1 00:00:01,040 --> 00:00:04,960 You're about to join Niels Kaastrup-Larsen on a raw and honest 2 00:00:04,960 --> 00:00:08,280 journey into the world of systematic investing and learn about 3 00:00:08,280 --> 00:00:11,840 the most dependable and consistent yet often overlooked investment 4 00:00:12,000 --> 00:00:16,000 strategy. Welcome to the Systematic Investor Series. 5 00:00:23,280 --> 00:00:28,846 Welcome and welcome back to this week's edition of the Systematic 6 00:00:28,846 --> 00:00:33,400 Investor series with Rob Carver and I, Niels Kaastrup-Larsen, 7 00:00:33,400 --> 00:00:39,472 where each week we take the pulse of the global markets through 8 00:00:39,472 --> 00:00:42,508 the lens of our rules-based investor. Rob,it is wonderful to 9 00:00:26,490 --> 00:00:34,194 be back with you this week. Don't think we've spoken since before 10 00:00:34,194 --> 00:00:42,540 the summer, actually. So how are you? How are things in the UK? 11 00:00:42,860 --> 00:00:47,284 Yeah, I've been on a very long summer break as I'm very lucky to 12 00:00:47,284 --> 00:00:52,024 be in a position to do so and enjoying what was a very nice sunny 13 00:00:52,024 --> 00:00:56,132 summer in Britain and also a bit of Europe that I visited. And 14 00:00:56,132 --> 00:00:59,608 it's nice and sunny today, which is good as we're obviously 15 00:00:59,608 --> 00:01:03,716 welcoming VIP visitors to the UK at the moment. But I won't mention 16 00:01:03,716 --> 00:01:08,140 the name of these people, but you know, but we like to be good 17 00:01:08,140 --> 00:01:10,984 hosts so it's nice that they've got good weather. 18 00:01:11,440 --> 00:01:15,604 Yeah, that is absolutely true. We’ve got a good lineup of topics 19 00:01:15,604 --> 00:01:19,768 as we always do, of course, and we got three questions coming 20 00:01:19,768 --> 00:01:24,973 in or that came in, I should say. And before we get into all of 21 00:01:24,973 --> 00:01:29,831 that, I was going to ask you, as I normally do, sort of what's 22 00:01:29,831 --> 00:01:33,995 been on your radar, and I imagine the last couple of weeks, 23 00:01:33,995 --> 00:01:36,077 not the whole summer holiday recap. 24 00:01:36,090 --> 00:01:52,925 Yeah, when I was a child at school, the first thing you always 25 00:01:52,925 --> 00:02:09,760 have to do, when you came back in September, was write an essay 26 00:02:09,760 --> 00:02:24,005 entitled What I Did in my Summer Holidays. And because I've 27 00:02:24,005 --> 00:02:39,545 got quite a vivid imagination, I always felt the ability to make 28 00:02:39,545 --> 00:02:53,790 stuff up, which I hope amused my teachers immensely, rather than 29 00:02:53,790 --> 00:03:05,445 just reading about, you know, the normal standard stuff. Yeah,it's 30 00:01:38,756 --> 00:01:53,419 actually an anniversary of an important event, and that is Lehman 31 00:01:53,419 --> 00:02:10,748 Day. So happy Lehman Day to those who celebrate – so, 17 years 32 00:02:10,748 --> 00:02:28,077 to a couple of days ago. It was the sort of financial crisis, 33 00:02:28,077 --> 00:02:46,739 and I find it astonishing that it's been 17 years. It makes me feel 34 00:02:46,739 --> 00:03:02,735 incredibly old to say that. It's still very fresh in my mind, 35 00:03:02,735 --> 00:03:05,401 but yeah. So,if you follow the Financial Times Alphaville blog 36 00:01:43,974 --> 00:01:55,362 (which is free by the way, unlike the rest of FT that's behind 37 00:01:55,362 --> 00:02:04,998 an extraordinarily expensive paywall, as I know by my cost), they 38 00:02:04,998 --> 00:02:16,386 did an article last week. And they've got all kinds of a treasure 39 00:02:16,386 --> 00:02:25,146 trove of historical documents that you can read about, including 40 00:02:25,146 --> 00:02:36,534 (I won't mention the name of the bank) a note from a buy-side 41 00:02:36,534 --> 00:02:45,294 analyst, sorry, a sell-side analyst saying that Lehman's is a 42 00:02:45,294 --> 00:02:54,930 great investment at this price and you should definitely buy shares 43 00:02:54,930 --> 00:03:05,442 in them, that was written about 10 days before they went bust. 44 00:03:06,250 --> 00:03:10,956 Yeah, well, let me be more precise, because I did look up the 45 00:03:10,956 --> 00:03:14,576 treasure trove, when you mentioned Lehman Brothers, and I 46 00:03:14,576 --> 00:03:17,834 will mention it's public knowledge. So, apparently, it's a 47 00:03:17,834 --> 00:03:21,816 note from Morgan Stanley, it's dated June 30th, and it initiates 48 00:03:21,816 --> 00:03:24,712 coverage of Lehman Brothers with an overweight recommendation 49 00:03:24,712 --> 00:03:29,056 that is iterated, apparently, on July 14 and August 27. Now the 50 00:03:29,056 --> 00:03:33,762 lead author (okay, here I will blank out the name, people can look 51 00:03:33,762 --> 00:03:35,572 it up if they want)… 52 00:03:35,630 --> 00:03:36,908 Yeah, let's not kick him down. 53 00:03:37,310 --> 00:03:43,547 He left Morgan Stanley in 2009 and joined the Congressional Oversight 54 00:03:43,547 --> 00:03:48,650 Panel for Market Monitoring, Regulation, and Top Management, and 55 00:03:48,650 --> 00:03:54,320 spent five years at the Treasury Department where he was 56 00:03:54,320 --> 00:03:58,856 First Executive Director of the Financial Stability Oversight 57 00:03:58,856 --> 00:03:59,423 Council. 58 00:03:59,450 --> 00:04:08,206 Well, I suppose you could argue that having made mistakes in 59 00:04:08,206 --> 00:04:16,166 the private sector, he said he'd have learned valuable lessons 60 00:04:16,166 --> 00:04:25,718 that then the public sector could take advantage of in his new 61 00:04:25,718 --> 00:04:33,678 job. Let's be charitable about this. Everyone deserves a second 62 00:04:33,678 --> 00:04:40,842 chance and everyone makes mistakes. So, these things happen. 63 00:03:59,450 --> 00:04:07,354 Yeah,they do say that if you're a hedge fund manager and you've 64 00:04:07,354 --> 00:04:14,042 lost a lot of money, then actually people are desperate to 65 00:04:14,042 --> 00:04:22,554 give you money. First of all, if people gave you a lot of money 66 00:04:22,554 --> 00:04:29,242 to begin with then you must have some credibility. But secondly, 67 00:04:29,242 --> 00:04:35,322 they're like, well, you must have learned something from that 68 00:04:35,322 --> 00:04:40,794 and hopefully it'll be much better in the future. 69 00:04:40,850 --> 00:04:44,370 Sometimes the world just seems upside down, frankly. It's a little 70 00:04:44,370 --> 00:04:47,570 bit like George Costanza in Seinfeld: the opposite, you know, 71 00:04:47,570 --> 00:04:52,050 kind of works. You lose money and then people line up to give you 72 00:04:52,050 --> 00:04:56,210 money. I mean, it doesn't work in our... Funny enough, in the CTA 73 00:04:56,210 --> 00:04:59,410 world, it doesn't seem to be the way things work. 74 00:04:59,580 --> 00:05:02,726 A sneak preview for some of the topics which I'm out with later 75 00:05:02,726 --> 00:05:05,630 is that will be one of the themes for today, definitely: should 76 00:05:05,630 --> 00:05:08,292 you give money to people after they lost money or not? 77 00:05:08,300 --> 00:05:09,620 Ah, okay, fair enough. 78 00:05:09,620 --> 00:05:10,140 So, yeah. 79 00:05:10,220 --> 00:06:07,928 Okay. All right. Well, I mean, on my radar, it's not that I spend 80 00:06:07,928 --> 00:06:53,270 a lot of time looking for sensational stories, but even in 81 00:06:53,270 --> 00:07:42,734 the not so surprising rate-cut we had yesterday from the Fed, the 82 00:07:42,734 --> 00:08:28,076 25 basis points that everybody was expecting, there are a couple 83 00:08:28,076 --> 00:09:01,052 of things I thought were kind of interesting. Oneof the big headlines, 84 00:05:28,255 --> 00:06:07,932 of course, was that the new Fed governor, Stephen Miran, who 85 00:06:07,932 --> 00:06:51,216 was just sworn in like, I think, the day before, he dissented 86 00:06:51,216 --> 00:07:38,107 the decision. He wanted to go for 50 basis points - no surprise 87 00:07:38,107 --> 00:08:21,391 there. And I think even Powell dodged a question about Fed independence 88 00:08:21,391 --> 00:08:57,461 during the press conference because of Miran's presence on the 89 00:08:57,461 --> 00:09:01,068 FOMC. Butone thing that actually also that was kind of interesting, 90 00:05:53,263 --> 00:06:44,132 and that is that Christopher Waller, who I think is one of the 91 00:06:44,132 --> 00:07:35,001 contenders to be the next Fed Chair, he did not descend. So, he 92 00:07:35,001 --> 00:08:21,957 actually was in line with the 25 basis points. Probably not something 93 00:08:21,957 --> 00:08:53,261 that, necessarily, the administration would have liked to 94 00:08:53,261 --> 00:09:01,087 have seen. Soanyways, it'll be interesting to follow. It didn't 95 00:06:07,937 --> 00:07:24,893 sound like, from what I've read, that Powell was lining up complete 96 00:07:24,893 --> 00:08:41,849 surrender with lots of rate cuts coming our way. So, we'll have 97 00:08:41,849 --> 00:09:01,088 to follow it. Theother thing, and this is not news but it's something 98 00:05:48,115 --> 00:06:29,455 that I've noticed lately, is that I find it very interesting that 99 00:06:29,455 --> 00:06:57,015 some of these large institutions, the BlackRocks, the 100 00:06:57,015 --> 00:07:28,020 Fidelities, have entered the CTA trend following space lately. 101 00:07:28,020 --> 00:08:09,360 And actually, Fidelity put out a paper, I think in June, written 102 00:08:09,360 --> 00:08:43,810 by Roberto Crouch and Sladja Carton, about beautiful managed futures 103 00:08:43,810 --> 00:09:01,100 as a powerful portfolio diversifier. Idon'tthink there 104 00:05:19,640 --> 00:05:45,545 was anything new. I will read you the conclusion though because 105 00:05:45,545 --> 00:06:06,740 they say in the paper, “managed futures strategies represent 106 00:06:06,740 --> 00:06:30,290 one of the most time tested, research backed, and structurally 107 00:06:30,290 --> 00:06:49,130 unique investment approaches available today. Their ability to 108 00:06:49,130 --> 00:07:10,325 dynamically respond to trends, maintain low correlation to traditional 109 00:07:10,325 --> 00:07:31,520 asset classes, and protect capital during crisis makes them 110 00:07:31,520 --> 00:07:52,715 a valuable component to both traditional and alternative investments. 111 00:07:52,715 --> 00:08:13,910 While they may not always outperform in every environment, 112 00:08:13,910 --> 00:08:32,750 we believe their consistent contribution to portfolio resilience, 113 00:08:32,750 --> 00:08:53,945 over time, justifies serious consideration by any investor seeking 114 00:08:53,945 --> 00:09:01,010 long-term risk-aware diversification.” Now,that's all 115 00:05:19,580 --> 00:05:53,900 good. There's nothing new in that. These are all the arguments 116 00:05:53,900 --> 00:06:34,460 that we would otherwise have put out in the last couple of decades. 