numbers in a second. We just 145 00:09:33,480 --> 00:09:41,460 finished July, which was, I would say, a month that, I mean, 146 00:09:41,460 --> 00:09:49,440 it's not that it was overall that bad, but there was definitely 147 00:09:49,440 --> 00:09:55,425 some dispersion among managers, which of course there always 148 00:09:55,425 --> 00:10:04,070 is, but still maybe more than I would normally expect in a single 149 00:10:04,070 --> 00:10:08,725 calendar month. Any observations, any takeaways from 150 00:10:08,725 --> 00:10:16,705 your side so far, whether it's August, July, whether it's year to 151 00:10:16,705 --> 00:10:22,690 date, anything that you find interesting about the environment 152 00:10:22,690 --> 00:10:26,680 for trend following at the moment? 153 00:10:27,900 --> 00:10:52,305 Yeah, I mean July was, if you look at July, it was kind of characterized 154 00:10:52,305 --> 00:11:18,337 by sort of two big themes. I would say you had a bit of a stopper 155 00:11:18,337 --> 00:11:36,234 on the AI trade, as we already briefly discussed, with obviously 156 00:11:36,234 --> 00:11:55,758 one hedge fund taking some pretty big losses. You also, of course, 157 00:11:55,758 --> 00:12:07,147 had kind of rallying energy markets again. And so, I think from 158 00:10:33,120 --> 00:10:43,560 a trend model perspective, if you're sort of taking sensitivity 159 00:10:43,560 --> 00:10:54,000 around three months, like a continuous sort of moving average, 160 00:10:54,000 --> 00:11:05,484 you probably would have been long markets like the oil complex, 161 00:11:05,484 --> 00:11:19,056 you would have been long stocks. So, I think this is why you're 162 00:11:19,056 --> 00:11:30,540 seeing such dispersion. So, choices about how much to have within 163 00:11:30,540 --> 00:11:44,112 each asset class, how much risk weight to have in that in the 164 00:11:44,112 --> 00:11:55,596 long run; choices of whether to exclude certain markets or whole 165 00:11:55,596 --> 00:12:07,080 asset classes would have created a lot of dispersion in July. 166 00:10:27,900 --> 00:10:40,176 So, I would expect people long equities would have been… that would 167 00:10:40,176 --> 00:10:51,429 have been tough, in general, particularly for some of those trend 168 00:10:51,429 --> 00:11:03,705 following kind of themes as well. So, if you were trend following 169 00:11:03,705 --> 00:11:14,958 themes and you were long, let's say, sort of tech hardware 170 00:11:14,958 --> 00:11:23,142 companies, you're long semiconductors, and you were short 171 00:11:23,142 --> 00:11:36,441 kind of software, that would have hurt us as well. So, I think 172 00:11:36,441 --> 00:11:48,717 that kind of explains a bit of that dispersion. You're seeing that 173 00:11:48,717 --> 00:11:59,970 Niels - more energies would have been good, more stocks generally 174 00:11:59,970 --> 00:12:07,131 bad. So yeah, big dispersion in July. 175 00:12:09,351 --> 00:12:39,696 I don't know if you managed to listen to it but last week I had 176 00:12:39,696 --> 00:13:05,995 Rich and Dave Dredge on the show and there was one thing that 177 00:13:05,995 --> 00:13:28,248 really, it wasn't surprising in that sense but still when you 178 00:13:28,248 --> 00:13:54,547 hear it you think, wow! And that was this example that Dave brought, 179 00:13:54,547 --> 00:14:14,777 up early on in the conversation, about, I think it's 180 00:14:14,777 --> 00:14:37,030 South Korean stock, SK Hynix or something like that, where the 181 00:14:37,030 --> 00:15:05,352 stock had gone up like 120% and then down to flat for over a 182 00:15:05,352 --> 00:15:31,651 3 month period. But the 2x levered ETF was down like 40+% for 183 00:15:31,651 --> 00:15:49,858 that period. Again, obviously highlighting the dangers of leverage 184 00:15:49,858 --> 00:16:14,134 and how some of these “retail products” are put together. But anyways, 185 00:16:14,134 --> 00:16:32,341 it certainly has been an interesting summer so far. But as 186 00:12:24,827 --> 00:14:20,897 I said I think for a lot of people it was really the yen that 187 00:14:20,897 --> 00:15:53,753 kind of defined how you're going to finish the month of July. 188 00:15:53,753 --> 00:16:32,443 So, let's look at that. So, the BTOP50… Obviously, we did the 189 00:12:25,941 --> 00:12:56,751 July figures last week, but as I said, actually, it's a little bit 190 00:12:56,751 --> 00:13:27,561 soft getting into August. So, we are down 67 basis points for the 191 00:13:27,561 --> 00:13:58,371 BTOP 50 (and that would be as of Tuesday night, we are recording 192 00:13:58,371 --> 00:14:26,811 on Thursday this week), it's still up 7.17% so far this year. 193 00:14:26,811 --> 00:14:59,991 The SocGen CT index is down 88 basis points, up 7.15% so far this 194 00:14:59,991 --> 00:15:30,801 year. The Trend index down about 1% or so - 94 basis points, 195 00:15:30,801 --> 00:15:56,871 up 6.89% for the year. And the Short-Term Traders Index actually 196 00:15:56,871 --> 00:16:27,681 up for the month 14 basis points, but up just 3.32% so far 197 00:16:27,681 --> 00:16:32,421 this year. If we look at the traditional markets, we have the 198 00:12:35,391 --> 00:13:09,243 MSCI World Index (and this is as of yesterday) up 2.77% in August, 199 00:13:09,243 --> 00:13:37,887 up 12.45 for the year (stripping out US and Canada based 200 00:13:37,887 --> 00:14:09,135 on the proposal from a listener) the E index is up 1.83% 201 00:14:09,135 --> 00:14:42,987 in August, up 11.8% for the year. US S&P US Aggregate Bond index 202 00:14:42,987 --> 00:15:16,839 up 64 basis points in August, up 31 basis points this year. And 203 00:15:16,839 --> 00:15:53,295 the S&P 500 total return of 3.12% so far this month, and up 12.83 204 00:15:53,295 --> 00:16:27,147 this year, a really strong, by the way, start this month to equities 205 00:16:27,147 --> 00:16:32,355 for sure. I mean, from my perspective, before we move on, not 206 00:12:44,431 --> 00:13:26,527 a massive amount of change, directional change in kind of a trend 207 00:13:26,527 --> 00:14:01,607 following portfolio. There are certainly changes in terms of the 208 00:14:01,607 --> 00:14:33,179 levels of commitment but, generally speaking, CTAs are still 209 00:14:33,179 --> 00:15:08,259 long equities, probably long metals, probably (as you said) long 210 00:15:08,259 --> 00:15:43,339 energies, probably long the dollar I would imagine, and then 211 00:15:43,339 --> 00:16:18,419 still, most likely, short fixed income, to some extent. So, 212 00:16:18,419 --> 00:16:32,451 not a lot there. When I look at another thing that you can comment 213 00:12:23,641 --> 00:12:45,076 on if you want, you don't have to. When I look at like a standard 214 00:12:45,076 --> 00:12:59,366 trend following model and I compare just different speeds, because 215 00:12:59,366 --> 00:13:16,514 you mentioned speed could have been a determinant, and I think that 216 00:13:16,514 --> 00:13:35,091 that's very true. So, I tend to look at a traditional trend following 217 00:13:35,091 --> 00:13:49,381 model and looking at say four different speeds, meaning lookback 218 00:13:49,381 --> 00:14:09,387 periods, 20 days, 60 days, 130 days, 260 days. And if you had chosen 219 00:14:09,387 --> 00:14:25,106 a 20-day lookback, again this is just from this specific model, 220 00:14:25,106 --> 00:14:43,683 you would actually have been down this year, and if you had chosen 221 00:14:43,683 --> 00:15:05,118 a 60 day you would also have been down but not as much. If you 222 00:15:05,118 --> 00:15:23,695 had chosen 130 days you would have been slightly up and if you 223 00:15:23,695 --> 00:15:43,701 had chosen an even further out 260 days you would have been up a 224 00:15:43,701 --> 00:16:00,849 little bit more. That's kind of how I see speed impacting, at 225 00:16:00,849 --> 00:16:17,997 the moment, the environment or the results. I don't know if you 226 00:16:17,997 --> 00:16:32,287 have anything that you want to say, comment, on here. 227 00:16:32,451 --> 00:16:42,783 Yeah, I would say that completely makes intuitive sense 228 00:16:42,783 --> 00:16:57,707 to me. So, if you're running a slower trend system in 2026, you 229 00:16:57,707 --> 00:17:09,187 would have been fairly more persistently sort of short bonds 230 00:17:09,187 --> 00:17:22,963 which I would have certainly paid in the past few months. Once 231 00:17:22,963 --> 00:17:35,591 you built into those long energy positions, you tended to hold 232 00:17:35,591 --> 00:17:49,367 them a little bit longer which on a year-to-date basis would have 233 00:17:49,367 --> 00:18:00,847 helped. You might have suffered a little bit more volatility 234 00:18:00,847 --> 00:18:07,735 there, but would be also adding. And then in stocks, I mean 235 00:16:44,619 --> 00:17:08,955 the longest models have done the best there as well. So, very 236 00:17:08,955 --> 00:17:25,179 long lookbacks. You've been persistently long stocks. Again, 237 00:17:25,179 --> 00:17:51,543 we've had some wobbles along the way, but if you look at stock 238 00:17:51,543 --> 00:18:07,767 indices, in general, across the board, they're up. So, it's certainly 239 00:16:34,536 --> 00:16:43,571 been a year so far. I would say slower has been beneficial. We 240 00:16:43,571 --> 00:16:50,521 haven't had a turning point that's persisted for long enough 241 00:16:50,521 --> 00:16:59,556 for the fastest speeds to kind of shine. So, echoes a little bit 242 00:16:59,556 --> 00:17:08,591 of what we've seen over the past few years. That slow has tended 243 00:17:08,591 --> 00:17:16,236 to outperform a bit. Again, we've done research on this. You've 244 00:17:16,236 --> 00:17:25,966 discussed it on your show. I mean, my view is always to have some 245 00:17:25,966 --> 00:17:33,611 dynamism there that you can still capture some of those turning 246 00:17:33,611 --> 00:17:41,951 points. But yeah, I completely agree with what you said, in general, 247 00:17:41,951 --> 00:17:48,901 you've on longer out which eventually means you've held onto 248 00:17:48,901 --> 00:17:57,241 your long energy positions, you tended to hold on to your short 249 00:17:57,241 --> 00:18:03,496 bond positions, and you've just been long stocks throughout 250 00:18:03,496 --> 00:18:07,666 the year which has worked well. 251 00:18:08,651 --> 00:18:12,237 One thing I don't personally follow so closely, I would imagine 252 00:18:12,237 --> 00:18:16,475 you do more so, and that's because at DUNN we just trade kind 253 00:18:16,475 --> 00:18:20,387 of the most liquid futures. But when the other thing that drives 254 00:18:20,387 --> 00:18:24,299 performance, as we often talk about before we get into the topics, 255 00:18:24,299 --> 00:18:27,559 is of course market universe. You've written beautiful paper about 256 00:18:27,559 --> 00:18:31,145 this that was published in late January actually. So, I don't 257 00:18:31,145 --> 00:18:35,057 know if you remember this, THIS is not something we kind of 258 00:18:35,057 --> 00:18:39,621 plan to talk about. But do you think there's been a lot of dispersion 259 00:18:39,621 --> 00:18:42,229 based on market universe this year as well? 260 00:18:43,281 --> 00:19:05,409 Yeah, absolutely. I would say this year you have really seen your 261 00:19:05,409 --> 00:19:25,693 kind of old versus traditional markets like diverge to a fairly 262 00:19:25,693 --> 00:19:47,821 large extent. And I think there are two places where this is 263 00:19:47,821 --> 00:20:09,949 most evident, or maybe three actually. So, the three that we could 264 00:20:09,949 --> 00:20:28,389 touch on would be in alternative energies, in fixed income, 265 00:20:28,389 --> 00:20:37,609 and then also in stock. So, in