117 00:06:34,460 --> 00:07:11,900 What's interesting to me is, why now? Why are these huge institutions 118 00:07:11,900 --> 00:07:49,340 coming out with their products now, because as they say in this 119 00:07:49,340 --> 00:08:29,900 article, this is not new, so they must have known about it. So 120 00:08:29,900 --> 00:09:01,100 why didn't they launch these products 10, 15 years ago? Now,I 121 00:05:19,274 --> 00:06:09,071 know this could be personnel driven. Maybe Roberto, I mean, I 122 00:06:09,071 --> 00:07:07,922 actually spoke to him at some point, so I know he's been working 123 00:07:07,922 --> 00:07:57,719 with trend for a while, and obviously not necessarily at Fidelity, 124 00:07:57,719 --> 00:08:47,516 but also people like BlackRock, and so on, and so forth. 125 00:08:47,516 --> 00:09:01,100 It's just interesting. Ithinkit's great news for the industry 126 00:05:41,004 --> 00:06:19,484 that some of these really heavyweights in our industry and 127 00:06:19,484 --> 00:06:54,116 gatekeepers of huge model portfolios (where most likely they're 128 00:06:54,116 --> 00:07:36,444 going to be using their own products) give people exposure to 129 00:07:36,444 --> 00:08:22,620 trend. So that's wonderful news. But it is kind of funny to 130 00:08:22,620 --> 00:09:01,100 me that it's all happening in 2025 and not sooner. 131 00:09:02,540 --> 00:09:05,444 Yeah, actually, I had that article flagged as one for us to 132 00:09:05,444 --> 00:09:07,864 potentially discuss, but I'd assumed, because it was quite old 133 00:09:07,864 --> 00:09:10,284 now, that someone else would already have brought it up. 134 00:09:10,620 --> 00:09:11,820 We haven't discussed it. 135 00:09:11,900 --> 00:09:37,436 Yeah, so I did actually look at it. I think you're right in saying 136 00:09:37,436 --> 00:09:59,324 that there's nothing really new there. Like, if you look at all 137 00:09:59,324 --> 00:10:21,212 the kind of graphs that they use, it's the very standard. It's 138 00:10:21,212 --> 00:10:44,924 almost like if you're someone who works for a CTA and your job 139 00:10:44,924 --> 00:11:10,460 is to sell CTAs, you've got a kind of classic deck that you use 140 00:11:10,460 --> 00:11:14,108 of graphs. So,one is the correlation matrix. They've got a 141 00:09:23,600 --> 00:09:36,600 correlation matrix in there, you know, to both equities, bonds, 142 00:09:36,600 --> 00:09:50,900 and also 60/40. One is the sort of selective performance where 143 00:09:50,900 --> 00:10:05,200 you say, well, what happened during this GFC, which we've been 144 00:10:05,200 --> 00:10:18,200 talking about? What happened in 2022, you know, what happened 145 00:10:18,200 --> 00:10:33,800 in… I think they've even gone back to 2002, the.com crisis, as 146 00:10:33,800 --> 00:10:48,100 well. Exactly, yeah, so, that's a classic kind of thing to 147 00:10:48,100 --> 00:11:03,700 show. In those situations, how did bonds do? How did equities do? 148 00:11:03,700 --> 00:11:14,100 How did 60/40 do? How did trend do? Andthey've also got the 149 00:09:23,650 --> 00:09:44,800 classic, you know, what happens to mean standard deviation 150 00:09:44,800 --> 00:10:13,000 drawdowns and Sharpe ratio when you add, say, I think this 10%? 151 00:10:13,000 --> 00:10:43,550 Yeah, when you put 10% of a 60/40 portfolio into trend, it does 152 00:10:43,550 --> 00:11:09,400 better. So yeah, it's the classic deck, and it's not particularly 153 00:11:09,400 --> 00:11:14,100 anything new. Butyou're right, the significance is, that 154 00:09:24,297 --> 00:09:45,549 this is Fidelity, like, one of the biggest asset managers in the 155 00:09:45,549 --> 00:10:03,259 world. You know, you've got BlackRock, you've got Vanguard, you've 156 00:10:03,259 --> 00:10:26,282 got Fidelity, and I guess, in terms of just sheer number of assets, 157 00:10:26,282 --> 00:10:47,534 Pimco is probably out there as well. So, we're talking about the 158 00:10:47,534 --> 00:11:07,015 top four, top five asset managers in the world and they're 159 00:11:07,015 --> 00:11:14,099 suddenly interested in this. Soyeah, it's curious and it may 160 00:09:33,838 --> 00:10:14,580 just be a weird coincidence, or it may be because (and we'll get 161 00:10:14,580 --> 00:10:52,188 to this) trend hasn't done so well recently and who knows, maybe 162 00:10:52,188 --> 00:11:14,126 this is buying the dip. We'll see. 163 00:11:14,460 --> 00:11:43,596 Yeah, absolutely. And for those who may have missed it, I think 164 00:11:43,596 --> 00:12:15,160 maybe it was not last week's episode but the week before, I did 165 00:12:15,160 --> 00:12:44,296 mentioned the paper that the firm I work with, Dunn Capital, produced 166 00:12:44,296 --> 00:13:13,432 and released. You can also find it in the latest Sunday newsletter 167 00:13:13,432 --> 00:13:40,140 that I publish, if you're subscribed to that. And I think, 168 00:13:40,140 --> 00:14:06,848 also, I put a link, and I think it was toptradersunplugged.com/5voltrend 169 00:14:06,848 --> 00:14:18,988 I think was the link. Butin any event, we didn't spend time, 170 00:11:37,900 --> 00:12:07,200 actually, repeating all these arguments. We just looked at the 171 00:12:07,200 --> 00:12:42,360 evidence and we compared it to all the major hedge fund strategies 172 00:12:42,360 --> 00:13:14,590 because we hadn't seen that being done before. So, if people 173 00:13:14,590 --> 00:13:46,820 are interested to see how our industry versus ‘hedge fund strategies’ 174 00:13:46,820 --> 00:14:19,050 have done, then they can look that up in the paper. We'vealready 175 00:11:22,150 --> 00:12:08,290 touched a little bit on managed futures, so let me just get 176 00:12:08,290 --> 00:12:50,585 into that now. My own trend barometer finished yesterday at 41. 177 00:12:50,585 --> 00:13:32,880 So, that's an improvement. Still not very strong, I have to 178 00:13:32,880 --> 00:14:19,020 say, but it's an improvement in the last 10 days or so. Andyou 179 00:11:18,908 --> 00:11:41,148 know, so far September… although the last couple of days 180 00:11:41,148 --> 00:12:10,060 leading into the Fed cut, and the Fed cut itself, has not been 181 00:12:10,060 --> 00:12:38,972 great. I see performance down a little bit the last couple of days, 182 00:12:38,972 --> 00:13:07,884 but it's still a positive month. And some of the kind of, how 183 00:13:07,884 --> 00:13:34,572 should I say, usual suspects from the last couple of months such 184 00:13:34,572 --> 00:13:59,036 as equity, such as gold, have been really pulling the performance 185 00:13:59,036 --> 00:14:19,052 forward and so we see a continuation of that. Interms of 186 00:11:21,213 --> 00:11:50,476 hard numbers, let me find them here, BTOP50 is, as of Tuesday this 187 00:11:50,476 --> 00:12:19,739 week, up 2.43% for the month, now down only 45 basis points for 188 00:12:19,739 --> 00:12:44,500 the year. So, a great comeback. SocGen CTA index up 2.85 189 00:12:44,500 --> 00:13:13,763 for the month, down only 3.62% for the year. SocGen Trend up almost 190 00:13:13,763 --> 00:13:40,775 4%, down only 4% and change for the year. And the Short-Term 191 00:13:40,775 --> 00:14:03,285 Traders Index, actually a good improvement, 2.13% for the month, 192 00:14:03,285 --> 00:14:19,042 down just 4.37% so far this year. Now,of course, the traditional 193 00:11:30,370 --> 00:12:05,372 world continues to deliver. MSCI world up another 2.2% in September, 194 00:12:05,372 --> 00:12:46,738 now up 16.6% for the year. The US Aggregate Bond index from S&P 195 00:12:46,738 --> 00:13:31,286 is up 1.25% for the month, up 6.25% for the year. And the S&P500 196 00:13:31,286 --> 00:14:12,652 Total Return index up 2.25% for the month, and up 13.29% so far 197 00:14:12,652 --> 00:14:19,016 this year. Notnecessarily specifically about you... Well, what 198 00:11:47,591 --> 00:12:49,120 are your takeaways from trend, looking at your own kind of way of 199 00:12:49,120 --> 00:13:41,183 dealing with these signals this year, over the summer? I imagine 200 00:13:41,183 --> 00:14:19,047 you've also seen some improvement over the summer. 201 00:14:19,860 --> 00:14:23,940 Yeah, actually, unfortunately my computer is temporarily broken. 202 00:14:23,940 --> 00:14:26,100 That’s fine, just talk about from what you remember, Rob. 203 00:14:26,100 --> 00:14:29,334 Yeah, it's fine, I can't give you P&L numbers. I can give you risk 204 00:14:29,334 --> 00:14:30,489 numbers though, so that's 10… 205 00:14:30,500 --> 00:14:32,900 Well, just give me your thoughts. 206 00:14:33,350 --> 00:14:44,802 Yeah, so, last time I checked I was kind of, you know, making some 207 00:14:44,802 --> 00:14:53,800 positive performance. I'm not sure whether I'm flat for the year 208 00:14:53,800 --> 00:15:04,434 as quite yet, but it's not too far off hopefully, yeah. So, I'm 209 00:15:04,434 --> 00:15:13,432 currently long equities, long FX, long bonds, a little bit short 210 00:15:13,432 --> 00:15:22,430 energies. So, that's kind of the overall pattern, and risk is 211 00:15:22,430 --> 00:15:29,792 running at, I guess, equivalent numbers to you really. 212 00:15:29,792 --> 00:15:42,062 So, it would be about 50 %or 60% of my kind of typical kind of 213 00:15:42,062 --> 00:15:50,242 risk level. So, still not putting on like massive positions. 214 00:15:50,242 --> 00:15:59,240 It doesn't look like there's still very clear trends out there. 215 00:14:33,350 --> 00:14:45,791 Itlooks like I'm short Bitcoin though, so that's nice to 216 00:14:45,791 --> 00:15:00,494 see. I do like it when my system agrees with my own prejudices 217 00:15:00,494 --> 00:15:14,066 and biases. So, that's a nice pleasant sight. But yeah, I mean 218 00:15:14,066 --> 00:15:26,507 it does feel like the market is… Although the traditional markets 219 00:15:26,507 --> 00:15:41,210 are doing very well, it does feel like there's an awful lot of 220 00:15:41,210 --> 00:15:55,913 bad news out there that isn't, to me, priced in. I mean, particularly 221 00:15:55,913 --> 00:15:59,320 in interest rates. Imean,in any other kind of universe the fact 222 00:14:56,580 --> 00:15:19,810 that Fed independence is being seriously threatened, as you mentioned, 223 00:15:19,810 --> 00:15:43,040 should be something to cause serious consternation. I mean, so 224 00:15:43,040 --> 00:15:59,301 where am I with bonds? Hang on… 225 00:15:59,720 --> 00:16:04,491 I think (while you've looked that up), and may I may have mentioned 226 00:16:04,491 --> 00:16:09,629 it last week, I think right now in the CTA space, bonds are where 227 00:16:09,629 --> 00:16:13,666 there might be some divergence in positioning because I see the 228 00:16:13,666 --> 00:16:18,070 daily moves between some of these ETFs and mutual funds and it 229 00:16:18,070 --> 00:16:22,474 almost feels like that when bonds go up, some managers do much 230 00:16:22,474 --> 00:16:27,612 better than others, and so on, and so forth. So, I think maybe bonds, 231 00:16:27,612 --> 00:16:32,016 people or models just disagree on. I think equities, and gold, and 232 00:16:32,016 --> 00:16:35,319 silver, and all that stuff, people are fully aligned. 233 00:16:35,380 --> 00:16:44,046 Yeah, I mean, one reason for that might be that carry is a much 234 00:16:44,046 --> 00:16:50,855 more significant factor in bond price movements than it is in 235 00:16:50,855 --> 00:16:58,283 equities. And so, the weight you have to carry will make quite 236 00:16:58,283 --> 00:17:04,473 a big difference to your positions versus trend. But yeah, 237 00:17:04,473 --> 00:17:12,520 for what it's worth, I'm long bonds. My risk is about half what 238 00:17:12,520 --> 00:17:19,948 is in equities. So, I'm long BTPs, which is the Italian 10-year, 239 00:17:19,948 --> 00:17:25,519 obviously. I'm long US 30-years. I'm long 10-year Canadian. 240 00:17:25,519 --> 00:17:32,947 So yeah, I basically have a sort of modest size but consistently 241 00:17:32,947 --> 00:17:35,423 long position in bonds. So,you know, if there are serious 242 00:16:42,513 --> 00:16:49,646 concerns about Fed independence and inflation, then 243 00:16:49,646 --> 00:17:02,893 unfortunately, if I'm right, in the sense of my kind of gut feeling 244 00:17:02,893 --> 00:17:17,159 that Trump is bad for the bond markets, and bad for the equity markets, 245 00:17:17,159 --> 00:17:27,349 then, at least in my systematic portfolio, I'm going to 246 00:17:27,349 --> 00:17:35,501 lose money. So, there you go, that's life. 247 00:17:36,981 --> 00:17:40,798 A fair point, but actually it's not just about one individual. 248 00:17:40,798 --> 00:17:43,921 I think, with the fiscal policies being implemented right 249 00:17:43,921 --> 00:17:47,738 now and everybody getting massive amounts of debt, I think 250 00:17:47,738 --> 00:17:51,208 you could probably argue that many bond markets probably should 251 00:17:51,208 --> 00:17:54,331 have a higher risk premia than what they have. 252 00:17:54,361 --> 00:17:57,865 Yeah, but I mean, the bond market selling off with debt concerns 253 00:17:57,865 --> 00:18:01,369 is the dog that's never barked, really, to be honest. I mean, 254 00:18:01,369 --> 00:18:04,581 apart from, you know, Liz Truss a few years ago, politicians 255 00:18:04,581 --> 00:18:06,917 in developed markets, generally speaking, get away with 256 00:18:06,917 --> 00:18:10,421 levels of debt that a few years ago people thought would be 257 00:18:10,421 --> 00:18:10,713 inconceivable. 258 00:18:10,961 --> 00:18:15,758 Yeah, you say that, fair enough. But in the UK we did hit 259 00:18:15,758 --> 00:18:20,186 a 27-year high yields only a couple of weeks ago, higher than 260 00:18:20,186 --> 00:18:24,614 when Liz Truss did. So, maybe someone in there is actually aligning 261 00:18:24,614 --> 00:18:28,304 with your thoughts that this is not a great idea. 262 00:18:28,341 --> 00:18:34,072 We also have persistently high inflation. I mean, we've sort of 263 00:18:34,072 --> 00:18:39,282 maintained quite a bit of inflation from 2022. Whereas, say, 264 00:18:39,282 --> 00:18:46,576 in the EU it's kind of mean reverted a fair bit. But yeah, I 265 00:18:46,576 --> 00:18:52,828 mean, if you look at the spread to bonds, or even actually 266 00:18:52,828 --> 00:18:58,559 PTPs, I mean, you're better off lending to the Italian government 267 00:18:58,559 --> 00:19:01,164 than the British government apparently. Butthat doesn't work 268 00:18:32,205 --> 00:18:43,797 with the debt story because Italian debt to GDP ratios are much 269 00:18:43,797 --> 00:18:57,321 worse than the UK. So, it can't just be a debt story, it must 270 00:18:57,321 --> 00:19:01,185 be an inflation story. 271 00:19:02,071 --> 00:19:05,129 And the whole inflation story, I mean depending on which country 272 00:19:05,129 --> 00:19:07,909 and how they measure inflation, it's very hard to get. 273 00:19:07,911 --> 00:19:11,401 Anyway, let's stop pretending to be macro economists and experts 274 00:19:11,401 --> 00:19:15,240 on this because people will think they've tuned into the wrong 275 00:19:15,240 --> 00:19:15,589 podcast. 276 00:19:15,831 --> 00:19:17,769 Let's move on to the questions. 277 00:19:17,771 --> 00:19:19,729 Let’s move on to things that we actually know about, yeah. 278 00:19:24,851 --> 00:19:29,895 Question number one, from John (it came in a few days ago), “Hey 279 00:19:29,895 --> 00:19:33,775 Rob, big reader and implementer of your work. My question 280 00:19:33,775 --> 00:19:36,491 is about adapting your continuous forecast-based position 281 00:19:36,491 --> 00:19:39,983 sizing for a cash-only portfolio like stocks or crypto, 282 00:19:39,983 --> 00:19:43,863 where total notional exposure cannot exceed 100% of capital, the 283 00:19:43,863 --> 00:19:46,967 formula correctly scales a position-based on forecast strength. 284 00:19:46,967 --> 00:19:51,623 A forecast of 20 would target double the exposure of an average 285 00:19:51,623 --> 00:19:56,279 forecast of 10. When the sum of these ideal positions across all 286 00:19:56,279 --> 00:20:00,159 instruments exceeds available capital, what's the best way to scale 287 00:20:00,159 --> 00:20:03,263 them down? Should all positions be reduced proportionally 288 00:20:03,263 --> 00:20:07,531 to meet the 100% capital limit? Or is there another method 289 00:20:07,531 --> 00:20:10,247 you would recommend for managing the excess?” 290 00:20:11,691 --> 00:20:14,373 This is a really good question, actually. It's extremely 291 00:20:14,373 --> 00:20:17,353 relevant because I'm currently writing a book about trading without 292 00:20:17,353 --> 00:20:20,929 leverage. So, there's my book plug for the day. Niels, you can 293 00:20:20,929 --> 00:20:22,419 take that off the list. 294 00:20:22,451 --> 00:20:23,211 Is that a promise? 295 00:20:23,371 --> 00:21:55,467 Yeah, (laughter) well, unless there's another question related 296 00:21:55,467 --> 00:23:04,539 to it, in which case, sorry. Soyeah. I mean, so obviously, if 297 00:20:37,651 --> 00:21:00,091 you can't increase your position size when you get more confident 298 00:21:00,091 --> 00:21:20,491 about your positions, then you've kind of have two options, 299 00:21:20,491 --> 00:21:44,971 really. So, one is to kind of just effectively cut everything in 300 00:21:44,971 --> 00:22:09,451 half. And what that would mean, in practice, is that your average 301 00:22:09,451 --> 00:22:33,931 investment would be 50% cash. On average, you'll be using 50% of 302 00:22:33,931 --> 00:22:56,371 your exposure. And then, if your forecast is really strong, you'd 303 00:22:56,371 --> 00:23:04,531 be up to 100%. Theproblem with that is that, well it depends 304 00:20:43,801 --> 00:21:08,771 on what you're trading. So, if you're trading very illiquid, risky 305 00:21:08,771 --> 00:21:29,201 emerging market stocks that have, say, an annualized standard 306 00:21:29,201 --> 00:21:54,171 deviation of like 50%, then actually trading with half of that, 307 00:21:54,171 --> 00:22:14,601 on average (which would be 25%, ignoring diversification effects), 308 00:22:14,601 --> 00:22:39,571 wouldn't necessarily be a bad thing. That would be a reasonable 309 00:22:39,571 --> 00:23:04,541 risk target to run, like 25%. That's what I run myself. Onthe 310 00:20:26,101 --> 00:20:41,116 other hand, if you're trading, say, like, liquid large-cap S&P 500 311 00:20:41,116 --> 00:20:56,131 stocks, which have an analyzed standard deviation of like 25%, 30%, 312 00:20:56,131 --> 00:21:13,876 then you're going down from, say, 25% to 12.5%, which is a very 313 00:21:13,876 --> 00:21:27,526 low-risk target, and is unlikely to be optimal. So basically, 314 00:21:27,526 --> 00:21:46,636 what you do in practice is you just have to say, well, I'm going 315 00:21:46,636 --> 00:22:04,381 to live with the fact that I can't use forecast strength in such 316 00:22:04,381 --> 00:22:20,761 a clean way to actually construct my signals. So, if my forecasts 317 00:22:20,761 --> 00:22:38,506 become really weak, then yes, I will cut my position. But if they 318 00:22:38,506 --> 00:22:53,521 get stronger than average, then I can't really do anything about 319 00:22:53,521 --> 00:23:04,441 it. I just have to live with it. So,what I would probably do 320 00:20:35,161 --> 00:20:58,741 is something like, rather than cutting… At the moment, I cut my 321 00:20:58,741 --> 00:21:22,321 forecasts off at 20%, which is twice the average. I would probably 322 00:21:22,321 --> 00:21:47,866 cut them off at 10%, which is the average. And what that would 323 00:21:47,866 --> 00:22:11,446 mean in practice is that, if you had instruments you were holding 324 00:22:11,446 --> 00:22:33,061 where you had less than the average forecast, then you'd have 325 00:22:33,061 --> 00:22:54,676 a smaller position in them. So, you basically would have cash 326 00:22:54,676 --> 00:23:04,501 creeping back into your portfolio. Thequestion is whether 327 00:20:38,719 --> 00:21:17,089 it's then optimal to reallocate that cash to other positions? 328 00:21:17,089 --> 00:21:59,296 Then it gets complicated. So, in theory, yes. In practice, unless 329 00:21:59,296 --> 00:22:37,666 it was a big effect, I probably wouldn't bother because 330 00:22:37,666 --> 00:23:04,539 it would get a bit too complicated. So,for more details… 331 00:23:05,181 --> 00:23:05,741 Yeah. 332 00:23:05,741 --> 00:23:06,541 Buy the book. 333 00:23:06,621 --> 00:23:08,301 There is a book out there somewhere. 334 00:23:08,301 --> 00:23:10,281 Yeah, there probably is another book, But specifically for 335 00:23:10,281 --> 00:23:12,481 this problem, I'll definitely answer that question for you, but 336 00:23:12,481 --> 00:23:14,681 you just have to wait a few months, I'm afraid. 337 00:23:14,781 --> 00:23:19,214 Okay, great. All right. Well, there is another question. This is 338 00:23:19,214 --> 00:23:23,647 from Absolute Gnosis, which I imagine is some kind of handle. 