energies, if you go down that alternative 266 00:18:56,637 --> 00:19:11,477 market spectrum, you're starting to look at things like Nordic 267 00:19:11,477 --> 00:19:27,801 Power or kind of Italian Power. Those markets have just tended 268 00:19:27,801 --> 00:19:48,577 to be a little bit more range bound than we've seen in the more 269 00:19:48,577 --> 00:20:01,933 traditional energy markets this year. So, again mid frequency 270 00:20:01,933 --> 00:20:13,805 models tending to be profitable year-to-date in things 271 00:20:13,805 --> 00:20:33,097 like crude and heating oil, etc., and less so in some of the 272 00:20:33,097 --> 00:20:37,549 old energy use. In fixed income, fixed income has been pretty 273 00:18:57,345 --> 00:19:18,441 difficult for a few years for trend. Right? As you know, you've 274 00:19:18,441 --> 00:19:37,779 had these kind of competing forces in fixed income. You've got 275 00:19:37,779 --> 00:19:55,359 the administration shouting for rate cuts, rate cuts. You have 276 00:19:55,359 --> 00:20:11,181 markets follow that occasionally. Then you look at the 277 00:20:11,181 --> 00:20:30,519 inflation print, you look at rising deficits, and you've got these 278 00:20:30,519 --> 00:20:37,551 kind of two forces. And I think that's been even more pronounced 279 00:18:56,729 --> 00:19:16,901 as you move down the spectrum and away from your big traditional 280 00:19:16,901 --> 00:19:33,711 government bond futures, so away from 10-year treasury and your 281 00:19:33,711 --> 00:19:53,883 gold, and you go and look at, for example, Polish interest rate 282 00:19:53,883 --> 00:20:12,374 swaps or other markets like this. Again, those kind of gyrations 283 00:20:12,374 --> 00:20:34,227 have just been more pronounced and they tended to be a little bit 284 00:20:34,227 --> 00:20:37,589 more difficult. The final one to talk on, and we did discuss this 285 00:19:07,932 --> 00:19:34,824 to an extent in the paper, and actually my colleagues have recently 286 00:19:34,824 --> 00:20:06,198 written a paper on this as well if we get time to talk about 287 00:20:06,198 --> 00:20:35,331 it today, great, but otherwise we can of course save it for another 288 00:20:35,331 --> 00:20:37,572 time. 289 00:20:37,611 --> 00:20:38,371 Save it for next time. 290 00:20:38,371 --> 00:20:46,291 Essentially there, typically, when you move away from your stock 291 00:20:46,291 --> 00:20:55,795 indices, so move away from just like S&P and NASDAQ, Euro stocks, 292 00:20:55,795 --> 00:21:05,299 and you start to look at cash equity sectors, for example. So, 293 00:21:05,299 --> 00:21:14,803 you might look at for example like STEMIS versus software, as we 294 00:21:14,803 --> 00:21:24,307 mentioned in the intro. You might also look at things like value, 295 00:21:24,307 --> 00:21:31,435 momentum, quality. Those are much more cross sectional constructions. 296 00:21:31,435 --> 00:21:40,147 Right? So, as we were discussing, a slow model would have 297 00:21:40,147 --> 00:21:49,651 been long the indices. The indices have gone up this year and 298 00:21:49,651 --> 00:21:52,027 that's worked well. These cross-sectional models, I mean, from 299 00:20:42,505 --> 00:20:50,773 a high level, if course they have also been profitable just in 300 00:20:50,773 --> 00:20:59,041 a very different way. Right? So, if you think, if you construct 301 00:20:59,041 --> 00:21:07,309 a quality factor, quality has really struggled this year on a year-to-date 302 00:21:07,309 --> 00:21:15,577 basis. That factor you can imagine may have added as well. So, 303 00:21:15,577 --> 00:21:22,467 you've got three areas, I think, which have contributed like 304 00:21:22,467 --> 00:21:31,424 a big portion of the kind of variance between what you might call 305 00:21:31,424 --> 00:21:38,314 your traditional futures and forwards versus your more kind of 306 00:21:38,314 --> 00:21:45,893 OTC type instruments. So yeah, great note to notice that Niels, 307 00:21:45,893 --> 00:21:52,094 and I think that's absolutely worth a comment there. 308 00:21:57,471 --> 00:22:02,187 All right, well, let's move on to the topics you brought along. 309 00:22:02,187 --> 00:22:07,689 And before we get into kind of the papers, so to speak, I think 310 00:22:07,689 --> 00:22:12,012 you very nicely laid out something that ties into the initial 311 00:22:12,012 --> 00:22:17,121 part of our conversation, both in terms of AI and macro. So why 312 00:22:17,121 --> 00:22:21,444 don't you take us into today's main feature, so to speak? 313 00:22:22,351 --> 00:23:00,481 Yeah, absolutely. Thanks, Niels. And I think this was kind 314 00:23:00,481 --> 00:23:50,050 of motivated by a few things. Of course, the events in the past 315 00:23:50,050 --> 00:24:35,806 couple of weeks with the sort of unwind and people stepping in 316 00:24:35,806 --> 00:25:25,375 to buy this AI trade that hit the headlines, but also really motivated 317 00:25:25,375 --> 00:25:59,692 by your conversations you've been having on diversification. So, 318 00:25:59,692 --> 00:26:45,448 with David Dredge and Richard Brennan last week, and then a couple 319 00:26:45,448 --> 00:27:31,204 of weeks back with Nick and with Yoav. From a trend following 320 00:27:31,204 --> 00:28:16,960 perspective, looking at the AI trade and AI theme in markets, it's 321 00:28:16,960 --> 00:29:10,342 a very interesting time to kind of be part of it and be living 322 00:29:10,342 --> 00:29:17,968 through it. Because on the one hand you have all of these comparisons 323 00:23:04,690 --> 00:23:54,727 back to the railroads, back to laying cable in the dot com bust, 324 00:23:54,727 --> 00:24:48,613 and these forecasts that it has to be a bubble. This is one side 325 00:24:48,613 --> 00:25:42,499 of the argument, it has to be a bubble. It's got all the echoes, 326 00:25:42,499 --> 00:26:28,687 it's got all the hallmarks of these prior periods where CapEx went 327 00:26:28,687 --> 00:27:11,026 through the roof. There's mania about a new thing that's appeared, 328 00:27:11,026 --> 00:28:12,610 and it all kind of all ends in tears, so to say. Not to say that 329 00:28:12,610 --> 00:28:47,251 there isn't some very useful infrastructure that's left behind, 330 00:28:47,251 --> 00:29:18,043 but from a market's perspective, it's very painful. That's 331 00:22:26,869 --> 00:23:21,085 kind of one side of the argument. The thing is, I always 332 00:23:21,085 --> 00:24:15,301 try and really stay balanced. Right? Like that is a gripping narrative. 333 00:24:15,301 --> 00:25:09,517 It's a great story. It's a gripping narrative to kind of tell 334 00:25:09,517 --> 00:26:12,769 to a client in a meeting. But you've also got to consider the other 335 00:26:12,769 --> 00:27:06,985 scenario, the kind of upside scenario. And if I look within Man 336 00:27:06,985 --> 00:28:01,201 Group, we are integrating AI across all of our teams, across all 337 00:28:01,201 --> 00:28:50,899 of our processes. We're seeing benefits to quant research – just 338 00:28:50,899 --> 00:29:18,007 a huge amount of productivity gains. What do you want your researchers 339 00:23:13,249 --> 00:25:03,528 to be doing? You mostly want them to be thinking and being creative 340 00:25:03,528 --> 00:26:45,324 rather than writing lots and lots of Python code, where the Python 341 00:26:45,324 --> 00:28:44,086 code is now kind of table stakes and you can do it much, much 342 00:28:44,086 --> 00:29:18,018 faster with these tools. Do I see the benefits here? And you hear 343 00:22:45,301 --> 00:23:10,801 obviously the Frontier Labs, as they would say, talking about 344 00:23:10,801 --> 00:23:43,951 what the future might look like and so, I always try and stay 345 00:23:43,951 --> 00:24:12,001 balanced on this. Maybe there is this huge upside scenario where 346 00:24:12,001 --> 00:24:47,701 the trade that we saw go wrong last month is the right trade and 347 00:24:47,701 --> 00:25:18,301 it works over the next 10 years. There's this kind of real 348 00:25:18,301 --> 00:25:48,901 level of uncertainty and I think it links to your tagline of 349 00:25:48,901 --> 00:26:19,501 the show – being prepared for what you just really can't predict. 350 00:26:19,501 --> 00:26:52,651 And that's how I think kind of trend following comes into it. Because 351 00:26:52,651 --> 00:27:23,251 if you think, you're trying to build a portfolio to be resilient 352 00:27:23,251 --> 00:27:48,751 in this environment. Right? So, you're allocating to many different 353 00:27:48,751 --> 00:28:14,251 asset classes. You probably, in an allocated portfolio, have a 354 00:28:14,251 --> 00:28:44,851 decent amount of equities, you have a decent amount of bonds, you 355 00:28:44,851 --> 00:29:05,251 probably hold potentially some diversifiers like commodities, you 356 00:29:05,251 --> 00:29:18,001 might hold some private assets. Now in that portfolio, AI 357 00:23:06,571 --> 00:24:26,167 trade keeps going, economic growth keeps rising, CapEx keeps 358 00:24:26,167 --> 00:26:03,451 rising. That portfolio is rising. Now on paper you look diversified. 359 00:26:03,451 --> 00:27:40,735 These assets, they seem to have, in normal times, somewhat low 360 00:27:40,735 --> 00:29:09,175 correlations with each other. You feel diversified and you feel 361 00:29:09,175 --> 00:29:18,019 comfortable. Now, in a crack to that trade or a full blow-up. 362 00:23:48,624 --> 00:25:38,426 And again, I mean here like the S&P isn't down 5% or 10%, it's 363 00:25:38,426 --> 00:27:36,071 down 30% or 40% over a period. When you look at that book that you 364 00:27:36,071 --> 00:29:18,030 own, what in there is going to protect you against that huge drop? 365 00:22:22,351 --> 00:23:10,993 So, you would say, well the equities, we've just said, potentially 366 00:23:10,993 --> 00:24:04,057 they're down 30% or 40%, private assets - you can slow your 367 00:24:04,057 --> 00:25:01,543 marks, they may mark down more slowly but at some point you're going 368 00:25:01,543 --> 00:25:59,029 to take hits there. Right? And we've all seen the slowdown of return 369 00:25:59,029 --> 00:26:52,093 of capital. We've seen the FT and others running articles on what 370 00:26:52,093 --> 00:27:45,157 a client's going to do when they're getting a lot less capital 371 00:27:45,157 --> 00:28:29,377 back than they expect. You're certainly not getting any long-term 372 00:28:29,377 --> 00:29:18,019 protection there. You might get some paper protection for a period. 