339 00:23:24,221 --> 00:23:26,141 I mean, it could be their real name. 340 00:23:26,671 --> 00:23:41,257 It could be their real name, and I will stress that I normally 341 00:23:41,257 --> 00:23:54,721 would only take questions that come from people with a real name. 342 00:23:54,721 --> 00:24:02,575 Just so people are aware of that. Anyways,the question is, “Is 343 00:23:31,396 --> 00:23:40,846 short/long symmetry optimal for a strategy in which the price 344 00:23:40,846 --> 00:23:51,241 of the traded instruments is measured in units, “ fiat currencies”, 345 00:23:51,241 --> 00:24:02,581 that are inflated brackets devalued at 8% to 14% year per year?” 346 00:23:26,671 --> 00:23:53,611 It'sa very specific question. Maybe you can make more sense of 347 00:23:53,611 --> 00:24:02,591 it than I can. 348 00:24:03,801 --> 00:25:08,061 Okay, so we obviously have someone who's a fan of crypto here 349 00:25:08,061 --> 00:26:06,966 because they're using the term ‘fiat currency’ and that term is 350 00:26:06,966 --> 00:27:00,516 never used except pejoratively by crypto people. Until about five 351 00:27:00,516 --> 00:28:04,776 years ago, I thought that fiat currency was the money you handed 352 00:28:04,776 --> 00:29:19,746 over when you went to buy a small Italian car. But no, when they 353 00:29:19,746 --> 00:30:02,586 say that, they mean traditional currencies like pounds, 354 00:30:02,586 --> 00:30:18,651 euros, dollars, etc. Now,I would be curious where they got this 355 00:24:46,506 --> 00:25:52,936 figure of 8% to 14% a year, because as far as I'm aware, there's 356 00:25:52,936 --> 00:26:35,641 no major economy where inflation's running at that level 357 00:26:35,641 --> 00:27:27,836 (when I say major, I mean developed market), or whether that 358 00:27:27,836 --> 00:28:15,286 is what they think is the long-run appreciation of bitcoin 359 00:28:15,286 --> 00:29:02,736 versus fiat currencies is. Again, I don't know. Bitcoin has 360 00:29:02,736 --> 00:30:04,421 gone up by more than that recently, so I don't know if figures 361 00:30:04,421 --> 00:30:18,656 come from there. Butanyway, so let's reframe the question to 362 00:24:26,345 --> 00:24:57,343 terms that make more sense, I think. And that's the following. 363 00:24:57,343 --> 00:25:25,523 Let's suppose that I'm trading an instrument that's not denominated 364 00:25:25,523 --> 00:26:02,157 in my home currency. So, for example, I could be a UK investor 365 00:26:02,157 --> 00:26:41,609 (as I am), I'm trading S&P 500 futures (which I am), which are denominated 366 00:26:41,609 --> 00:27:15,425 in US dollars. Now, let's also assume, for the moment, that I'm 367 00:27:15,425 --> 00:27:46,423 trading the cash instrument and not the future because it gets 368 00:27:46,423 --> 00:28:20,239 a bit different with futures. But let's just, for the moment, assume 369 00:28:20,239 --> 00:28:56,873 I want to buy a US listed ETF that's denominated in dollars with 370 00:28:56,873 --> 00:29:39,143 my British pounds. And I want to say, well, I want to do some kind 371 00:29:39,143 --> 00:30:15,777 of trend following on that thing. So, I need the price of that 372 00:30:15,777 --> 00:30:18,595 thing. Thequestion is, which price should I use? Should I use 373 00:25:21,593 --> 00:26:53,529 the dollar price? So, I just get the dollar price, and I plot 374 00:26:53,529 --> 00:28:39,609 it on a chart, and I've had a wiggly line, and I apply my moving 375 00:28:39,609 --> 00:29:57,401 averages, or my Bollinger bands, or whatever it is I'm using 376 00:29:57,401 --> 00:30:18,617 to measure trends. Orshould I convert that price into a pound sterling 377 00:25:15,881 --> 00:26:56,793 price? So I get the wiggly line, and then I multiply it by the 378 00:26:56,793 --> 00:28:23,289 GBP USD FX rate, to get a sterling price, which obviously is 379 00:28:23,289 --> 00:29:56,993 going to look different. So, to kind of return to the terms used 380 00:29:56,993 --> 00:30:18,617 in the question. Iffor example, the pound had depreciated 381 00:24:50,652 --> 00:26:10,968 against the dollar, then the trend on that ETF would look stronger, 382 00:26:10,968 --> 00:27:31,284 because it would have had a sort of systematic drift added to 383 00:27:31,284 --> 00:28:51,600 it, which is the depreciation of the currency effect on top of 384 00:28:51,600 --> 00:30:18,609 it. And that would make me more likely to want to buy that. 385 00:24:03,801 --> 00:25:39,351 So,the question is, should we do that? And the answer is, well, 386 00:25:39,351 --> 00:26:52,851 if you do that, you're essentially trend following two things: 387 00:26:52,851 --> 00:28:13,701 an exchange rate and a currency. And you're kind of munging 388 00:28:13,701 --> 00:29:34,551 them into one thing and you're trend following that thing together 389 00:29:34,551 --> 00:30:18,651 rather than trend following them separately. Anotherway, an 390 00:24:14,022 --> 00:24:54,906 equivalent way of doing it would be to trend follow the dollar 391 00:24:54,906 --> 00:25:22,162 price, but also then, simultaneously have another trend 392 00:25:22,162 --> 00:25:56,232 following model which looks at GBP USD. Now, generally speaking, 393 00:25:56,232 --> 00:26:26,895 diversification is better than less diversification. By adding those 394 00:26:26,895 --> 00:27:00,965 two things together and creating the pound version of the 395 00:27:00,965 --> 00:27:35,035 S&P ETF price, what you're essentially doing is removing the 396 00:27:35,035 --> 00:28:02,291 potential for diversifying across two instruments, a currency 397 00:28:02,291 --> 00:28:39,768 and a stock market index, and just having a single instrument. 398 00:28:39,768 --> 00:29:24,059 Now that will work if and only if trend following will work better 399 00:29:24,059 --> 00:30:01,536 on the combined thing than on the two things separately, which 400 00:30:01,536 --> 00:30:18,571 will have a diversification benefit. Now,I haven't checked this, 401 00:24:21,316 --> 00:25:03,352 but generally speaking, I'm a big believer of the fact that it's 402 00:25:03,352 --> 00:25:31,376 quite hard to pick out statistically significant differences 403 00:25:31,376 --> 00:25:59,400 in performance of trend following of different instruments. 404 00:25:59,400 --> 00:26:41,436 So, when people say, oh yes, everyone knows that X trends better 405 00:26:41,436 --> 00:27:23,472 than Y, my first response is okay, what's the evidence? And they'll 406 00:27:23,472 --> 00:28:16,017 say, oh look, well, look, here you go. And I say, well, if I put 407 00:28:16,017 --> 00:29:01,556 these two return distributions on top of each other and do a statistical 408 00:29:01,556 --> 00:29:40,089 test, I find there's no difference between them. So, go away 409 00:29:40,089 --> 00:30:18,622 and come back to me when you've got a better idea. Andyes, 410 00:24:14,989 --> 00:25:10,929 if you check 250 different instruments, then yeah, you will 411 00:25:10,929 --> 00:26:01,275 find a few that are statistically significant. But that's 412 00:26:01,275 --> 00:27:08,403 just because you've checked so many. Then that's just the way that 413 00:27:08,403 --> 00:28:09,937 statistics work. So, you always find one or two by chance. 414 00:28:09,937 --> 00:29:33,847 So as a rule, I would not do that. I would not combine those two 415 00:29:33,847 --> 00:30:18,599 things together. I try to trade them separately. Now,things 416 00:24:09,895 --> 00:24:43,412 are more complicated when you think about futures because in futures, 417 00:24:43,412 --> 00:25:29,117 if I'm trading and I want to buy an SPY ETF, say, I don't actually 418 00:25:29,117 --> 00:26:02,634 have to convert the entire notional value of my pound account 419 00:26:02,634 --> 00:26:42,245 into dollars to buy that. I just have to convert enough for the 420 00:26:42,245 --> 00:27:24,903 margin. And if the margin is, say, I don't know, 20% or 25% or 421 00:27:24,903 --> 00:27:55,373 whatever (which is kind of, roughly, an average futures margin 422 00:27:55,373 --> 00:28:31,937 tends to come in at that), then my exposure to the currency 423 00:28:31,937 --> 00:29:11,548 risk is only a quarter of what it would be otherwise. Of course, 424 00:29:11,548 --> 00:29:45,065 because I'm trading futures, I can then separately go and trade 425 00:29:45,065 --> 00:30:15,535 the currencies as a separate thing, and get the diversification 426 00:30:15,535 --> 00:30:18,582 benefit. So,if you reframe the question as, you know, should 427 00:25:45,111 --> 00:27:16,290 we currency convert price series before trend following? My 428 00:27:16,290 --> 00:29:17,862 answer is no. I don't think you should do that because you’re 429 00:29:17,862 --> 00:30:18,648 kind of muddling two things together. Andactually if you're 430 00:24:50,657 --> 00:27:11,225 also trading GBP USD as well as the currency converted SPY, the 431 00:27:11,225 --> 00:28:44,937 correlation between those two things will increase significantly 432 00:28:44,937 --> 00:30:18,649 and you lose all the diversification benefit, essentially. 433 00:24:03,801 --> 00:25:24,117 Ifyou map that back to the original question, again, I would 434 00:25:24,117 --> 00:26:31,047 say no. I wouldn't, for example, reconvert all prices into 435 00:26:31,047 --> 00:27:31,284 bitcoin prices before running trend following things over them. 436 00:27:31,284 --> 00:28:51,600 Although, as I said, the whole premise of the question, the number 437 00:28:51,600 --> 00:30:18,609 in it - the 14%, I don't really know where that comes from. 438 00:30:18,901 --> 00:30:24,801 That's fair, that's fair. All right, last question. We'll make 439 00:30:24,801 --> 00:30:31,881 it quick because we've got all your wonderful topics to get to. 440 00:30:31,881 --> 00:30:40,141 This is a question that came in from Lynn just a few minutes ago 441 00:30:40,141 --> 00:30:46,631 before we started recording. I'm going to read all the questions 442 00:30:46,631 --> 00:30:54,891 and then you can just quickly answer the ones that you may be able 443 00:30:54,891 --> 00:30:56,071 to answer. Lynnsimply writes, “I've tested a few strategies but 444 00:30:32,293 --> 00:30:50,149 they're not working well with crypto. Got any ideas?” I think that's 445 00:30:50,149 --> 00:30:56,101 a pretty open question. Thenthere is one, “Is your new book 446 00:30:48,677 --> 00:30:56,121 about cryptocurrencies?” “Areyou considering a themed focus 447 00:30:43,711 --> 00:30:56,116 on stock picking?” 448 00:30:58,361 --> 00:31:08,670 So, for the first question, I do know at least one person, who's 449 00:31:08,670 --> 00:31:18,979 an Australian guy, who's very active on X, who has made an awful 450 00:31:18,979 --> 00:31:26,116 lot of money using trend following on cryptocurrencies. And 451 00:31:26,116 --> 00:31:34,839 certainly, I've also, myself, have this sort of side hustle where 452 00:31:34,839 --> 00:31:45,941 I do a bit of consultancy for a crypto hedge fund. I've also tested 453 00:31:45,941 --> 00:31:52,285 trend following on cryptocurrencies and it seems to 454 00:31:52,285 --> 00:32:01,801 work fine. And this hedge fund, again, is doing very well trend 455 00:32:01,801 --> 00:32:03,387 following cryptocurrencies. So,trend following on cryptocurrency 456 00:31:13,141 --> 00:31:45,657 seems to work pretty well. There's no guarantee of course, about 457 00:31:45,657 --> 00:32:03,393 the future, but there we go. Yes,(sorry about the second book 458 00:31:04,271 --> 00:31:16,091 plug, Niels, but it was forced to me through the question) yeah, 459 00:31:16,091 --> 00:31:25,941 there is. The book I'm currently writing is about trading 460 00:31:25,941 --> 00:31:33,821 without leverage and that includes trading cryptocurrency without 461 00:31:33,821 --> 00:31:44,656 leverage. So, it's in there. And yeah, it's also about trading 462 00:31:44,656 --> 00:31:54,506 individual liquid equities, so you can call that stock picking, 463 00:31:54,506 --> 00:32:03,371 if you like, as well. So, there we go. 464 00:32:03,411 --> 00:32:04,051 Okay, cool. 465 00:32:04,051 --> 00:32:05,251 Something for everybody there. 466 00:32:06,051 --> 00:32:10,071 Now I'm excited about the next few topics that you brought along, 467 00:32:10,071 --> 00:32:13,756 although we kind of exchanged ideas fairly late this time around. 468 00:32:13,756 --> 00:32:18,446 So, I will leave it up to you to guide us through because there 469 00:32:18,446 --> 00:32:22,131 are a few different articles that you're sort of combining into 470 00:32:22,131 --> 00:32:26,821 a bigger theme that we can then relate back to also to our world 471 00:32:26,821 --> 00:32:30,841 of trend following. So, I really need to just give you the 472 00:32:30,841 --> 00:32:33,186 floor and we'll see where we go. 473 00:32:34,031 --> 00:33:37,115 Yeah, so it's always hard for me, coming back from summer holidays, 474 00:33:37,115 --> 00:34:29,685 because I've not thought about, you know, markets, or trend 475 00:34:29,685 --> 00:35:32,769 following, or anything for the best part of two months. It's quite 476 00:35:32,769 --> 00:36:41,110 hard for me to get my head back into, you know, like, what's 477 00:36:41,110 --> 00:37:44,194 going on. So, yeah, apologies for it all being a bit last-minute. 478 00:32:34,031 --> 00:33:17,893 But,yeah, so, there are a few articles that kind of piqued my fancy. 479 00:33:17,893 --> 00:34:04,888 And I realized there was a bit of a theme to them. And then there 480 00:34:04,888 --> 00:34:39,351 was something else that happened that kind of tied into it. 481 00:34:39,351 --> 00:35:16,947 Anyway, so, the first article I'm going to mention is on Jeffrey 482 00:35:16,947 --> 00:35:57,676 Ptak. So, I think that's how you pronounce it. It's Ptak is how 483 00:35:57,676 --> 00:36:38,405 it's written. I don't know if the P is silent. Okay, Jeffrey, apologies 484 00:36:38,405 --> 00:37:19,134 if Neil's or I have got it wrong, but we've said it differently 485 00:37:19,134 --> 00:37:44,198 so one of us bound to be correct. Thearticle is titled Wish 486 00:33:31,471 --> 00:36:12,303 I Was Making This Up. And the headline says it all. An ETF gains 487 00:36:12,303 --> 00:37:44,207 42% a year. Its investors still lose money. Nowit's perhaps 488 00:32:50,948 --> 00:33:58,616 not worth talking about the ETF itself in detail, for what it's 489 00:33:58,616 --> 00:34:55,006 worth, it's called the Yield Max Coin Option Income Strategy. 490 00:34:55,006 --> 00:36:08,313 So, that's going to be some kind of ETF that is selling volatility 491 00:36:08,313 --> 00:37:21,620 to gain income that may, in turn, have its own risks. But that's 492 00:37:21,620 --> 00:37:44,176 not really the point. Thepoint is the thing did make money, 493 00:33:49,223 --> 00:35:42,011 42%, jolly good. But the investors in it lost money. Now how, 494 00:35:42,011 --> 00:37:34,799 you may ask, is this possible? Let's think about a really simple 495 00:37:34,799 --> 00:37:44,198 example. Let'simagine a fund that does something really strange. 496 00:33:04,370 --> 00:33:44,822 For 364 days of the year, it loses 1%... Actually, that's not 497 00:33:44,822 --> 00:34:18,532 mathematically possible. Well it is, actually, if you have geometric 498 00:34:18,532 --> 00:34:58,984 losses. But let's change the math slightly. So, for 51 weeks of 499 00:34:58,984 --> 00:35:39,436 the year it loses 1%. So, it's down basically half pretty much. 500 00:35:39,436 --> 00:36:30,001 And on the last day of the year it makes 200%. So that's a doubling 501 00:36:30,001 --> 00:37:03,711 and then another doubling. The first doubling removes the original 502 00:37:03,711 --> 00:37:40,792 loss. So, basically over the year it's made 100%. Woohoo, brilliant, 503 00:37:40,792 --> 00:37:44,163 fantastic. Now,let's imagine three investors. The first investor 504 00:33:47,015 --> 00:35:54,737 is buy and hold. Buys the fund beginning of the year, holds it till 505 00:35:54,737 --> 00:37:44,213 the end of the year, makes 100%. They're a very happy person. 506 00:32:34,031 --> 00:33:19,516 Okay,the other two investors think that they're geniuses at market 507 00:33:19,516 --> 00:34:13,271 timing. So, the first one buys the fund at the beginning of the 508 00:34:13,271 --> 00:35:07,026 year. After about 40 weeks they just go, oh my God, this is 509 00:35:07,026 --> 00:35:56,646 ridiculous, I've lost 40% on this thing, why am I still holding 510 00:35:56,646 --> 00:36:37,996 onto this piece of absolute garbage? And they basically sell 511 00:36:37,996 --> 00:37:40,021 it and then just sit in cash the rest of the year. So, they lose 512 00:37:40,021 --> 00:37:44,156 40%. Thefinal investor has some kind of God given talent for 513 00:33:38,403 --> 00:35:00,331 timing, and they wait till the very last week of the year and they 514 00:35:00,331 --> 00:36:10,555 buy into the fund, they make, you know, couple of hundred percent, 515 00:36:10,555 --> 00:37:20,779 maybe lots, minus 1%. So, they make quadruple their money. So, they're 516 00:37:20,779 --> 00:37:44,187 a even happier person. So,that's three investors. And out 517 00:33:12,011 --> 00:34:21,641 of those three investors, only one of them has actually managed 518 00:34:21,641 --> 00:35:43,931 to match the return of the fund over the year. The other two 519 00:35:43,931 --> 00:36:53,561 have quite different returns. One's down 40%, the other one's up 520 00:36:53,561 --> 00:37:44,201 200%. So, we've got very different return profiles. So,obviously 521 00:32:43,723 --> 00:33:41,875 that's a kind of made up, an extreme example, but hopefully you 522 00:33:41,875 --> 00:34:40,027 can see that it's going to be impossible for any individual investor 523 00:34:40,027 --> 00:35:38,179 to actually earn the returns of a fund unless they literally buy 524 00:35:38,179 --> 00:36:41,177 and hold, and hold the whole time, and make no kind of withdrawals 525 00:36:41,177 --> 00:37:34,483 or additional income. All of the investors will receive a slightly 526 00:37:34,483 --> 00:37:44,175 different return. So,we'll call those returns the percentage 527 00:32:59,875 --> 00:33:40,487 return and the cash return. Okay, so percentage return is the 528 00:33:40,487 --> 00:34:24,791 headline amount the fund makes over the year and the cash return 529 00:34:24,791 --> 00:35:01,711 is the money that any individual investor makes in it. 530 00:35:01,711 --> 00:35:42,323 But we can actually create an aggregate cash return because we 531 00:35:42,323 --> 00:36:19,243 can basically look at all the individual investors together and 532 00:36:19,243 --> 00:37:03,547 add up all of their withdrawals, and so on, and so forth, 533 00:37:03,547 --> 00:37:44,159 and work out how much they actually made in that year. So,for 534 00:32:40,297 --> 00:33:14,760 example, you could hopefully see that if 99% of the investors 535 00:33:14,760 --> 00:34:01,755 in the fund were of type 2; in other words, they held for 40 weeks 536 00:34:01,755 --> 00:34:45,617 and then sold. And only 1% of the investors was type 3; in other 537 00:34:45,617 --> 00:35:20,080 words, they were the lucky person at the end. Then overall, 538 00:35:20,080 --> 00:35:57,676 the cash return on that fund for all investors aggregated is going 539 00:35:57,676 --> 00:36:44,671 to be of the order of minus 39%, is going to be minus 40% plus 540 00:36:44,671 --> 00:37:28,533 a little bit for the guy at the end who was really lucky. So, 541 00:37:28,533 --> 00:37:44,198 that's exactly what Jeffrey does. We'llstick to his first name 542 00:32:52,457 --> 00:33:32,380 because I think we can agree that it's a lot easier to pronounce 543 00:33:32,380 --> 00:34:09,232 Jeffrey. He looks at the annual reports and the kind of cash 544 00:34:09,232 --> 00:34:52,226 movements and looks at the net flows in and out. And the issue is 545 00:34:52,226 --> 00:35:29,078 that there are a lot more people are putting money in just 546 00:35:29,078 --> 00:36:05,930 before the thing goes down than are putting money in when it's 547 00:36:05,930 --> 00:36:42,782 about to go up, and vice versa. So, they're taking money out 548 00:36:42,782 --> 00:37:22,705 before gains and vice versa. So, yeah, as the net result of that 549 00:37:22,705 --> 00:37:44,202 is that they do not make 42%. 550 00:37:45,021 --> 00:37:49,661 Well, specifically, according to his accounting, he says that US$2.5 551 00:37:49,661 --> 00:37:54,765 billion dollars were coming in in net inflows. But the investors 552 00:37:54,765 --> 00:37:59,405 lost a combined US$35.5 million over the very same period. 553 00:37:59,405 --> 00:38:01,261 I mean, that's extraordinary. 554 00:38:01,341 --> 00:38:04,575 Yeah. I mean, so that's not a big percentage, but it is still a 555 00:38:04,575 --> 00:38:07,116 loss, you know. So, was it US$2.5 billion and US$35 million, 556 00:38:07,116 --> 00:38:07,809 did you say? 557 00:38:08,631 --> 00:38:09,511 That's what the number said. 558 00:38:09,511 --> 00:38:27,321 That's like a loss of 1.5%, isn't it? Yeah, so, yeah. So obviously, 559 00:38:27,321 --> 00:38:46,501 a 1.5% loss is very different from a, a 41%, 42% gain. And you 560 00:38:46,501 --> 00:39:02,941 know, this is going to be more problematic in funds whose returns 561 00:39:02,941 --> 00:39:22,121 are very, let's say, patchy or spiky or to be technical for a second, 562 00:39:22,121 --> 00:39:35,821 that exhibits strong negative or positive skew. You know, it's 563 00:39:35,821 --> 00:39:52,261 the fund I described earlier, they’re going to lose money for 51 564 00:39:52,261 --> 00:40:00,481 weeks and makes a massive gain. That'sa kind of extreme example 565 00:38:33,301 --> 00:39:12,951 of a positive skewed strategy. Whereas, something like trend following, 566 00:39:12,951 --> 00:40:00,531 for example, is a less extreme example of a positive skewed strategy. 