373 00:22:22,351 --> 00:22:54,943 Then in bonds and gold, earlier in the year we saw, again, 374 00:22:54,943 --> 00:23:27,535 both of those sell off in a fairly correlated manner. The difficulty 375 00:23:27,535 --> 00:24:05,559 as well is if you look at the environment today and you look at 376 00:24:05,559 --> 00:24:40,867 the Fed in the US and you say well inflation's now been running 377 00:24:40,867 --> 00:25:16,175 above target for many years. It's not just like it's a short period, 378 00:25:16,175 --> 00:25:51,483 it's many years. Do they have the tools to really cut rates like 379 00:25:51,483 --> 00:26:29,507 we've seen in the past to kind of buoy markets? And if you're worried, 380 00:26:29,507 --> 00:26:59,383 like maybe they do, but if you're slightly worried about that 381 00:26:59,383 --> 00:27:29,259 then you might be in an environment where your bonds are 382 00:27:29,259 --> 00:28:07,283 also selling off very hard as well. And this is where I think trend 383 00:28:07,283 --> 00:28:37,159 kind of comes into the equation because, as we know, trend 384 00:28:37,159 --> 00:29:09,751 doesn't have a long-term view on any of these markets. Right? It 385 00:29:09,751 --> 00:29:17,899 trades them all. Okay, it doesn't trade private assets, but 386 00:22:48,328 --> 00:23:18,016 you’re trading equities, you're trading bonds, gold, oil, 387 00:23:18,016 --> 00:23:58,837 etc. It doesn't have a long-term view on those. It's just 388 00:23:58,837 --> 00:24:35,947 simply following the prices and saying, well, what I'm seeing 389 00:24:35,947 --> 00:25:27,901 in the last two weeks, in the last four weeks, in the last two 390 00:25:27,901 --> 00:26:05,011 months, is actually the narrative is changing. And I don't 391 00:26:05,011 --> 00:26:49,543 actually want to be long bonds and long equities. I actually want 392 00:26:49,543 --> 00:27:34,075 to be short bonds and short equities. If it's an inflation shock, 393 00:27:34,075 --> 00:28:11,185 you want to be long commodities. And so, trend actually 394 00:28:11,185 --> 00:28:55,717 moves into those positions. And now that's not to say it's always 395 00:28:55,717 --> 00:29:17,983 going to get it right. Right? It's going to cause you some pain 396 00:22:49,294 --> 00:23:35,482 from time to time. It's really being prepared for this tail, this 397 00:23:35,482 --> 00:24:21,670 left tail - the scenario I described at the start, when there 398 00:24:21,670 --> 00:25:11,707 is a crack, there's a break, maybe AI does deliver, but it's just 399 00:25:11,707 --> 00:26:01,744 not at the level that people have priced in. And having an asset 400 00:26:01,744 --> 00:26:47,932 like trend following in your book that can go short bonds, it 401 00:26:47,932 --> 00:27:34,120 can go short equities as we described, it might even go short 402 00:27:34,120 --> 00:28:24,157 the AI theme. Right? It might start to short SEMIS and short the 403 00:28:24,157 --> 00:29:02,647 software, the hardware, the tech hardware builders like you could 404 00:29:02,647 --> 00:29:18,043 go short those names. And really it's just offering that something 405 00:23:04,519 --> 00:24:10,783 completely different in your book that, as an investor, when you 406 00:24:10,783 --> 00:25:17,047 look and you say, okay, my assets are uncorrelated, but are 407 00:25:17,047 --> 00:26:23,311 they uncorrelated in that left tail shock? Like, what does your 408 00:26:23,311 --> 00:27:29,575 covariance matrix look like in that shock? And typically it's trend 409 00:27:29,575 --> 00:28:35,839 following that is zero or kind of negatively correlated. And other 410 00:28:35,839 --> 00:29:18,007 asset classes of correlations typically pick up. 411 00:29:18,691 --> 00:29:44,587 Yeah. I mean, you're kind of tempted to say that the beauty of 412 00:29:44,587 --> 00:30:06,499 trend following is that it's aware of the situation, it's situationally 413 00:30:06,499 --> 00:30:28,411 aware. But I think that name is taken, unfortunately, or maybe 414 00:30:28,411 --> 00:30:36,379 fortunately in this case. Now let me ask you something. You mentioned 415 00:29:26,763 --> 00:29:38,871 something that because I'm not a quant, I find it interesting. You 416 00:29:38,871 --> 00:29:49,970 mentioned this thing about the productivity gains that you see from 417 00:29:49,970 --> 00:30:05,105 having AI tools, and so on, and so forth. Just sort of on a very 418 00:30:05,105 --> 00:30:15,195 basic level, I understand the gains that someone like yourself 419 00:30:15,195 --> 00:30:25,285 who were taught how to program, you can essentially tell 420 00:30:25,285 --> 00:30:36,384 it what to do and you can also review the output. Are you, when 421 00:29:21,883 --> 00:29:35,715 you sit in your quiet moment in the underground, on your way home, 422 00:29:35,715 --> 00:29:47,419 are you thinking, well, what about the next generation where they 423 00:29:47,419 --> 00:30:01,251 will use AI as kind of their first tool, but they actually don't 424 00:30:01,251 --> 00:30:10,827 have the background and experience in having done programming 425 00:30:10,827 --> 00:30:25,723 myself, so how are they going to tell if what the output was is 426 00:30:25,723 --> 00:30:36,363 any good? Is that something you think about at all? 427 00:30:37,911 --> 00:31:12,747 Yeah, absolutely. Niels. And I think to best answer this, if we 428 00:31:12,747 --> 00:31:56,292 go back and kind of the story of when I joined Man Group, so I'm 429 00:31:56,292 --> 00:32:25,322 originally an actuary and basically in actuarial work 10, 15 430 00:32:25,322 --> 00:33:00,158 years ago, programming is not a big part of the syllabus. Right? 431 00:33:00,158 --> 00:33:37,897 So, when I joined Man, the first thing I did was go headfirst 432 00:33:37,897 --> 00:34:12,733 into all the developed courses run here and spent hundreds of hours 433 00:34:12,733 --> 00:34:44,666 in Python being able to do financial analysis, this sort of 434 00:34:44,666 --> 00:34:59,181 simple pandas, data manipulation, etc. Now today, when 435 00:30:53,910 --> 00:31:57,906 I look at the tools available like that was essentially a complete 436 00:31:57,906 --> 00:33:01,902 waste of time. To the level that I can program it's absolutely 437 00:33:01,902 --> 00:34:05,898 a complete, useless waste of 200 hours sort of seven years ago 438 00:34:05,898 --> 00:34:59,228 now. So, at that level the tools are absolutely fantastic. Now, 439 00:30:41,395 --> 00:31:19,719 if you think of a quant researcher and what you're really 440 00:31:19,719 --> 00:32:01,527 employing them for, what's the value of them in their seat? Right? 441 00:32:01,527 --> 00:32:50,303 So, 5, 6, 7 years ago, how's your time spent? Well, it might be 442 00:32:50,303 --> 00:33:28,627 50% doing the creative stuff, thinking, and 50% (maybe more than 443 00:33:28,627 --> 00:34:06,951 50%) is on implementation details - being good at Python, knowing 444 00:34:06,951 --> 00:34:48,759 how to run and set up a backtest, maybe replicating a test 445 00:34:48,759 --> 00:34:59,211 across multiple assets. Now, if you think where the value is, 446 00:30:58,607 --> 00:31:29,651 the value is not really in that second 50%. There are many, 447 00:31:29,651 --> 00:32:05,869 many more people, maybe 5, maybe 10 to 1 times as many who you 448 00:32:05,869 --> 00:32:39,500 can employ to do the second 50%. It's kind of table stakes in 449 00:32:39,500 --> 00:33:07,957 quantitative research - being good at Python, being able to run 450 00:33:07,957 --> 00:33:41,588 a backtest. Where the value really is, is that if you could split 451 00:33:41,588 --> 00:34:15,219 that 50/50 and make it more like 90/10 so that your top researchers 452 00:34:15,219 --> 00:34:48,850 have plenty more time to think and be creative. I think that's a 453 00:34:48,850 --> 00:34:59,198 big AI enabler. Right? AI, while it can summarize everything 454 00:31:06,417 --> 00:31:58,678 that's been written before, it can summarize the tax code, it’s 455 00:31:58,678 --> 00:32:55,690 very, very difficult to come up with brand new kind of quantitative 456 00:32:55,690 --> 00:33:47,951 high alpha ideas. That's what you want your researchers to do. 457 00:33:47,951 --> 00:34:40,212 And it's then what I would call table stakes jobs, speeding 458 00:34:40,212 --> 00:34:59,216 up and picking up. Now how that links to hiring. I mean, I think 459 00:31:19,391 --> 00:32:09,167 this is one we're all still trying to figure out. Because those 460 00:32:09,167 --> 00:32:50,647 very junior level graduate type roles, where you would spend 461 00:32:50,647 --> 00:33:40,423 more time on the kind of BAU implementation, etc., it's kind of 462 00:33:40,423 --> 00:34:26,051 being squeezed out by these tools. And we're automating a lot 463 00:34:26,051 --> 00:34:59,235 of processes that used to require more humans. So, it's a great 464 00:30:49,523 --> 00:31:18,553 question, like, how do you really interview for these skills? 465 00:31:18,553 --> 00:31:53,389 And I think it's, again, on the optimistic side, the people coming 466 00:31:53,389 --> 00:32:34,031 in now have grown up and been more involved with the kind of frontier 467 00:32:34,031 --> 00:33:11,770 models. Maybe they are more AI literate than kind of all of us. 468 00:33:11,770 --> 00:33:43,703 And so, they have that advantage. I think the big disadvantage 469 00:33:43,703 --> 00:34:18,539 is that just temptation at every moment. The answer to the problem 470 00:34:18,539 --> 00:34:53,375 you want to solve, especially when you're junior, is just one button 471 00:34:53,375 --> 00:34:59,181 click away. Whereas you and I, we had to go and dig out the papers, 472 00:31:42,014 --> 00:32:41,186 we had to read the theory, we maybe did some examples ourselves. 473 00:32:41,186 --> 00:33:40,358 I mean the actuarial exams you literally hand write out, like, ruin 474 00:33:40,358 --> 00:34:34,599 theory you were talking about last week, you're writing out all 475 00:34:34,599 --> 00:34:59,254 of the maths behind that. The new generations will never have to 476 00:30:56,202 --> 00:31:30,171 do that. And you have to have some serious discipline to go and 477 00:31:30,171 --> 00:32:06,753 dig the book out when you know you can get a perfect one-page summary 478 00:32:06,753 --> 00:32:40,722 with a click. So, I think you've got to test, like, give someone 479 00:32:40,722 --> 00:33:09,465 an AI generated piece, get them to go and critically analyze 480 00:33:09,465 --> 00:33:43,434 it, like what's wrong with it? What would have I improved on it? 481 00:33:43,434 --> 00:34:17,403 Because it's just a fact of life now. Locking someone in a Wi-Fi 482 00:34:17,403 --> 00:34:51,372 free, computer free box is not a good test of what working life 483 00:34:51,372 --> 00:34:59,211 looks like today. 484 00:34:59,341 --> 00:35:09,289 Right? Yeah, that's very true. That's very true. I don't know about 485 00:35:09,289 --> 00:35:19,237 you, but I actually often get the question during sort of more 486 00:35:19,237 --> 00:35:28,356 due diligence type meetings. And people might ask, well, do all 487 00:35:28,356 --> 00:35:37,475 your researchers do their own programming? And of course, the answer 488 00:35:37,475 --> 00:35:49,081 so far has always been yes, of course they do. That's part of what 489 00:35:49,081 --> 00:36:01,516 we want them to be able to do. And they need to review each other's 490 00:36:01,516 --> 00:36:07,319 code and all of that stuff, right? And I guess soon we'll just 491 00:35:09,805 --> 00:35:32,477 say no, they don't know how to program, and we'll see what face 492 00:35:32,477 --> 00:35:55,149 they put up, the investors, if that's a confidence face or a less 493 00:35:55,149 --> 00:36:07,357 confident face, but only time will tell. Now, before we move to 494 00:35:03,591 --> 00:35:14,641 the paper, I want to make sure, I think there might be more 495 00:35:14,641 --> 00:35:26,541 things. Do you want to link to more things before we dive into the 496 00:35:26,541 --> 00:35:36,741 first paper? Because we're going to talk about a new paper that 497 00:35:36,741 --> 00:35:45,241 you co-wrote with a few colleagues, Jonathan Smith and Chris 498 00:35:45,241 --> 00:35:57,141 Pi, I think it was, which is on Portable Alpha. But is there something 499 00:35:57,141 --> 00:36:07,341 more you wanted to bring up before we move into that, Harry? 500 00:36:08,081 --> 00:36:58,215 Yeah, I think we can jump into the paper. Maybe just to set the 501 00:36:58,215 --> 00:37:48,349 scene a little bit, I guess a couple of the things that the paper 502 00:37:48,349 --> 00:38:31,321 really dives into you sort of discussed last week, the week before. 