567 00:38:09,511 --> 00:38:28,013 Sellingoption vol, as in the fund that Jeffrey describes, is, 568 00:38:28,013 --> 00:38:44,833 generally speaking, a negative skewed strategy. So, it's got the 569 00:38:44,833 --> 00:39:05,017 opposite pattern. That again, it's going to have a pattern of lots 570 00:39:05,017 --> 00:39:25,201 of small gains and then big losses. That's kind of, roughly speaking, 571 00:39:25,201 --> 00:39:43,703 what you expect to see. It's slightly different from that for 572 00:39:43,703 --> 00:40:00,523 this fund because it’s obviously not purely options selling. 573 00:38:09,511 --> 00:38:32,996 Butagain, if you're selling immediately before one of those big 574 00:38:32,996 --> 00:38:58,616 losses, you're going to do very well. But if you're buying before 575 00:38:58,616 --> 00:39:26,371 one of those big losses, on the back of what probably look like 576 00:39:26,371 --> 00:39:51,991 persistent gains in the past, you're going to do very badly. So 577 00:39:51,991 --> 00:40:00,531 that's the first article. So,a question; I think whenever 578 00:38:24,313 --> 00:38:46,516 you see an example of something happening, anecdote, as 579 00:38:46,516 --> 00:39:18,587 a kind of systematic person, as a quant, if you like, you should 580 00:39:18,587 --> 00:39:48,191 say, well, this is generally true. Or are the investors in this 581 00:39:48,191 --> 00:40:00,526 particular fund just uniquely stupid? So,there was a nice piece 582 00:38:42,817 --> 00:39:49,429 of work done by, I think it's Morningstar, Jeffrey Ptak. And it's 583 00:39:49,429 --> 00:40:00,531 Jeffrey again. 584 00:40:01,011 --> 00:40:04,371 It's the same guy. So now we know he works at Morningstar. 585 00:40:04,371 --> 00:40:52,461 We know he works at Morningstar. We still don't know 586 00:40:52,461 --> 00:41:50,169 how to pronounce his name, but we know he works at Morningstar, 587 00:41:50,169 --> 00:42:47,877 and we know that he has a thing about this difference between 588 00:42:47,877 --> 00:43:26,349 percentage return and cash return or investor return. So,in 589 00:40:12,139 --> 00:40:54,863 this piece of analysis, he looks at all US Investment funds. 590 00:40:54,863 --> 00:41:37,587 So, I guess that's basically mutual funds. It's not hedge funds, 591 00:41:37,587 --> 00:42:20,311 it's mutual funds. So, stuff that ordinary people buy and sell. 592 00:42:20,311 --> 00:43:06,919 And he looks at the difference between the investor return and the 593 00:43:06,919 --> 00:43:26,339 cash return over those funds. Andbasically, in this context, a 594 00:40:29,619 --> 00:41:07,491 negative number means that the investors underperformed the fund, 595 00:41:07,491 --> 00:41:57,987 if you like. So, the previous example, that number would be like 596 00:41:57,987 --> 00:42:35,859 minus 40 something percent, because it's obviously a massive 597 00:42:35,859 --> 00:43:17,939 difference. In my stylized example, it would be even bigger 598 00:43:17,939 --> 00:43:26,355 than that. So,the numbers here aren't as big, which is kind 599 00:40:41,771 --> 00:41:26,651 of what you'd expect because this is an average across the whole 600 00:41:26,651 --> 00:42:04,051 industry. And obviously, in some funds, people will do better 601 00:42:04,051 --> 00:42:45,191 than the funds because their timing happens to have been good. 602 00:42:45,191 --> 00:43:26,331 There's probably no skill here. It probably is just by luck. 603 00:40:04,371 --> 00:41:01,454 Youknow, there are so many funds out there. At least with some 604 00:41:01,454 --> 00:41:54,146 of them, the investors would have got lucky. So, it's an average. 605 00:41:54,146 --> 00:42:46,838 The numbers are going to be smaller than for any individual fund, 606 00:42:46,838 --> 00:43:26,357 but they're still pretty damn impressive, to be honest. So.if 607 00:40:07,913 --> 00:40:27,394 we look at the worst, for example, which is called equity, 608 00:40:27,394 --> 00:40:46,875 that's kind of, I guess, US equities, although there's a confusingly… 609 00:40:46,875 --> 00:41:04,585 Oh, sorry, sector equity, apologies. So, that would be something 610 00:41:04,585 --> 00:41:25,837 like a fund that has a thematic focus on say, oil firms, 611 00:41:25,837 --> 00:41:41,776 or tech, or something like that. There, the underperformance 612 00:41:41,776 --> 00:42:06,570 is 4.4% a year. Whilst it's not 40% a year, if you're going to 613 00:42:06,570 --> 00:42:27,822 underperform by 4.4% a year, and this is over 10 years, that's 614 00:42:27,822 --> 00:42:41,990 a pretty depressing long underperformance. Most people, in 615 00:42:41,990 --> 00:42:59,700 a situation where they were underperforming by an average of 616 00:42:59,700 --> 00:43:26,265 4.4% a year for 10 years, would not be in a job for much longer. 617 00:40:04,371 --> 00:40:44,762 Andthat's the biggest figure. But if you look at the whole universe 618 00:40:44,762 --> 00:41:22,046 of funds, the underperformance is 1.7% a year, which may not sound 619 00:41:22,046 --> 00:41:56,223 much, but again, over 10 years, that is a significant difference. 620 00:41:56,223 --> 00:42:36,614 And if you do a little maths, and see the effect on compounded 621 00:42:36,614 --> 00:43:07,684 returns over 10 years with the 1.7% underperformance, you're going 622 00:43:07,684 --> 00:43:26,326 to see a pretty substantial underperformance. AndI guess if 623 00:40:13,875 --> 00:40:44,763 I was to do some speculative thought on this, I think the reason 624 00:40:44,763 --> 00:41:10,899 why it's particularly bad in equity sectors is that's where I 625 00:41:10,899 --> 00:41:41,787 think people are going to be most affected by kind of news and 626 00:41:41,787 --> 00:42:10,299 emotion and stuff and things like, oh, the AI revolution is coming, 627 00:42:10,299 --> 00:42:43,563 let's put all our money in AI. And all the AI stocks have already 628 00:42:43,563 --> 00:43:21,579 gone up a lot and then they go down and it turns out to be spectacularly 629 00:43:21,579 --> 00:43:26,331 poor timing. Sogenerally speaking, people are quite poor at 630 00:40:23,603 --> 00:40:45,239 predicting. Well, they're quite poor at predicting the future, 631 00:40:45,239 --> 00:41:09,279 full stop. Otherwise, we'd all be, gazillionaires. But they're quite 632 00:41:09,279 --> 00:41:33,319 poor at predicting whether a particular fund or particular manager, 633 00:41:33,319 --> 00:42:06,975 if you like, is going to do better in the near future. So, they're 634 00:42:06,975 --> 00:42:38,227 pretty poor at timing; when to invest in things and when to take 635 00:42:38,227 --> 00:43:02,267 money out of things. That's the conclusion of Jeffrey's work, 636 00:43:02,267 --> 00:43:26,307 I would say. So, yeah. I found those articles interesting. 637 00:43:26,371 --> 00:43:32,797 Yeah, and it kind of ties into a third article from the FT where 638 00:43:32,797 --> 00:43:37,846 they are kind of rebottling an article that was originally posted 639 00:43:37,846 --> 00:43:41,977 in Investment Advisor Association, and I think it's called 640 00:43:41,977 --> 00:43:44,272 Debunking the Persistent Score Card. 641 00:43:44,371 --> 00:44:33,332 Yeah, I mean, this is interesting. I'd never heard of the 642 00:44:33,332 --> 00:45:17,842 Persistent Scorecard, and maybe that's because I'm not a regular 643 00:45:17,842 --> 00:46:02,352 reader of that particular, what would you call it, publication, 644 00:46:02,352 --> 00:46:20,156 I suppose. Publication, indeed. So,yeah, I guess the idea 645 00:43:55,627 --> 00:44:14,387 behind this is, the whole passive versus active debate, which 646 00:44:14,387 --> 00:44:38,775 is the debate in finance that will never die. And you could argue 647 00:44:38,775 --> 00:44:57,535 that passive has gradually been winning that debate. And if 648 00:44:57,535 --> 00:45:20,047 you just look at the numbers, the percentage of assets that are 649 00:45:20,047 --> 00:45:40,683 passively managed, or managed by indices, has just kind of gone 650 00:45:40,683 --> 00:46:03,195 up persistently over quite a long period of time now. So that 651 00:46:03,195 --> 00:46:20,079 seems like an argument that’s been pretty much won. Butone of 652 00:43:49,867 --> 00:44:13,683 the things related to that is, well, if you are going to invest 653 00:44:13,683 --> 00:44:35,667 in active managers, should you invest in ones that have done well 654 00:44:35,667 --> 00:44:53,987 recently? In other words, is performance something that is persistent? 655 00:44:53,987 --> 00:45:14,139 And that kind of relates to it. Because if manager performance 656 00:45:14,139 --> 00:45:32,459 is persistent or has some signal, there's two options. So, 657 00:45:32,459 --> 00:45:54,443 one is that good managers continue to do well. The other is 658 00:45:54,443 --> 00:46:12,763 that good managers mean revert, and become less good managers 659 00:46:12,763 --> 00:46:20,091 or even bad managers. So,to get technical for a second, if you 660 00:44:07,348 --> 00:44:37,984 look at the returns of a particular fund or a particular strategy, 661 00:44:37,984 --> 00:45:03,514 do their returns show positive auto correlation? In other words, 662 00:45:03,514 --> 00:45:29,044 good performance follows good performance. Or do they show negative 663 00:45:29,044 --> 00:45:49,468 autocorrelation? In which case, bad performance follows good 664 00:45:49,468 --> 00:46:17,551 performance? So, do managers have persistent skill or is it just 665 00:46:17,551 --> 00:46:20,104 luck? Like,you know, a manager does particularly well in 666 00:44:29,596 --> 00:45:29,896 one year, are they likely to continue doing well in the future? 667 00:45:29,896 --> 00:46:15,121 So, apparently that's what this for persistent scorecard talks 668 00:46:15,121 --> 00:46:20,146 about. Soyeah, it's quite interesting because it sort of brought 669 00:44:09,491 --> 00:44:44,659 it to my attention. It’s taken a while to go into the debate, but 670 00:44:44,659 --> 00:45:07,267 basically, someone did an analysis of this persistent scorecard 671 00:45:07,267 --> 00:45:42,435 and the FT sort of said, well, this is a very poor analysis, but 672 00:45:42,435 --> 00:46:07,555 actually ultimately it doesn't really kind of settle the argument 673 00:46:07,555 --> 00:46:20,115 one way or the other. So,I'd encourage people to read it because 674 00:44:25,915 --> 00:45:23,038 it’s a very interesting article, and I think it's a particularly 675 00:45:23,038 --> 00:46:20,161 hot topic as we will get to in a moment, indeed. 