503 00:38:31,321 --> 00:39:21,455 And if I go back to as well, Rob and Katy, you mentioned some 504 00:39:21,455 --> 00:40:15,170 of the topics there as well. So, I think a good way to lead in 505 00:40:15,170 --> 00:40:58,142 it to start with kind of the cash efficiency of trend following 506 00:40:58,142 --> 00:41:23,209 because that really builds into portable alpha. And I mean all 507 00:36:44,101 --> 00:38:32,161 of your audience already know this, but trend is running on futures, 508 00:38:32,161 --> 00:40:29,226 right? You don't need to put up a hundred dollars of capital to 509 00:40:29,226 --> 00:41:23,256 get your hundred dollars of exposure. Now this unlocks quite 510 00:36:17,629 --> 00:36:46,273 a few interesting features as an investor, right? If I only need 511 00:36:46,273 --> 00:37:17,304 to use US$20 or US$25 to get my US$100 of trend following, it 512 00:37:17,304 --> 00:37:53,109 gives me a few new levers and things that I can do with the freed 513 00:37:53,109 --> 00:38:24,140 up cash. One common thing I hear is, I like my portfolio. I've 514 00:38:24,140 --> 00:38:57,558 been happy with it. Like I'm going to have to really kind of sell 515 00:38:57,558 --> 00:39:28,589 something down to free up some space for trend. And because of that 516 00:39:28,589 --> 00:39:57,233 capital efficiency, well, you don't have to sell very much or if 517 00:39:57,233 --> 00:40:25,877 anything at all, right? So, you can get an allocation, you can 518 00:40:25,877 --> 00:40:44,973 get the potential benefits, the defensiveness, the convexity, 519 00:40:44,973 --> 00:41:11,230 the things that all your audience know very well for very, 520 00:41:11,230 --> 00:41:23,165 very little capital up front. Now, the next thing I want to mention, 521 00:36:38,825 --> 00:37:24,941 and, again, it links to the paper that Rob and Katy discussed 522 00:37:24,941 --> 00:38:07,214 was on, it's cash efficient. You're getting this big benefit from 523 00:38:07,214 --> 00:38:41,801 directionally following markets. And in the paper that they 524 00:38:41,801 --> 00:39:20,231 discussed, they said, well, what is the additional benefit of 525 00:39:20,231 --> 00:39:50,975 the cross-sectional trend models? Could I replace cross-sectional 526 00:39:50,975 --> 00:40:33,248 trend models with, for example, carry? And I think the paper 527 00:40:33,248 --> 00:41:19,364 goes into what other alpha sources might you want to put alongside 528 00:41:19,364 --> 00:41:23,207 trend? Now, because it's so cash efficient, right? You're not 529 00:36:33,182 --> 00:37:09,439 having to stump up the full amount of dollars that frees up cash 530 00:37:09,439 --> 00:37:42,907 to go and, I would argue, keep the cross-sectional parts, but it 531 00:37:42,907 --> 00:38:10,797 had lower correlation, it still had positive Sharpe, it was 532 00:38:10,797 --> 00:38:38,687 still additive and add additional alpha models on top. So, 533 00:38:38,687 --> 00:39:20,522 in this sense you can sort of have your cake and eat it too, right? 534 00:39:20,522 --> 00:39:45,623 You're getting the full benefit of your trend allocation. 535 00:39:45,623 --> 00:40:21,880 You can fund it with, let's say, your US$20 to US$25. You can 536 00:40:21,880 --> 00:41:00,926 use that freed up cash to add some extra alpha in there. So, that's 537 00:41:00,926 --> 00:41:23,238 one way of looking at the problem. Right? Imagine, again, you're 538 00:36:25,919 --> 00:37:43,217 not getting your cash flows back as fast as you want from your 539 00:37:43,217 --> 00:38:36,731 private allocations. Maybe you have less equity, therefore less 540 00:38:36,731 --> 00:39:30,245 cash available to invest. Trend following is potentially a 541 00:39:30,245 --> 00:40:53,489 nice way to get access because you don't need all of that cash back 542 00:40:53,489 --> 00:41:23,219 to get the full allocation. Now, the way this then links into 543 00:36:23,950 --> 00:36:48,887 the paper is essentially the paper is fully focused on portable 544 00:36:48,887 --> 00:37:18,358 alpha. So, we've talked about a few ways that you could maybe take 545 00:37:18,358 --> 00:37:45,562 the cash and invest in other alphas. You could just simply use 546 00:37:45,562 --> 00:38:12,766 less cash to allocate to trend. What the paper is doing is 547 00:38:12,766 --> 00:38:37,703 it's jumping into the portable alpha topic and saying, well, what 548 00:38:37,703 --> 00:39:07,174 I'm going to do here is I'm going to sell essentially no equity 549 00:39:07,174 --> 00:39:34,378 exposure. I'm going to keep my portfolio exactly as it was. And 550 00:39:34,378 --> 00:40:08,383 what I'm going to do is I'm going to take a portion of my equities, 551 00:40:08,383 --> 00:40:40,121 let's say I own some passive S&P and I'm going to hold that via 552 00:40:40,121 --> 00:41:09,592 a future or a swap, and I'm going to port my trend following 553 00:41:09,592 --> 00:41:23,194 or my other allocation on top. Now, before we kind of dive in, the 554 00:36:34,617 --> 00:37:14,421 reason that we wrote this, and Jonathan and Chris were superb as 555 00:37:14,421 --> 00:37:57,542 well in contributing a lot of the analysis and some of the structuring 556 00:37:57,542 --> 00:38:43,980 as well to this paper. But the reason we wrote this is that there's 557 00:38:43,980 --> 00:39:20,467 tons of introductory material on portable alpha now. If you go 558 00:39:20,467 --> 00:40:03,588 and search, you'll be able to find how should you do it? Should 559 00:40:03,588 --> 00:40:46,709 you use turnkey? Should you do it yourself? What are the kind of 560 00:40:46,709 --> 00:41:19,879 risks of combining over 100% exposure? There's plenty of introductory 561 00:41:19,879 --> 00:41:23,196 material. What I found, sitting in meetings and listening 562 00:36:27,321 --> 00:36:51,371 to people asking questions about portable alpha, was the kind 563 00:36:51,371 --> 00:37:22,636 of next level down, what I would call this sort of second and 564 00:37:22,636 --> 00:37:49,091 third derivative just wasn't covered very well. So, if you go 565 00:37:49,091 --> 00:38:15,546 online and you search, for example, how should I rebalance portable 566 00:38:15,546 --> 00:38:46,811 alpha, or how much cash should I hold in a portable alpha structure, 567 00:38:46,811 --> 00:39:15,671 you don't find great answers and you don't really find much analysis 568 00:39:15,671 --> 00:39:46,936 on it. That was really the impetus to go out and write this. 569 00:39:46,936 --> 00:40:15,796 Let's use some simple examples to kind of make those points and 570 00:40:15,796 --> 00:40:44,656 bring them home. And yeah, we published it this week and yeah, 571 00:40:44,656 --> 00:41:15,921 happy to dive in. But if there are any questions up front, Niels, 572 00:41:15,921 --> 00:41:23,136 please go ahead. 573 00:41:23,261 --> 00:41:44,057 No, I mean, I think you're absolutely right. I think it's an 574 00:41:44,057 --> 00:42:01,387 important point that. Because I remember first coming across portable 575 00:42:01,387 --> 00:42:25,649 alpha, back in the mid to late ‘90s, talking about it. At the time, 576 00:42:25,649 --> 00:42:46,445 the firm I work for, we never launched anything but the concepts 577 00:42:46,445 --> 00:43:10,707 were there. And then, for me, I think really it came to the forefront 578 00:43:10,707 --> 00:43:29,770 again when Eric Crittenden and his group of people launched the 579 00:43:29,770 --> 00:43:47,100 Standpoint product in, I think, 2020, just before COVID, if 580 00:43:47,100 --> 00:43:52,299 I'm not mistaken. That's really where it came back into light 581 00:41:36,365 --> 00:41:54,383 and says, yeah, this is interesting because, to be very frank, 582 00:41:54,383 --> 00:42:14,039 there is definitely a slide that has been sitting in my slide 583 00:42:14,039 --> 00:42:35,333 deck for many years. And that is just to visualize for people what 584 00:42:35,333 --> 00:42:54,989 happens when you combine trend and equities. Because a lot of people 585 00:42:54,989 --> 00:43:16,283 think, well, if I put 50% here and 50% there, I'll probably get 586 00:43:16,283 --> 00:43:35,939 an average of the two. But actually, you get almost the sum 587 00:43:35,939 --> 00:43:52,319 of the two return streams, which is the crazy thing. And so, 588 00:41:28,149 --> 00:41:52,589 obviously, as you say, it's become popular again. Many people 589 00:41:52,589 --> 00:42:17,029 have launched products. Corey and his friends have really made 590 00:42:17,029 --> 00:42:43,913 it very public with the return stacking concept. So, that's another 591 00:42:43,913 --> 00:43:15,685 way of thinking about it. So, they've done a great job in that 592 00:43:15,685 --> 00:43:49,901 and had big success in raising assets for it. So yeah, it's an important 593 00:43:49,901 --> 00:43:52,345 point. I also have some pushback that I get in meetings, 594 00:41:47,701 --> 00:42:21,917 which I'll save and ask you a little bit later. But what I really 595 00:42:21,917 --> 00:42:53,689 loved about your paper is the fact that it actually does ask questions 596 00:42:53,689 --> 00:43:25,461 that even when I saw it, I thought, yeah, we never really talk 597 00:43:25,461 --> 00:43:52,345 about these topics. We should, because they are really, really important. 598 00:41:23,261 --> 00:42:03,191 So, for from my point, I would just love for you to go through these 599 00:42:03,191 --> 00:42:27,149 questions that you so eloquently ask and answer because 600 00:42:27,149 --> 00:42:59,093 I think people will be… And obviously, people should go and download 601 00:42:59,093 --> 00:43:28,375 the paper and read it, of course. But they're truly important 602 00:43:28,375 --> 00:43:52,333 in the context of this conversation, without a doubt. 603 00:43:53,641 --> 00:44:04,537 Yeah. Thanks so much, Niels. And it really was to answer some 604 00:44:04,537 --> 00:44:14,525 of those questions, actually. When I first kicked off doing some 605 00:44:14,525 --> 00:44:27,237 of this work, we thought, oh, this will be a week and it'll be 606 00:44:27,237 --> 00:44:39,949 so fast. And as you do it, you realize just how many questions you 607 00:44:39,949 --> 00:44:47,213 need to answer to build a resilient structure. So, yeah, if 608 00:43:55,489 --> 00:44:04,113 we dive in firstly on the title. I know last time I was on 609 00:44:04,113 --> 00:44:11,505 with Katy, and we were sort of touting writing a portable alpha 610 00:44:11,505 --> 00:44:17,665 paper. I think the working title was something like Covariance 611 00:44:17,665 --> 00:44:24,441 of Alpha and Beta in Stressed Markets - some horrible mouthful. 612 00:44:24,441 --> 00:44:30,601 Fortunately, we have a very good editorial and marketing team 613 00:44:30,601 --> 00:44:37,993 here and the title is now Portable Alpha, Ask the Hard Questions, 614 00:44:37,993 --> 00:44:46,001 which I think is a much more clickable title than what we had 615 00:44:46,001 --> 00:44:47,233 penciled originally. 616 00:44:47,581 --> 00:44:50,694 Sometimes it's good when you just say it actually does exactly 617 00:44:50,694 --> 00:44:54,656 what it says on the tin. And that's kind of what this title illustrates, 618 00:44:54,656 --> 00:44:54,939 yeah. 619 00:44:54,941 --> 00:45:08,539 Absolutely. And so, if we dive in, what were we really trying to 620 00:45:08,539 --> 00:45:20,045 answer here? So, there's a few things that kind of continuously 621 00:45:20,045 --> 00:45:34,689 come up in portable alpha. Now we know that in 2008 many of these 622 00:45:34,689 --> 00:45:44,103 structures failed and the lines down which those structures 623 00:45:44,103 --> 00:45:54,563 failed, those risks still exist. Right? So, not enough liquidity, 624 00:45:54,563 --> 00:46:03,977 correlations between your alpha and your beta, those things 625 00:46:03,977 --> 00:46:12,345 can still blow up and hurt you today. So very much not saying, oh, 626 00:45:05,747 --> 00:45:27,359 don't worry about it, like the world is different to 2008. It's 627 00:45:27,359 --> 00:45:45,369 very much saying, well, how can we build resilient structures 628 00:45:45,369 --> 00:46:06,981 that would survive docs like the GFC, like the dot com bubble, 629 00:46:06,981 --> 00:46:12,384 and like COVID. Now, the first thing I would pull out is essentially 630 00:45:08,291 --> 00:45:24,311 what we found on cash buffers in these strategies. Now, if you 631 00:45:24,311 --> 00:45:37,661 think of a portable alpha structure, you're buying your beta. 632 00:45:37,661 --> 00:45:53,681 Let's use S&P 500 as an example. You're then buying an alpha. 