676 00:46:21,872 --> 00:46:26,108 Sure, all right, well, let's get to that moment now, because the 677 00:46:26,108 --> 00:46:30,697 last, or the second to last question (we'll see if we get to 678 00:46:30,697 --> 00:46:34,933 the last one). But the next topic was very cryptic. When you 679 00:46:34,933 --> 00:46:39,522 sent it to me, you said, oh, I've been having a really interesting 680 00:46:39,522 --> 00:46:42,699 conversation with a CTA manager. So, tell me more. 681 00:46:44,422 --> 00:48:50,314 Well, yeah, so earlier this week I had lunch with an old friend 682 00:48:50,314 --> 00:50:56,206 who's also a CTA manager, and I'm not going to mention their name 683 00:50:56,206 --> 00:53:02,098 because I've not asked them if I can do this. And their employer 684 00:53:02,098 --> 00:54:58,306 might not necessarily be happy about it anyway, but they know who 685 00:54:58,306 --> 00:55:17,674 they are. So,we sort of got to talking about the fact that performance 686 00:49:22,354 --> 00:52:13,447 has not been great. Obviously, with, as we said right at the top 687 00:52:13,447 --> 00:54:38,218 of the program, it's recovered a bit recently, but overall it's 688 00:54:38,218 --> 00:55:17,701 not been great. AndI knew performance was bad because normally, 689 00:47:45,710 --> 00:49:25,303 well, let's just put it this way, we split the bill. Which, you 690 00:49:25,303 --> 00:50:41,913 know, I think all things considered means that the economic 691 00:50:41,913 --> 00:52:06,184 situation in CTAs must be pretty bad. Whereas, you know, I 692 00:52:06,184 --> 00:53:30,455 suppose if things get really, really bad, then I'd be expected 693 00:53:30,455 --> 00:54:47,065 to pay for the whole thing. But unfortunately, things haven't 694 00:54:47,065 --> 00:55:17,709 gotten quite that bad. Anyway,we kind of got to this kind 695 00:47:26,750 --> 00:48:19,660 of eternal question, which is, you know, has everything changed 696 00:48:19,660 --> 00:49:23,152 or is this just a temporary period of poor performance? And I 697 00:49:23,152 --> 00:50:26,644 think this is, you know, every time there's an inevitable poor performance 698 00:50:26,644 --> 00:51:35,427 in CTAs, and for that matter, in any kind of strategy, equity value, 699 00:51:35,427 --> 00:52:38,919 for example, any strategy that will go through long periods of underperformance 700 00:52:38,919 --> 00:53:21,247 or negative performance, inevitably, at some point, people 701 00:53:21,247 --> 00:54:24,739 are going to be asking that question. And I'm sure, Niels, it's 702 00:54:24,739 --> 00:55:17,649 a question that you've been asked many times as well. So,I sort 703 00:47:04,681 --> 00:48:32,470 of sat down and I thought, well, you know, okay, let's be systematic, 704 00:48:32,470 --> 00:49:46,753 let's be quants about this. So ultimately, what’s the answer to 705 00:49:46,753 --> 00:51:01,036 the question, has something changed or not? Well, we can waffle 706 00:51:01,036 --> 00:52:08,566 and say, well, you know, Trump's changed everything, or more 707 00:52:08,566 --> 00:53:16,096 algo trading has changed everything. Or, you know, we could 708 00:53:16,096 --> 00:54:37,132 say that, oh, just over time, markets get more efficient and therefore 709 00:54:37,132 --> 00:55:17,650 there's less alpha to be collected. Youcan come up with all 710 00:47:21,082 --> 00:48:28,292 kinds of plausible reasons why things might have changed. But ultimately, 711 00:48:28,292 --> 00:49:59,942 if I was to say, let's just boil it down to a really simple problem. 712 00:49:59,942 --> 00:50:54,932 Let's suppose I'm an alternative allocator of assets and 713 00:50:54,932 --> 00:52:08,252 I've got two options for my portfolio. Let's return to that earlier 714 00:52:08,252 --> 00:53:21,572 paper, the Fidelity paper. My two options are I've got a 60/40 715 00:53:21,572 --> 00:54:22,672 portfolio, and I've got potential to put some trend following 716 00:54:22,672 --> 00:55:17,662 in there. How should I do that portfolio optimization? Theone answer 717 00:46:56,452 --> 00:47:48,582 is to say, well, you should just take all the data you have, 718 00:47:48,582 --> 00:48:44,722 25, 30, 40… I don't know how long the trend index has been going, 719 00:48:44,722 --> 00:49:28,832 but obviously it's possible to do something like (oh my goodness, 720 00:49:28,832 --> 00:50:12,942 I can't believe I'm saying this), you could do a replication 721 00:50:12,942 --> 00:50:57,052 of the trend index, which is something I don't normally advocate. 722 00:50:57,052 --> 00:51:41,162 But if you wanted, for example, to create a backfill trend 723 00:51:41,162 --> 00:52:29,282 index that goes back much further, then you can do that by 724 00:52:29,282 --> 00:53:05,372 replicating the trend index and then using underlying futures 725 00:53:05,372 --> 00:54:01,512 prices. So, it's not a hard thing to do, or you could just create 726 00:54:01,512 --> 00:54:45,622 your own trend index. It doesn't really matter, to be honest, 727 00:54:45,622 --> 00:55:17,702 as long as you've got something that's investable. Let'ssay 728 00:47:08,864 --> 00:49:47,737 you've got 50 years. My own backtest goes back to 1970, so let's 729 00:49:47,737 --> 00:52:38,831 call it around 50 years. It's got 50 years of data. You just use 730 00:52:38,831 --> 00:55:17,704 the whole of the data, and you treat all of that data equally. So,you 731 00:47:06,264 --> 00:49:17,316 basically say, I think that, you know, 2025 is just like 1975, 732 00:49:17,316 --> 00:51:28,368 it's just like 1985, just like 1995. So, I'm saying things haven't 733 00:51:28,368 --> 00:53:50,341 changed. I'm saying that to do any kind of optimization I want to 734 00:53:50,341 --> 00:55:17,709 use as much data as I possibly can. So,I'm just going to use all 735 00:48:21,526 --> 00:51:35,734 of my data and I'm going to treat it all equally. So that's kind 736 00:51:35,734 --> 00:54:22,198 of the starting point. The question is whether that can be beaten 737 00:54:22,198 --> 00:55:17,686 by doing something else. So,what kinds of things might you 738 00:47:33,639 --> 00:48:58,011 do if you genuinely think that the world has changed? Well, one 739 00:48:58,011 --> 00:50:29,414 thing you could do of course is arbitrarily say, oh, I think the 740 00:50:29,414 --> 00:52:00,817 world changed three years ago and I'm just going to use the last 741 00:52:00,817 --> 00:53:32,220 three years of data, or one year ago, or some period of time 742 00:53:32,220 --> 00:55:03,623 and just say, right, I'm going to use that, that's what I'm going 743 00:55:03,623 --> 00:55:17,712 to do. Ithinkthe world's changed, and if I use, say, if we 744 00:48:22,186 --> 00:49:51,803 go back say 2.5 years, and obviously we'll miss the outperformance 745 00:49:51,803 --> 00:51:29,567 of 2022, and that means your CTA allocation would go down significantly. 746 00:51:29,567 --> 00:52:51,037 Depending on your exact methodology it might go to zero, 747 00:52:51,037 --> 00:54:45,095 but certainly it will be a lot lower than it would be using the 748 00:54:45,095 --> 00:55:17,683 full 50 year span. Andyou kind of smugly say, well you know, 749 00:47:17,893 --> 00:47:58,802 I'm not putting any money into trend following because I've done 750 00:47:58,802 --> 00:48:35,992 this incredibly sophisticated analysis, and this is what I've done, 751 00:48:35,992 --> 00:49:20,620 and this is great. Now, the problem with that is you've done 752 00:49:20,620 --> 00:49:57,810 something that is not back testable because you have essentially 753 00:49:57,810 --> 00:50:42,438 created an arbitrary point. I mean it doesn't have to be arbitrary. 754 00:50:42,438 --> 00:51:30,785 You could come up with some logic for why you've chosen 2.5 years, 755 00:51:30,785 --> 00:52:15,413 or 3 years, or whatever. But you've been able to do something 756 00:52:15,413 --> 00:53:07,479 that you couldn't have done 3 years ago, which is to know 3 years 757 00:53:07,479 --> 00:53:44,669 ago that trend following performance in the next not quite 758 00:53:44,669 --> 00:54:36,735 3 years, but 2, 2.5 years was going to be very poor. So, you've 759 00:54:36,735 --> 00:55:10,206 effectively kind of polluted your so-called rigorous testing with 760 00:55:10,206 --> 00:55:17,644 future information. So,you actually need to do something that 761 00:47:50,646 --> 00:49:38,260 could be back testable. So, you need to have some way, a testable 762 00:49:38,260 --> 00:51:01,040 way of identifying when the performance of a particular strategy, 763 00:51:01,040 --> 00:52:40,376 whether it be 60/40 or trend following, or any other hedge fund 764 00:52:40,376 --> 00:54:27,990 strategy, is likely to go down in the future. And as we've discussed, 765 00:54:27,990 --> 00:55:17,658 people are really bad at that. Andif people are really, really, 766 00:47:35,746 --> 00:49:26,948 really bad at that, it means it's probably going to be quite hard 767 00:49:26,948 --> 00:51:01,042 to find any kind of algorithm that's truly in-sample, that's not 768 00:51:01,042 --> 00:52:09,474 being polluted with future information, or conditioning variable, 769 00:52:09,474 --> 00:53:43,568 or something like that, that's actually going to say, well, your 770 00:53:43,568 --> 00:55:17,662 performance in the future is going to be better or worse. Andif 771 00:47:02,432 --> 00:48:23,477 you do look at things like persistence, and autocorrelation, 772 00:48:23,477 --> 00:49:53,527 of things like that, generally speaking the effects aren't especially 773 00:49:53,527 --> 00:51:32,582 strong. And you know, it's really, really hard to say, well, 774 00:51:32,582 --> 00:52:44,622 yes, given recent CTA performance, what CTA performance 775 00:52:44,622 --> 00:54:41,687 is likely to be over, say, the next 12 months. It's really, really 776 00:54:41,687 --> 00:55:17,707 hard to do that. So,the other thing you might do is say, well you 777 00:47:47,859 --> 00:48:57,063 know what, I'm just going to take a much more simplistic approach 778 00:48:57,063 --> 00:50:12,034 which is to say, rather than using the last 50 years of data, 779 00:50:12,034 --> 00:51:27,005 I'm only ever going to fit using the last 10 years of data, 780 00:51:27,005 --> 00:52:59,277 or the last 5 years of data, or the last 3 years of data, or the 781 00:52:59,277 --> 00:54:14,248 last 2 years of data. And that's sort of better from the kind 782 00:54:14,248 --> 00:55:17,685 of point of view you're no longer cheating with future information. 