633 00:45:53,681 --> 00:46:09,701 And now, throughout the paper, we use SG Trend, obviously, as a 634 00:46:09,701 --> 00:46:12,371 trend podcast. 635 00:46:12,391 --> 00:46:14,231 Slightly biased here. 636 00:46:15,991 --> 00:48:32,061 But fits nicely. You could pick multi strategy, you could pick 637 00:48:32,061 --> 00:50:48,131 other alphas in there. And we've used trend in the analysis, 638 00:50:48,131 --> 00:52:51,831 but internally we've done this analysis across many different alpha 639 00:52:51,831 --> 00:55:07,901 structures. And so, you're buying 100% of your beta and then 640 00:55:07,901 --> 00:56:34,491 you are layering the alpha on top. So, what you're aiming to do 641 00:48:28,531 --> 00:52:09,431 is replicate the passive index. You're aiming to deliver the 642 00:52:09,431 --> 00:56:34,511 S&P to your investors plus whatever alpha you can get on top. 643 00:46:15,991 --> 00:48:17,486 Now, obviously when you put your S&P onto a derivative, onto 644 00:48:17,486 --> 00:50:41,071 a future, or a swap, you're no longer putting up the full US$100 645 00:50:41,071 --> 00:53:04,656 to get your US$100 exposure. Now you can put as little as US$10 646 00:53:04,656 --> 00:55:39,286 in to get your US$100. Now, that frees up US$90 to go and buy 647 00:55:39,286 --> 00:56:34,511 lots and lots of alpha. The problem is, in a stress scenario 648 00:47:09,443 --> 00:48:25,803 where your beta starts to sell, well, your counterparties are 649 00:48:25,803 --> 00:49:57,435 going to start calling you and saying, hey, your position is losing, 650 00:49:57,435 --> 00:51:29,067 you need to send us more margin. Now, if you've only supported 651 00:51:29,067 --> 00:53:00,699 your S&P with US$10 of capital, and the S&P falls by 10%, 652 00:53:00,699 --> 00:54:39,967 all of your margin is gone. Right? And so, you need some unencumbered 653 00:54:39,967 --> 00:56:11,599 cash. Though it's a given, you have to have some unencumbered cash 654 00:56:11,599 --> 00:56:34,507 in the structure. Now what I think is really interesting is you 655 00:48:46,444 --> 00:52:07,048 might then say, well, okay, I'll have an extra US$10 of unencumbered 656 00:52:07,048 --> 00:56:01,086 cash. That should keep me nice and safe and I can weather a pretty 657 00:56:01,086 --> 00:56:34,520 big drawdown. Now, that 20% unencumbered cash, if you hold it 658 00:47:26,452 --> 00:48:44,742 and you stimulate this S&P plus trend portfolio through the 659 00:48:44,742 --> 00:50:18,690 last 26, 27 years, that gets fully drawn down three times. So, 660 00:50:18,690 --> 00:52:16,125 you get truly drawn down in the GFC, in the dot com, and in COVID. 661 00:52:16,125 --> 00:53:42,244 So, your structure will function fine most of the time. And 662 00:53:42,244 --> 00:55:08,363 then in these big left tail events, well, you're most likely 663 00:55:08,363 --> 00:56:34,482 going to have to close out some of your beta position. Now, 664 00:46:25,362 --> 00:48:27,185 this is a really good question as well because, is that bad? Now, 665 00:48:27,185 --> 00:50:19,637 if you are an allocator, and you have benchmarks, and from one 666 00:50:19,637 --> 00:52:21,460 day you are 100% long the S&P like your benchmark, and then the 667 00:52:21,460 --> 00:54:32,654 next day you are 20% long the S&P because you've had to sell a 668 00:54:32,654 --> 00:56:15,735 lot of that position. You have now massive tracking error versus 669 00:56:15,735 --> 00:56:34,477 your benchmark. The second point is that, over the long run, 670 00:47:19,243 --> 00:48:43,579 equities have tended to go up. And after big crises they've tended 671 00:48:43,579 --> 00:50:28,999 to rebound. And so, one part that we look at is if you were selling 672 00:50:28,999 --> 00:52:07,391 in the early part of 2009 at a low in your equity component and 673 00:52:07,391 --> 00:53:31,727 you're knocked out of that position, you miss kind of 29%, 30% 674 00:53:31,727 --> 00:55:10,119 of the, of upside. So, it can be a pretty bad outcome. It's certainly 675 00:55:10,119 --> 00:56:34,455 an outcome, as risk managers, we do not want to happen. Right? 676 00:46:15,991 --> 00:47:46,193 We want it to be a very, very left tail, very far away probability. 677 00:47:46,193 --> 00:49:22,838 So, what we look at is we say, well, what is a cash buffer that 678 00:49:22,838 --> 00:50:40,154 starts to really make it very unlikely this will happen (now I 679 00:50:40,154 --> 00:52:03,913 say unlikely, not impossible)? And in the paper we end up with essentially 680 00:52:03,913 --> 00:53:21,229 40% cash to support your beta. So that's 10% margin, 30% unencumbered. 681 00:53:21,229 --> 00:54:19,216 And our approximate calculations mean that will get shut 682 00:54:19,216 --> 00:55:42,975 out, as in you'll have to re-gear that beta about once in every 683 00:55:42,975 --> 00:56:34,519 1 in 200 years. So, it's not impossible. Essentially, what we're 684 00:46:53,860 --> 00:49:25,336 saying is, if you look at market data back through time, look 685 00:49:25,336 --> 00:51:56,812 at live data, it survives. But obviously, you've only had one run 686 00:51:56,812 --> 00:54:28,288 of markets. Again, you talked about this at length last week. You've 687 00:54:28,288 --> 00:56:34,518 only had one path of the truth that actually happened. Now, we know 688 00:46:40,732 --> 00:48:11,449 you could get bigger drawdowns than that in the full distribution. 689 00:48:11,449 --> 00:49:42,166 And so, by running Monte Carlo, we're able to sieve these 690 00:49:42,166 --> 00:51:37,624 sort of severe drawdowns. And we think that kind of 1 in 100, 1 691 00:51:37,624 --> 00:53:16,588 in 200 year sort of time frame is reasonable. It's very unlikely 692 00:53:16,588 --> 00:54:39,058 but manageable. If it happens, you'll have less equity exposure, 693 00:54:39,058 --> 00:56:18,022 you can re-gear it afterwards. But that's sort of where we looked 694 00:56:18,022 --> 00:56:34,516 at that. Now, what I think is the interesting finding for listeners 695 00:47:45,631 --> 00:49:42,163 is if you're looking at say a 20% buffer or suddenly you're massively 696 00:49:42,163 --> 00:51:20,767 increasing the chance that you get knocked out… And some people 697 00:51:20,767 --> 00:52:59,371 might be fine with that. They might have callable capital available, 698 00:52:59,371 --> 00:54:37,975 they might have a credit facility, they might be happy to 699 00:54:37,975 --> 00:56:07,615 get on the phone with counterparties and post that margin 700 00:56:07,615 --> 00:56:34,507 at short notice. Now, a lot of investors that we work with do not 701 00:48:11,304 --> 00:50:06,617 want that. They want us to take on the operational complexity, 702 00:50:06,617 --> 00:52:01,930 the cash flow complexity of risk management, etc. They do not 703 00:52:01,930 --> 00:54:18,209 want to be getting margin calls on a Monday - please send cash 704 00:54:18,209 --> 00:56:34,488 by end of day. So, that was one finding on the cash buffers. 705 00:46:15,991 --> 00:47:50,481 Now the next bit that we essentially were also looking at, 706 00:47:50,481 --> 00:49:42,151 I mean, firstly you can look at the alphas that you hold. Like 707 00:49:42,151 --> 00:51:25,231 why does trend make a nice alpha and portable alpha? Well, it's 708 00:51:25,231 --> 00:52:51,131 super liquid. Right? If you have something that's like weekly 709 00:52:51,131 --> 00:54:08,441 dealing or monthly dealing, you can recapitalize quickly. So, 710 00:54:08,441 --> 00:56:00,111 it makes trend a very, very nice candidate for it. But you can 711 00:56:00,111 --> 00:56:34,471 of course use others. And the next one we really dug into, which 712 00:47:27,361 --> 00:49:10,451 again, it started coming up in my meetings, was how should I rebalance 713 00:49:10,451 --> 00:51:01,471 my alpha and my beta? Now, at first this sounds like the most simple 714 00:51:01,471 --> 00:52:12,841 question. Well, obviously just pick monthly, or pick quarterly. 715 00:52:12,841 --> 00:53:40,071 It can't be that difficult, right? That's how I rebalance my 716 00:53:40,071 --> 00:55:15,231 alts portfolios, how I balance my equities and bonds. So, then I 717 00:55:15,231 --> 00:56:34,531 would say, okay, so we're going to pick monthly rebalancing. 718 00:46:15,991 --> 00:47:44,341 Now, at year end, do you want the return of the S&P and your beta 719 00:47:44,341 --> 00:49:00,911 component? And they would say yes. And I would say, well, as soon 720 00:49:00,911 --> 00:49:59,811 as you select a monthly rebalance, you no longer replicate 721 00:49:59,811 --> 00:51:16,381 the S&P. And this is actually quite interesting new news to a lot 722 00:51:16,381 --> 00:52:32,951 of people. And when I first was looking at this, I was like, 723 00:52:32,951 --> 00:53:43,631 oh yeah, this makes complete sense, right? If you own the S&P, 724 00:53:43,631 --> 00:55:06,091 you don't buy it and sell it every month, you just own the index. 725 00:55:06,091 --> 00:56:28,551 If it goes to 110 and then it makes another 10%, it compounds on 726 00:56:28,551 --> 00:56:34,441 itself. If it goes to 110 and you sell US$10 of it to rebalance, 727 00:49:10,789 --> 00:51:52,141 you're no longer tracking the index. When you get to year end, 728 00:51:52,141 --> 00:54:33,493 you're going to get a surprise that, oh, my portable alpha beta 729 00:54:33,493 --> 00:56:34,507 component has lots of tracking error versus the index. So, we've 730 00:46:44,105 --> 00:49:18,732 run a number of different rebalancing rules. If you want to 731 00:49:18,732 --> 00:51:53,359 track the index as close as possible, you should actually buy 732 00:51:53,359 --> 00:54:56,100 and hold that index. So, maybe do an annual rebound or even longer, 733 00:54:56,100 --> 00:56:34,499 just hold it and hold your alpha. Now in the paper we say we 734 00:47:30,639 --> 00:49:27,943 wouldn't do this ourselves, because as you buy and hold, you're 735 00:49:27,943 --> 00:51:35,911 going to slowly get this drift. You might own too much equities 736 00:51:35,911 --> 00:53:43,879 or too little. If other investors come in, how do you get 737 00:53:43,879 --> 00:56:02,511 them to the nav, do you give them this non one-to-one mix that 738 00:56:02,511 --> 00:56:34,503 they're looking for? So, we think, essentially, if you can let 739 00:48:01,271 --> 00:50:26,031 that beta replicate closely, so rebalance a little bit less but 740 00:50:26,031 --> 00:52:50,791 have some sensible tolerance. So, let's say not going to rebalance, 741 00:52:50,791 --> 00:55:55,031 but if my S&P and SG trend goes more than 10% out of whack, 742 00:55:55,031 --> 00:56:34,511 I will rebalance. Now, it's all very good then talking through 743 00:47:22,855 --> 00:49:03,151 this. We put some numbers on it to look at specific episodes. 744 00:49:03,151 --> 00:50:35,089 So, something like 2008, it pays to rebalance very slowly because 745 00:50:35,089 --> 00:51:50,311 your equities are falling, your trend allocation is rising, 746 00:51:50,311 --> 00:53:22,249 every time you rebalance, you rebuy, you increase your equity allocation 747 00:53:22,249 --> 00:54:45,829 into a continually falling equity market. You sell trend that's 748 00:54:45,829 --> 00:56:17,767 rising. Essentially all rules, other than a buy and hold, underperform 749 00:56:17,767 --> 00:56:34,483 in 2008. Now, that's just one example, right? And it's one that 750 00:47:30,511 --> 00:48:52,483 everyone quotes, like everyone who is in this business will always 751 00:48:52,483 --> 00:49:59,551 talk about 2008 and rebalancing. But what about over 752 00:49:59,551 --> 00:51:36,427 the long run. And over the long run we found that, in performance 753 00:51:36,427 --> 00:52:58,399 terms, it kind of doesn't matter. So, whether you do monthly, 754 00:52:58,399 --> 00:54:12,919 quarterly, annually, whether you do 5%, 10% tolerance, it doesn't 755 00:54:12,919 --> 00:55:27,439 matter over the very long-term. But you will get periods 756 00:55:27,439 --> 00:56:34,507 where it really matters, like 2008, potentially like COVID. And 757 00:46:28,877 --> 00:49:16,395 so, all we wanted to do was just to highlight, look, but it's 758 00:49:16,395 --> 00:52:16,799 a good question to ask, just have a look at the paths. Are there 759 00:52:16,799 --> 00:54:38,545 any paths that you really wouldn't like in there? And therefore 760 00:54:38,545 --> 00:56:34,519 that might push you towards a certain rebalancing rule. So, we 761 00:46:55,895 --> 00:51:15,271 look at that. Again, it comes up in so many meetings, we wanted 762 00:51:15,271 --> 00:54:54,743 to put something out. We've internally done it with other assets, 763 00:54:54,743 --> 00:56:34,503 but here we're doing trend. And the final bit of the paper, which 764 00:48:26,207 --> 00:51:41,531 again, if you Google, should I use swaps or futures in portable 765 00:51:41,531 --> 00:54:24,301 alpha? There’s not a ton amount of information. Right? So, 766 00:54:24,301 --> 00:56:34,517 that's another one that we looked at. So, essentially kind of 767 00:46:42,883 --> 00:48:12,523 no free lunch, no arbitrage assumption, which it obviously holds 768 00:48:12,523 --> 00:50:26,983 in a market as big as this. But I just wanted to see it proved. 769 00:50:26,983 --> 00:52:23,515 Because you come to this, and you say, hang on, so my counterparty 770 00:52:23,515 --> 00:54:11,083 bank's going to charge me 40 basis points over offer. Why don't 771 00:54:11,083 --> 00:56:25,543 I just go and buy the future where I don't see the 40 basis points 772 00:56:25,543 --> 00:56:34,507 over? Whereas there's an explicit cost in the swap, it's implicit 773 00:47:25,481 --> 00:49:02,767 in the future. And if you run this over time, it washes out. Right? 774 00:49:02,767 --> 00:50:12,257 You don't get a structural performance advantage. So, what we 775 00:50:12,257 --> 00:51:21,747 described there is essentially it comes down to operational preferences. 776 00:51:21,747 --> 00:52:24,288 Internally, we've run features, we've run swaps. We make 777 00:52:24,288 --> 00:53:40,727 the case that it's probably down to investor preference and there's 778 00:53:40,727 --> 00:54:15,472 some operational considerations. Obviously, futures, 779 00:54:15,472 --> 00:55:18,013 you don't have the counterparty risk, that might be 780 00:55:18,013 --> 00:56:34,452 a consideration for some in a structure like this as well. So, 781 00:46:25,655 --> 00:48:31,287 we dive into that and that's kind of where we end. And really, 782 00:48:31,287 --> 00:50:17,591 the conclusion is that there isn't a completely right or wrong 783 00:50:17,591 --> 00:52:23,223 answer for any of these, but you should be talking with your providers 784 00:52:23,223 --> 00:53:59,863 about these questions. You should understand kind of what will 785 00:53:59,863 --> 00:56:15,159 happen if you draw down 20% in a month. It's important to know ahead 786 00:56:15,159 --> 00:56:34,487 of time. 787 00:56:34,691 --> 00:56:47,567 Yeah, I mean, that's great. But I also think it's important because, 788 00:56:47,567 --> 00:56:59,370 I mean, obviously you do touch on correlation. We, obviously, talked 789 00:56:59,370 --> 00:57:12,246 initially about why trend is such a great candidate. But as these 790 00:57:12,246 --> 00:57:24,049 strategies become more and more popular, we have seen some other 791 00:57:24,049 --> 00:57:35,852 combinations that may look great in a backtest, but where there 792 00:57:35,852 --> 00:57:47,655 might be some risks that haven't been exposed in the last 793 00:57:47,655 --> 00:57:57,312 couple of decades, partly because of correlations, partly because 794 00:57:57,312 --> 00:58:12,334 of other things. But, I mean, you did a great job in touching on 795 00:58:12,334 --> 00:58:15,553 all of that. Now, in consideration of time, I think we 796 00:56:40,331 --> 00:56:49,496 have maybe another sort of 10, 12 minutes left. And I know there's 797 00:56:49,496 --> 00:56:57,251 another paper that we came across that you really liked and 798 00:56:57,251 --> 00:57:05,711 it's about the mathematics of… Turtle Trading is the name of the 799 00:57:05,711 --> 00:57:13,466 paper written by Nicholas Polson and Vadim Sokolov. And I would 800 00:57:13,466 --> 00:57:24,041 not mind if you could give us a little bit of a flavor of what's 801 00:57:24,041 --> 00:57:33,206 in the paper, why you liked it. So, maybe we don't have time 802 00:57:33,206 --> 00:57:39,551 for a full deep-dive deconstruction of the paper, but 803 00:57:39,551 --> 00:57:48,716 maybe kind of focus on some of the highlights and then people can 804 00:57:48,716 --> 00:57:59,291 go and read the paper. I'm sure I'm going to link to it in one 805 00:57:59,291 --> 00:58:07,751 of my weekly emails. But I know you found it very interesting, 806 00:58:07,751 --> 00:58:15,506 so why don't you talk a little bit about that, Harry? 807 00:58:16,721 --> 00:58:36,711 Yeah, absolutely. And thank you for flagging this one, Niels. 808 00:58:36,711 --> 00:59:02,698 It was a superb read. I really enjoyed it. It did absolutely take 809 00:59:02,698 --> 00:59:32,683 me back 10 plus years now to my actuarial days. A lot of the concepts 810 00:59:32,683 --> 00:59:56,671 in there, like you're tested on in the fairly abstract, and it's 811 00:59:56,671 --> 01:00:14,662 lovely to kind of see them all unified here. But essentially what 812 00:58:20,696 --> 00:58:36,596 the paper is doing is it's saying the Turtle rules that were 813 00:58:36,596 --> 00:58:52,496 applied in the ‘80s and then published in the early 2000s, like 814 00:58:52,496 --> 00:59:08,396 these were kind of rules of thumb, hand calculable, pen and paper, 815 00:59:08,396 --> 00:59:22,971 easy to do in a few minutes approximations of later decision 816 00:59:22,971 --> 00:59:36,221 theory mathematics that was kind of published in academia. And 817 00:59:36,221 --> 00:59:52,121 what the paper is doing is it is linking those original rules, 818 00:59:52,121 --> 01:00:06,696 from the ‘80s, to academic papers starting from the ‘50s, but 819 01:00:06,696 --> 01:00:14,646 all the way up to now. And so, it's a beautiful combination of some 820 00:58:36,801 --> 00:58:54,371 history, some financial history, behavioral economics. I 821 00:58:54,371 --> 00:59:21,981 know you've talked about kind of the Turtles and the behavioral 822 00:59:21,981 --> 00:59:52,101 side, but here it's really relating that to what was later published 823 00:59:52,101 --> 01:00:14,691 in kind of decision theory, probability theory by authors. So, 824 00:58:18,142 --> 00:58:33,773 when I read this, I also opened the original Turtle rules 825 00:58:33,773 --> 00:58:49,404 and read those original Turtle rules as well. And obviously, the 826 00:58:49,404 --> 00:59:06,456 story there, a group of people with limited or no trading experience, 827 00:59:06,456 --> 00:59:26,350 given a set of rules, follow these rules, and you will be a profitable 828 00:59:26,350 --> 00:59:44,823 trader. And we know some of those came out and were very successful. 829 00:59:44,823 --> 01:00:04,717 But I mean this isn't the area to touch on. You're a much bigger 830 01:00:04,717 --> 01:00:14,664 expert on that area than me, Niels. So, let's actually jump into 831 00:58:34,596 --> 00:59:21,071 the concepts that they bring out in the paper. And the first one, 832 00:59:21,071 --> 01:00:03,971 which is a favorite of the audience and all of us practitioners 833 01:00:03,971 --> 01:00:14,696 is estimating volatility. 834 01:00:15,501 --> 01:00:15,901 Right. 835 01:00:15,901 --> 01:01:02,596 So, the Turtles have a property N which is essentially the 836 01:01:02,596 --> 01:01:45,046 volatility of a market. And they use a range-based estimator, 837 01:01:45,046 --> 01:02:31,741 very simple to update each day. It's essentially like an exponential 838 01:02:31,741 --> 01:03:14,191 decay, half-life of 14. The people calculating this with pen 839 01:03:14,191 --> 01:03:56,641 and paper were simply looking at open/close, high/low, and getting 840 01:03:56,641 --> 01:04:43,336 some volatility estimate for each market. So, that's what we would 841 01:04:43,336 --> 01:05:42,766 today see as our vol estimate in crude or our vol estimate in treasuries. 842 01:00:15,901 --> 01:01:18,986 Now, what's interesting is the paper then links that to published 843 01:01:18,986 --> 01:02:10,601 research. For example, Parkinson who says look, this is 844 01:02:10,601 --> 01:03:13,686 actually an efficient way of estimating vol. Their kind of pen 845 01:03:13,686 --> 01:04:33,976 and paper calcs that they were doing on the fly is backed up in 846 01:04:33,976 --> 01:05:42,796 academia as like, yep, this is a legitimate way to estimate vol. 847 01:00:15,901 --> 01:02:07,727 Now, where I think it starts to get more interesting is on, well, 848 01:02:07,727 --> 01:03:50,951 you now have vol estimates for your markets. How do you position 849 01:03:50,951 --> 01:05:42,777 size? Like, how do I take that information and risk weight my positions? 850 01:00:15,901 --> 01:00:59,256 And so, the original Turtles, they had this idea of a unit such 851 01:00:59,256 --> 01:01:32,606 that one unit of each commodity or financial asset they 852 01:01:32,606 --> 01:02:05,956 were trading would be equivalent in risk terms. You can 853 01:02:05,956 --> 01:02:52,646 imagine a unit of crude oil, while you might need a unit of crude 854 01:02:52,646 --> 01:03:32,666 oil versus a unit of 10-year US treasury, they're equal in risk 855 01:03:32,666 --> 01:04:19,356 terms even if you have to buy five times the dollar amount of the 856 01:04:19,356 --> 01:04:56,041 treasury. And so, really what this is this volatility based position 857 01:04:56,041 --> 01:05:26,056 sizing, right? So, it's risk-based position sizing done in 858 01:05:26,056 --> 01:05:42,731 the ‘80s by the Turtles. Now, where this then starts, and what 859 01:00:34,969 --> 01:01:13,105 the paper does, is it then relates this to a lot of the academics 860 01:01:13,105 --> 01:01:40,345 around these position sizes, optimal bet sizing. And so, they 861 01:01:40,345 --> 01:02:07,585 bring in the Kelly criteria which, again, you've discussed at 862 01:02:07,585 --> 01:02:37,549 great length. They quote Ed Thorp. And for your listeners, if 863 01:02:37,549 --> 01:03:15,685 anyone hasn't read his A Man for All Markets. I mean that is probably, 864 01:03:15,685 --> 01:03:45,649 I would say, the best description of how to implement this 865 01:03:45,649 --> 01:04:12,889 in practice. And essentially, it's saying that look, the optimal 866 01:04:12,889 --> 01:04:48,301 bet that you want to take to compound and grow wealth over the 867 01:04:48,301 --> 01:05:20,989 time, you need to take some account for the probability of going 868 01:05:20,989 --> 01:05:42,781 