783 00:46:44,422 --> 00:47:39,802 Theproblem is if you do that, what you will find is that your performance 784 00:47:39,802 --> 00:48:16,722 will be really bad because, generally speaking, the more data 785 00:48:16,722 --> 00:49:01,026 you have the better it is for fitting an optimization. And there's 786 00:49:01,026 --> 00:49:37,946 a degradation of out-of-sample performance as you reduce the size 787 00:49:37,946 --> 00:50:29,634 of the data you have available to you. So, you can imagine if you 788 00:50:29,634 --> 00:51:13,938 just are really silly, and do an optimization based on the last 789 00:51:13,938 --> 00:52:05,626 10 days, you can see how that would be kind of crazy, even ignoring 790 00:52:05,626 --> 00:52:38,854 trading costs, because obviously you're going to be pulling 791 00:52:38,854 --> 00:53:26,850 your money in and out of funds nonstop, and putting it in 60/40, 792 00:53:26,850 --> 00:54:03,770 and vice versa. Even ignoring trading costs, you're just going 793 00:54:03,770 --> 00:54:51,766 to have mostly noise. It's just going to be, at best, it's going 794 00:54:51,766 --> 00:55:17,610 to be something just doing really badly. So,there's a degradation 795 00:47:03,430 --> 00:47:50,950 of performance and it's sort of nonlinear. So basically, using 796 00:47:50,950 --> 00:48:52,726 50 years of data is good. It doesn't make a lot of difference 797 00:48:52,726 --> 00:50:08,758 if you use 40 or even 30 years of data, or even 20 years of data. 798 00:50:08,758 --> 00:51:05,782 But when you're getting down to the point where you’re using the 799 00:51:05,782 --> 00:52:02,806 last two or three years of data because you think that something 800 00:52:02,806 --> 00:53:04,582 has changed in the last three years, and you test that, you find 801 00:53:04,582 --> 00:54:06,358 that it's not going to do anywhere near as well as just sticking 802 00:54:06,358 --> 00:55:03,382 to your guns and essentially assuming you can use all the data 803 00:55:03,382 --> 00:55:17,638 that you can. Sothat's kind of the theoretical answer to the 804 00:49:23,713 --> 00:52:56,101 question. And I'd hoped, after saying that, they would pick up the 805 00:52:56,101 --> 00:55:17,693 bill, but they didn't. But anyway, that's life. 806 00:55:19,472 --> 00:56:27,188 So, I'm curious, I obviously didn't know how long the lunch was, 807 00:56:27,188 --> 00:57:29,261 but I'm curious, did you conclude anything about has the world 808 00:57:29,261 --> 00:57:34,904 changed? Butlet me preface this with the fact that what surprises 809 00:55:33,134 --> 00:55:49,280 me is, when I hear this argument, which I do, as you rightly 810 00:55:49,280 --> 00:56:05,426 point out, I do that a lot, over the years some things changed, 811 00:56:05,426 --> 00:56:17,846 and usually it’s when performance is less than the average 812 00:56:17,846 --> 00:56:35,234 and all of that stuff. But my answer to that is, well, the world 813 00:56:35,234 --> 00:56:51,380 is always changing and if you look at, say, the company that I 814 00:56:51,380 --> 00:57:05,042 work with, we've been doing trend following for 51 years. Can 815 00:57:05,042 --> 00:57:21,188 you imagine going back to 1974 and think about how much has changed 816 00:57:21,188 --> 00:57:34,850 in that period of time? Yet the strategy continues to work. Now,clearly 817 00:55:26,600 --> 00:56:12,932 you need to adapt, and you need to do research. So, in that 818 00:56:12,932 --> 00:56:55,700 sense, I don't subscribe to the fact that you can just apply 819 00:56:55,700 --> 00:57:34,904 rules from the 70s today and expect necessarily the same outcome. 820 00:55:19,472 --> 00:55:46,554 ButI think the strength of this particular strategy is that 821 00:55:46,554 --> 00:56:13,636 it is adaptable and it has proven, over many, many decades, 822 00:56:13,636 --> 00:56:38,256 that it's capable of, over time, to produce not necessarily 823 00:56:38,256 --> 00:57:02,876 the same returns. But that's because people forget that actually 824 00:57:02,876 --> 00:57:25,034 volatility, the volatility of these benchmarks that we compare, 825 00:57:25,034 --> 00:57:34,882 has gone down significantly. Inthe paper I mentioned, that we 826 00:55:39,219 --> 00:56:13,071 did at Dunn, we went back and looked at the rolling volatility 827 00:56:13,071 --> 00:56:55,386 of the Barclays CTA index. And back in the 70s and the 80s, it was 828 00:56:55,386 --> 00:57:23,596 more than 25% annual volatility. Today that index is less 829 00:57:23,596 --> 00:57:34,880 than 5% annualized volatility. So,clearly returns should be lower 830 00:55:37,532 --> 00:56:10,642 if you're just looking at those kind of overall stats without 831 00:56:10,642 --> 00:56:49,772 looking into the stats. So, I mean, actually, I think it's a fair 832 00:56:49,772 --> 00:57:28,902 question to ask, but I think we have a really good answer, in 833 00:57:28,902 --> 00:57:34,922 our case. 834 00:57:35,242 --> 00:58:07,990 Yeah, I mean, the issue is that, if you think about different 835 00:58:07,990 --> 00:58:35,280 kinds of statistical testing, which essentially is if you sort 836 00:58:35,280 --> 00:59:02,570 of look at the rolling performance of trend following over 837 00:59:02,570 --> 00:59:35,318 time and say, well, has that degraded? Visually, you can see some 838 00:59:35,318 --> 00:59:38,047 degradation. AndI appreciate that you kind of have to volatility 839 00:57:56,052 --> 00:58:23,105 adjust, in a sense, because the markets of the 1970s were a lot 840 00:58:23,105 --> 00:58:41,834 more volatile than they are now, particularly commodity markets. 841 00:58:41,834 --> 00:59:06,806 So, even though we are risk targeting or trying to risk target, 842 00:59:06,806 --> 00:59:33,859 over time, to get the same level of risk, it's still going to 843 00:59:33,859 --> 00:59:38,021 be hard. Particularlyin the earlier days when there were fewer 844 00:57:47,797 --> 00:58:00,352 markets, there was less diversification. So that also made 845 00:58:00,352 --> 00:58:21,277 it harder to get a low vol. But even if you sort of accept the 846 00:58:21,277 --> 00:58:36,622 argument that, yeah, it looks like returns are degraded, you still 847 00:58:36,622 --> 00:58:47,782 don't see a statistically significant difference in returns. 848 00:58:47,782 --> 00:59:03,127 And because it's quite a low Sharpe strategy anyway, it's going 849 00:59:03,127 --> 00:59:19,867 to take quite a long time, possibly even another 50 years, to 850 00:59:19,867 --> 00:59:36,607 know for sure that things have changed and it has definitely ‘stopped 851 00:59:36,607 --> 00:59:38,002 working’. Sothat's the kind of… I mean, I've said this before 852 00:58:09,881 --> 00:58:44,520 in a podcast and somebody picked up on LinkedIn. But really, 853 00:58:44,520 --> 00:59:16,010 you're at the point where statistics can't really help you. 854 00:59:16,010 --> 00:59:38,053 You either believe or you don't believe. Theonly thing I can 855 00:57:43,427 --> 00:58:04,708 say is that, yeah, people have been really bad at sort of timing 856 00:58:04,708 --> 00:58:22,715 investment in and out of strategies, and there's lots of evidence 857 00:58:22,715 --> 00:58:42,359 of that. So, my kind of view is, generally speaking, your starting 858 00:58:42,359 --> 00:59:03,640 point with anything should be buy and hold. You know, buy and buy 859 00:59:03,640 --> 00:59:18,373 and hold underlying assets unless you're pretty sure, you've 860 00:59:18,373 --> 00:59:38,017 got a lot of confidence you can do better by trading them. Andif 861 00:57:39,708 --> 00:58:02,038 you're investing in funds, or strategies, or risk factors, then 862 00:58:02,038 --> 00:58:26,601 again, you should be buying and holding those funds, or strategies, 863 00:58:26,601 --> 00:58:48,931 or risk factors, unless you can prove very convincingly that 864 00:58:48,931 --> 00:59:17,960 you can do better. All the evidence I've seen is that it's quite 865 00:59:17,960 --> 00:59:38,057 the opposite. People do a terrible job of timing. 866 00:59:38,792 --> 00:59:43,112 Yeah, absolutely. Speaking of timing, Rob, about one hour now. 867 00:59:43,192 --> 00:59:46,754 Yes, I think it's a good time. We talked already about Fed independence, 868 00:59:46,754 --> 00:59:50,316 which was one of my final topics. So, I think we can leave 869 00:59:50,316 --> 00:59:50,864 it there. 870 00:59:50,872 --> 01:00:23,718 Yeah, yeah, absolutely. This was great. Great to catch up with 871 01:00:23,718 --> 01:00:59,550 you and thanks for all the articles you brought along and all 872 01:00:59,550 --> 01:01:05,522 of that. Andof course, for all of you listening, if you want 873 01:00:06,074 --> 01:00:26,804 to send a little bit of love to Rob, why don't you go to your 874 01:00:26,804 --> 01:00:42,006 favorite podcast platform and leave a rating and review so that 875 01:00:42,006 --> 01:00:59,972 more people can join us when Rob and all the other great co-hosts 876 01:00:59,972 --> 01:01:05,500 are on the show. Speakingof great co-hosts, the next one, next 877 01:00:00,344 --> 01:00:14,552 week, that is joining me will be Andrew Beer. So, definitely worth 878 01:00:14,552 --> 01:00:27,576 tuning in to. I'm sure we'll have our usual interesting conversations; 879 01:00:27,576 --> 01:00:39,416 proper CTAs versus people who replicate CTAs. And that's always 880 01:00:39,416 --> 01:00:54,808 fun. And he's a good sport, and I know he's been publishing some 881 01:00:54,808 --> 01:01:05,464 stuff recently that will be probably good discussion points. 882 00:59:50,872 --> 01:00:19,407 Ifyou have some questions for Andrew, which I hope you do, you 883 01:00:19,407 --> 01:00:32,577 can email them to info@toptradersunplugged.com and 884 01:00:32,577 --> 01:01:05,502 I'll do my very best to make sure that we get them in our conversation. 885 00:59:50,872 --> 01:00:17,821 FromRob and me, thanks ever so much for listening. We look forward 886 01:00:17,821 --> 01:00:44,770 to being back with you next week. And in the meantime, as usual, 887 01:00:44,770 --> 01:01:05,500 take care of yourself and take care of each other. 888 01:01:06,002 --> 01:01:16,442 Thanks for listening to the Systematic Investor podcast series. 889 01:01:16,442 --> 01:01:32,682 If you enjoy this series, go on over to iTunes and leave an honest 890 01:01:32,682 --> 01:01:47,762 rating and review. And be sure to listen to all the other episodes 891 01:01:47,762 --> 01:01:52,402 from Top Traders Unplugged. Ifyou have questions about systematic 892 01:01:15,284 --> 01:01:33,848 investing, send us an email with the word ‘question’ in the subject 893 01:01:33,848 --> 01:01:40,036 line to info@ptoptradersunplugged.com and 894 01:01:40,036 --> 01:01:52,412 we'll try to get it on the show. Andremember, all the discussion 895 01:01:08,697 --> 01:01:14,626 that we have about investment performance is about the past and 896 01:01:14,626 --> 01:01:19,477 past performance does not guarantee or even infer anything 897 01:01:19,477 --> 01:01:24,328 about future performance. Also, understand that there's a significant 898 01:01:24,328 --> 01:01:29,718 risk of financial loss with all investment strategies and you 899 01:01:29,718 --> 01:01:35,647 need to request and understand the specific risks from the investment 900 01:01:35,647 --> 01:01:40,498 manager about their products before you make investment decisions. 901 01:01:40,498 --> 01:01:46,966 Thanks for spending some of your valuable time with us and we'll 902 01:01:46,966 --> 01:01:52,356 see you on the next episode of the Systematic Investor.