bust, like the ruin theory that you've discussed. And as you 869 01:00:28,006 --> 01:01:08,356 increase your bet size, you increase your expected return, but 870 01:01:08,356 --> 01:02:00,811 you're always adding this tail that you might hit. I mean what as 871 01:02:00,811 --> 01:02:49,231 actuaries would call ruin. I think we have ruin theory is very 872 01:02:49,231 --> 01:03:37,651 applicable here. And what the Turtles did is they had stops in 873 01:03:37,651 --> 01:04:22,036 based on the risk of that position. Through this quantity N 874 01:04:22,036 --> 01:05:14,491 which is your estimate of vol at 2N, you take your position away 875 01:05:14,491 --> 01:05:42,736 - you stop out of that position. And what the paper is saying 876 01:00:36,547 --> 01:01:17,839 is, well, look, this 2N equates to 2% of your total equity 877 01:01:17,839 --> 01:02:06,013 in your book, i.e. that stop is set that you can't lose more than 878 01:02:06,013 --> 01:02:50,746 2% of your equity. And it links it to the Kelly criterion - 879 01:02:50,746 --> 01:03:28,597 optimal Kelly. Now, what the authors describe and what they actually 880 01:03:28,597 --> 01:04:13,330 show, and they do attest with their own data later on, is this 881 01:04:13,330 --> 01:04:58,063 is quite far on the kind of Kelly hill. You have some optimal 882 01:04:58,063 --> 01:05:32,473 Kelly which is under no uncertainty and that maximizes your 883 01:05:32,473 --> 01:05:42,796 long-term compound growth. Now, any prudent manager will be 884 01:00:37,453 --> 01:01:27,741 on the left side of that hill. Right? You are taking a little bit 885 01:01:27,741 --> 01:02:14,437 less of the upside for a much lower probability of ruin. And what 886 01:02:14,437 --> 01:02:57,541 the paper shows is that these Turtle traders, back in the ‘80s, 887 01:02:57,541 --> 01:03:37,053 were running a version of that, a pretty prudent version of 888 01:03:37,053 --> 01:04:20,157 that where their risk of ruin was much much reduced. So again, 889 01:04:20,157 --> 01:04:59,669 for listeners, I think the Ed Thorp book describes it fantastically 890 01:04:59,669 --> 01:05:42,773 well. The paper also kind of links all these concepts really nicely. 891 01:00:15,901 --> 01:01:14,449 Now, the next thing they look at, and again very interesting for 892 01:01:14,449 --> 01:02:08,118 a modern day trend follower, is in that position sizing obviously 893 01:02:08,118 --> 01:03:11,545 it's not just volatility that impacts that. As we all know we should 894 01:03:11,545 --> 01:03:55,456 be considering things like liquidity. We should especially be 895 01:03:55,456 --> 01:04:39,367 considering the correlation between these markets. So, any modern 896 01:04:39,367 --> 01:05:23,278 risk position sizing will also include, obviously, information on 897 01:05:23,278 --> 01:05:42,794 the correlation between markets. Now, remember, back in the 898 01:00:33,666 --> 01:01:12,749 ‘80s these traders were using pen and paper to calculate. Right? 899 01:01:12,749 --> 01:01:41,173 They weren't inverting covariance matrices and they had 900 01:01:41,173 --> 01:02:27,362 to have rules of firm. And so, the way they dealt with correlations 901 01:02:27,362 --> 01:03:06,445 was not in the actual construction but it was on position 902 01:03:06,445 --> 01:03:41,975 size caps. They correlated markets, they would cap, and, again, 903 01:03:41,975 --> 01:04:24,611 what the paper does is it essentially just links this to academics, 904 01:04:24,611 --> 01:04:53,035 more modern decision and portfolio construction theory, which 905 01:04:53,035 --> 01:05:39,224 is just, it's beautiful that they were kind of doing it in the 906 01:05:39,224 --> 01:05:42,777 ‘80s. And the final point I would touch on which was the (I mean 907 01:03:29,068 --> 01:05:42,799 there's a few more in the interest of time)… 908 01:05:43,211 --> 01:05:43,611 Sure. 909 01:05:44,811 --> 01:06:04,424 … they also had drawdown rules. Right? So, like many managers 910 01:06:04,424 --> 01:06:24,037 will have this certainly multi strategy pods, you have these very 911 01:06:24,037 --> 01:06:41,867 strict drawdown rules. Now, for anyone kind of really interested 912 01:06:41,867 --> 01:07:05,046 in the maths of kind of Kelly and drawdown control, you know, under 913 01:07:05,046 --> 01:07:30,008 your Kelly, if you are left of the top of the hill you shouldn't 914 01:07:30,008 --> 01:07:49,621 go into ruin. Right? You've got a decent probability of avoiding 915 01:07:49,621 --> 01:07:51,404 ruin. The issue is, if you're down 80% in your strategy, you still 916 01:05:59,847 --> 01:06:14,883 know it's going to work. If it's your strategy, and you're following 917 01:06:14,883 --> 01:06:26,160 the rules, and you are indestructible in your confidence, 918 01:06:26,160 --> 01:06:39,943 you don't suffer from human biases, you implement that, you should 919 01:06:39,943 --> 01:06:54,979 be fine. The issue is, and in the Turtles case, obviously they're 920 01:06:54,979 --> 01:07:10,015 being given a notional book of someone else's equity. In a manager's 921 01:07:10,015 --> 01:07:21,292 case, you are running allocators money, you are running 922 01:07:21,292 --> 01:07:32,569 the money of pension scheme participants, teachers, etc., you 923 01:07:32,569 --> 01:07:45,099 cannot stomach an 80% drawdown. So, what is an optimum 924 01:07:45,099 --> 01:07:51,364 rule for mitigating those drawdowns? And so, the Turtles back 925 01:05:58,881 --> 01:06:35,463 in the ‘80s, I'm just trying to recall, I think they lost 10% 926 01:06:35,463 --> 01:07:12,045 in the market, then they would cut their equity by 20%. So, you're 927 01:07:12,045 --> 01:07:34,557 essentially approaching a limit where you essentially can't 928 01:07:34,557 --> 01:07:51,441 lose any more at some point. Now, what the paper does is it links 929 01:05:59,931 --> 01:06:22,611 this to the later published work of Grossman and Zhoe in the 930 01:06:22,611 --> 01:06:41,511 ‘90s on optimal kind of drawdown control metrics. And again, 931 01:06:41,511 --> 01:07:00,411 it's like a beautiful link that these hand-count rules, back 932 01:07:00,411 --> 01:07:24,981 in the ‘80s, that were given to the kind of regular people, you 933 01:07:24,981 --> 01:07:45,771 can actually express that in maths very, very nicely and people 934 01:07:45,771 --> 01:07:51,441 do that later. So yeah, I really enjoyed reading. It is a bit 935 01:06:16,471 --> 01:06:57,629 of a history lesson as well as some very nice tie-ins there. Again, 936 01:06:57,629 --> 01:07:32,455 Niels, you're probably the expert on the history more than me, 937 01:07:32,455 --> 01:07:51,451 but I did enjoy this one. 938 01:07:54,122 --> 01:08:03,092 That's very kind of you to take us through that and the links 939 01:08:03,092 --> 01:08:10,682 to sort of research that's been published later on. But you're 940 01:08:10,682 --> 01:08:19,652 right, I mean, I think there was a time where people thought maybe 941 01:08:19,652 --> 01:08:27,242 that because the Turtles, they were not the first trend followers, 942 01:08:27,242 --> 01:08:35,522 there were trend followers in the ‘70s as well, but they became 943 01:08:35,522 --> 01:08:44,492 the most known in a sense because of the books that was written 944 01:08:44,492 --> 01:08:52,082 about them and the narrative. We know how strong narratives can 945 01:08:52,082 --> 01:08:52,772 be. 946 01:08:55,422 --> 01:09:09,634 And I think there was certainly a period of time where 947 01:09:09,634 --> 01:09:22,554 maybe people mistook the clarity of rules as being something 948 01:09:22,554 --> 01:09:39,350 that was simple. But actually, the genius was not so much in buying 949 01:09:39,350 --> 01:09:56,146 100 day breakout. The genius was, as you point out, in the way 950 01:09:56,146 --> 01:10:11,650 they approach risk. And I also think that sometimes with trend following 951 01:10:11,650 --> 01:10:27,154 and CTAs in general. I think it's not understood well enough that 952 01:10:27,154 --> 01:10:40,074 we are probably risk managers first and foremost. Actually, the 953 01:10:40,074 --> 01:10:54,286 returns that we generate is not something we control, but we 954 01:10:54,286 --> 01:11:09,790 do control to large extent, the risk we take. So, I've always 955 01:11:09,790 --> 01:11:26,586 felt that we, as an industry, we came from a risk management first 956 01:11:26,586 --> 01:11:29,170 and foremost. So, yeah, sort of building a system that takes small 957 01:09:22,882 --> 01:09:42,104 losses, and scales intelligently, and remains intact 958 01:09:42,104 --> 01:10:04,072 when the outsized opportunity finally arrives, that's essentially 959 01:10:04,072 --> 01:10:39,770 what we've been doing for the past five, six decades in our industry. 960 01:10:39,770 --> 01:11:09,976 And it's nice to see these things being continued, talked about, 961 01:11:09,976 --> 01:11:29,198 and written about in all of that. So, really, really grateful 962 01:09:04,334 --> 01:09:31,070 for you doing all of this heavy lifting today, Harry. It was 963 01:09:31,070 --> 01:09:53,350 really good. And also getting your additional perspective and thoughts. 964 01:09:53,350 --> 01:10:22,314 Let me, actually, go back one step, before we round out in the 965 01:10:22,314 --> 01:10:46,822 next few minutes, because there's one thing I forgot to ask 966 01:10:46,822 --> 01:11:13,558 you about the portable alpha. I was so excited about hearing your 967 01:11:13,558 --> 01:11:29,154 run down of the portable alpha paper. I do come across people who 968 01:09:06,402 --> 01:09:28,362 are interested in the concept, but where they feel a little bit 969 01:09:28,362 --> 01:09:46,662 worried about adding leverage, which is essentially what the product 970 01:09:46,662 --> 01:10:06,792 does. It combines two return streams and it's more than US$100 971 01:10:06,792 --> 01:10:26,922 in total. It becomes maybe US$150, maybe US$200. So, in their 972 01:10:26,922 --> 01:10:45,222 view, it's adding leverage and it's replacing to unlevered investments, 973 01:10:45,222 --> 01:11:10,842 so to speak. I asked the same question of Rich and Dave last week, 974 01:11:10,842 --> 01:11:29,142 but I'd love to hear your kind of brief perspective. And that is, 975 01:09:00,978 --> 01:09:26,906 and I don't want to make it too much of a leading question, but 976 01:09:26,906 --> 01:09:49,130 I think it's difficult for me not to. Why should people perhaps 977 01:09:49,130 --> 01:10:11,354 be less worried about the fact that you're stacking these two return 978 01:10:11,354 --> 01:10:35,430 streams on top of each other? You're putting the alpha on top of 979 01:10:35,430 --> 01:10:59,506 the beta so it looks like it's a leveraged product? Why should people 980 01:10:59,506 --> 01:11:21,730 be less concerned about that when we're talking about CTA or trend 981 01:11:21,730 --> 01:11:29,138 being the alpha generator? 982 01:11:29,772 --> 01:11:41,132 Yeah, Niels, it's a great question. It's like the leverage 983 01:11:41,132 --> 01:11:57,036 does just push up a bit of a kind of aversion barrier in some 984 01:11:57,036 --> 01:12:11,804 investors and I mean rightly so in some areas of the market. I 985 01:12:11,804 --> 01:12:24,300 think Dave and Rich, last week, were talking about the kind 986 01:12:24,300 --> 01:12:37,932 of three times levered ETF and the asymmetry of the return profile, 987 01:12:37,932 --> 01:12:43,612 like leverage there is scary. The way I see leverage in the way 988 01:11:35,924 --> 01:11:46,690 kind of consider it is that it is a tool, it's a portfolio construction 989 01:11:46,690 --> 01:11:53,611 tool, it's an investment management tool. And before linking 990 01:11:53,611 --> 01:12:02,070 it to trend and portable alpha, let's just take a portfolio 991 01:12:02,070 --> 01:12:12,067 of equities and bonds. So, if you start with a portfolio of 50% 992 01:12:12,067 --> 01:12:21,295 equities and 50% bonds, what you find is that pretty much all 993 01:12:21,295 --> 01:12:30,523 the risk of that book is coming from the equity portion. 85% 994 01:12:30,523 --> 01:12:40,520 of the risk is equities, 15% is bonds, even though on paper it 995 01:12:40,520 --> 01:12:43,596 says 50% / 50%. Now, the optimal risk weight or the optimal 996 01:11:42,908 --> 01:12:02,612 Sharpe ratio weights of those assets is going to be more bonds 997 01:12:02,612 --> 01:12:22,316 than equities to get more risk balance. It might be 30% equities 998 01:12:22,316 --> 01:12:38,736 and 70% bonds. Now, nobody likes that portfolio because the 999 01:12:38,736 --> 01:12:43,662 return's gone down. 1000 01:12:44,072 --> 01:12:44,472 Sure. 1001 01:12:44,632 --> 01:13:32,719 The bonds are less risky, they have lower return. So, if you allow 1002 01:13:32,719 --> 01:14:13,408 leverage in that portfolio, maybe you hold your 70% bonds, you're 1003 01:14:13,408 --> 01:15:05,194 still able to hold 50%, 60%, 70% of equities, you get a very risk 1004 01:15:05,194 --> 01:15:49,582 balanced portfolio, you get a higher Sharpe than you would by having 1005 01:15:49,582 --> 01:16:00,679 the US$100 constraint. Now, in portable alpha, the alpha that you 1006 01:13:12,640 --> 01:13:51,151 port is very important to your overall outcomes and your probability 1007 01:13:51,151 --> 01:14:29,662 of having a bad experience. So, in your alpha component, you're 1008 01:14:29,662 --> 01:15:11,674 layering alpha onto 100% beta. So, you already have your 100% S&P, 1009 01:15:11,674 --> 01:15:46,684 you've got your 100% equities, now you're introducing an additional 1010 01:15:46,684 --> 01:16:00,688 alpha driver on top. So, when you look at that alpha driver, what 1011 01:13:00,526 --> 01:13:21,718 alpha driver do you want? You see, almost certainly, that you don't 1012 01:13:21,718 --> 01:13:44,676 want to be introducing more beta. That's a risk in the same direction 1013 01:13:44,676 --> 01:14:02,336 that is really pushing out your volatility, it's pushing out 1014 01:14:02,336 --> 01:14:25,294 your potential of a big left tail loss. We also need to really 1015 01:14:25,294 --> 01:14:42,954 investigate not just that it's uncorrelated, but that it's uncorrelated 1016 01:14:42,954 --> 01:15:04,146 in the left tail. And the reason this is important is there's 1017 01:15:04,146 --> 01:15:21,806 plenty of alpha strategies that are kind of liquidity provision 1018 01:15:21,806 --> 01:15:44,764 and they look very good in normal times but have a big left 1019 01:15:44,764 --> 01:16:00,658 tail. Short vol is the most extreme of those. You could even 1020 01:12:53,410 --> 01:13:25,596 think of something like merger arbitrage or, I mean, the treasury 1021 01:13:25,596 --> 01:14:00,708 basis trade, something where if you have a very, very big shock, 1022 01:14:00,708 --> 01:14:35,820 that little rule that was making you money, I mean all the 1023 01:14:35,820 --> 01:15:13,858 deals might break in merger arb or in the basis trade, the liquidity 1024 01:15:13,858 --> 01:15:51,896 dries up and you get really hit because you no longer have the 1025 01:15:51,896 --> 01:16:00,674 liquidity you need. So, a good thing to do is actually look at the 1026 01:13:14,167 --> 01:13:46,387 returns of your alpha source, look at your returns of your beta 1027 01:13:46,387 --> 01:14:21,292 source, not just the whole data set but the worst 5% of beta 1028 01:14:21,292 --> 01:14:48,142 returns, the worst equity periods, and check that you still 1029 01:14:48,142 --> 01:15:20,362 have no correlation. Now why does trend sit nicely in there? Is 1030 01:15:20,362 --> 01:15:52,582 because in those worst periods trend actually tends to have a negative 1031 01:15:52,582 --> 01:16:00,637 correlation to equities. In your real tough times, for your structure, 1032 01:13:21,976 --> 01:14:17,992 where your equities are really getting hit over a three, four month 1033 01:14:17,992 --> 01:14:55,336 period, you're down decent amount there. Trend historically 1034 01:14:55,336 --> 01:15:51,352 has had a very load negative correlation. So, it makes a very 1035 01:15:51,352 --> 01:16:00,688 good complement. And back to your original question on the leverage. 1036 01:13:09,832 --> 01:13:35,032 It's extremely cash efficient. You're not actually deploying many 1037 01:13:35,032 --> 01:14:08,632 dollars to get that exposure. And also, when there's a big turning 1038 01:14:08,632 --> 01:14:36,632 point in markets, what happens? Vol in markets goes right 1039 01:14:36,632 --> 01:15:07,432 up. Your trend positions come right down. Like we just discussed 1040 01:15:07,432 --> 01:15:41,032 in the Turtle paper, you scale with volatility, risk has gone up, 1041 01:15:41,032 --> 01:16:00,632 your position size has actually come down. So, where people 1042 01:12:52,270 --> 01:13:15,184 are worried about margin calls, you'd be especially worried 1043 01:13:15,184 --> 01:13:43,190 about this in like a liquidity provision strategy like the basis 1044 01:13:43,190 --> 01:14:06,104 trade or other structures, trend is actually reducing positions. 1045 01:14:06,104 --> 01:14:39,202 It needs less margin to run. So yeah, that's a great point on 1046 01:14:39,202 --> 01:15:17,392 leverage. I see it as a tool and certainly a tool to be well understood. 1047 01:15:17,392 --> 01:15:42,852 But if it's increasing diversification then my view is that 1048 01:15:42,852 --> 01:16:00,674 it can be used to reduce risk. 1049 01:16:01,172 --> 01:16:15,268 Yeah, fantastic. Good stuff. Harry, wonderful, really wonderful. 1050 01:16:15,268 --> 01:16:36,412 Thoroughly enjoyed it. I know the audience will as well. And for 1051 01:16:36,412 --> 01:16:59,318 those of you listening in, if you did, show Harry how much you 1052 01:16:59,318 --> 01:17:18,700 appreciate all the preparation he did by going to your favorite 1053 01:17:18,700 --> 01:17:38,082 podcast platform and leave a rating and review because it really 1054 01:17:38,082 --> 01:17:53,940 does help more people discover these conversations and concepts. 1055 01:16:01,172 --> 01:16:40,976 And of course, as mentioned, make sure you go to the insights 1056 01:16:40,976 --> 01:17:20,780 page on Man's website and find the paper that we discussed today 1057 01:17:20,780 --> 01:17:53,950 and there are lots of other great resources on that. So, that 1058 01:16:02,842 --> 01:16:13,697 will give you enough to read until you join us next week where 1059 01:16:13,697 --> 01:16:24,552 I will be joined by Andrew Beer and Tom Robel where we will, 1060 01:16:24,552 --> 01:16:32,067 I'm sure, continue our conversations about many of these 1061 01:16:32,067 --> 01:16:42,087 topics. If you do have a question, feel free to suggest your 1062 01:16:42,087 --> 01:16:54,612 topic and send it over to me and I'll do my best. The email is 1063 01:16:54,612 --> 01:17:00,457 info@toptradersunplugged.com which actually happens to be also 1064 01:17:00,457 --> 01:17:11,312 the email address you can use, if you leave a really nice review, 1065 01:17:11,312 --> 01:17:19,662 as mentioned in last week's conversation, where you can actually 1066 01:17:19,662 --> 01:17:32,187 win a TTU merch in terms of a vest, a really nice vest (I have 1067 01:17:32,187 --> 01:17:43,042 to say). You might have seen some people wearing it on the show. 1068 01:17:43,042 --> 01:17:53,897 I need to make sure you get one as well, Harry, of course. But 1069 01:16:03,051 --> 01:16:19,962 anyways, there are some instructions in last week's episode 1070 01:16:19,962 --> 01:16:48,147 at the very end. So go and check it out because at the end of 1071 01:16:48,147 --> 01:17:12,574 this month we will find a winner and some of the reviews we've 1072 01:17:12,574 --> 01:17:38,880 had so far are actually very good. So, you need to up the game 1073 01:17:38,880 --> 01:17:53,912 if you want to be considered for that. Anyways, from Harry and 1074 01:16:13,364 --> 01:16:49,940 me, thanks ever so much for listening. We look forward to being 1075 01:16:49,940 --> 01:17:29,564 back with you next week. And until next time, as usual, take care 1076 01:17:29,564 --> 01:17:53,948 of yourself and take care of each other. 1077 01:17:56,762 --> 01:19:29,148 Thanks for listening to Top Traders Unplugged. If you feel you 1078 01:18:02,038 --> 01:18:15,228 learned something of value from today's episode, the best way 1079 01:18:15,228 --> 01:18:32,375 to stay updated is to go on over to your favorite podcast platform 1080 01:18:32,375 --> 01:18:50,841 and follow the show so that you'll be sure to get all the new 1081 01:18:50,841 --> 01:19:04,031 episodes as they're released. We have some amazing guests lined 1082 01:19:04,031 --> 01:19:21,178 up for you and to ensure our show continues to grow, please leave 1083 01:19:21,178 --> 01:19:29,092 us an honest rating and review. It only takes a minute and 1084 01:18:18,080 --> 01:19:07,822 it's the best way to show us you love the podcast. We'll see you 1085 01:19:07,822 --> 01:19:29,140 next time on Top Traders Unplugged. This podcast expresses 1086 01:18:02,657 --> 01:18:28,202 the views of its hosts and the guests appearing on the podcast as 1087 01:18:28,202 --> 01:18:53,747 of the date of its recording, and such views are subject to change 1088 01:18:53,747 --> 01:19:15,362 without notice. Top Traders Unplugged do not have any duty or 1089 01:19:15,362 --> 01:19:29,117 obligation to update the information contained herein. Furthermore, 1090 01:17:59,015 --> 01:18:21,545 Top Traders Unplugged make no representation to its accuracy and 1091 01:18:21,545 --> 01:18:46,328 it shall not be assumed that past investment performance is an 1092 01:18:46,328 --> 01:19:08,858 indication of future results. Moreover, wherever there is a potential 1093 01:19:08,858 --> 01:19:29,135 for profit, there is also the possibility of loss. This content 1094 01:18:07,026 --> 01:18:53,214 is made available for educational purposes only and should 1095 01:18:53,214 --> 01:19:29,138 not be used for any other purpose. The information contained 1096 01:18:04,460 --> 01:18:32,686 in this podcast does not constitute and should not be construed 1097 01:18:32,686 --> 01:19:03,478 as investment advice or an offer to sell or a solicitation to 1098 01:19:03,478 --> 01:19:29,138 buy any securities or related financial instruments in any jurisdiction. 1099 01:17:56,762 --> 01:18:14,241 Certain information contained herein concerning economic trends 1100 01:18:14,241 --> 01:18:39,211 and performance is based on or derived from information provided 1101 01:18:39,211 --> 01:18:59,187 by independent third-party sources. Top Traders Unplugged may 1102 01:18:59,187 --> 01:19:24,157 believe that the sources from which such information are obtained 1103 01:19:24,157 --> 01:19:29,151 are reliable. However, Top Traders Unplugged cannot guarantee 1104 01:18:05,420 --> 01:18:18,407 the accuracy of such information and has not independently 1105 01:18:18,407 --> 01:18:29,951 verified the accuracy or completeness of such information 1106 01:18:29,951 --> 01:18:45,824 or the assumptions on which such information is based. This podcast, 1107 01:18:45,824 --> 01:18:58,811 including the information contained herein, may not be reproduced, 1108 01:18:58,811 --> 01:19:16,127 copied, republished or posted in whole or in part in any form, 1109 01:19:16,127 --> 01:19:29,114 without the prior written consent of Top Traders Unplugged.