1 00:00:09,640 --> 00:00:13,360 Speaker 1: Hello, and welcome to another episode of the Odd Loots Podcast. 2 00:00:13,440 --> 00:00:18,599 Speaker 1: I'm Tracy Alloway and I'm Joe, So Joe. One thing 3 00:00:18,760 --> 00:00:22,239 Speaker 1: we talked about quite a lot on this podcast, especially lately, 4 00:00:22,600 --> 00:00:26,760 Speaker 1: has to be volatility. Yeah. Absolutely, we didn't seem much 5 00:00:26,880 --> 00:00:32,560 Speaker 1: volatility in any markets in and it's picked up more 6 00:00:32,800 --> 00:00:36,560 Speaker 1: in eighteen, though not dramatically, but we certainly have seen 7 00:00:36,600 --> 00:00:40,320 Speaker 1: some very interesting episodes across a range of markets, making 8 00:00:40,360 --> 00:00:43,120 Speaker 1: things quite a bit more interesting. Yeah. I love that 9 00:00:43,120 --> 00:00:45,640 Speaker 1: we have episodes when there's no volatility and then we 10 00:00:45,680 --> 00:00:48,720 Speaker 1: have episodes when there's lots of volatility. But do we 11 00:00:48,840 --> 00:00:54,720 Speaker 1: ever stop to ask about the backbone of volatility measuring 12 00:00:54,880 --> 00:00:58,920 Speaker 1: or volatility models. No, we don't really. I mean, we 13 00:00:59,080 --> 00:01:01,160 Speaker 1: talked a little bit it. It's like we always sort 14 00:01:01,200 --> 00:01:03,440 Speaker 1: of talk around it, don't you Like we talked about 15 00:01:03,960 --> 00:01:06,360 Speaker 1: how the people blew up when they were shorting the 16 00:01:06,520 --> 00:01:10,600 Speaker 1: vix and stuff like that. But it is true that 17 00:01:10,680 --> 00:01:13,920 Speaker 1: to some extent, we feel like these measures of volatility 18 00:01:13,920 --> 00:01:18,720 Speaker 1: are like handed down to us on stone tablets on 19 00:01:18,760 --> 00:01:21,520 Speaker 1: the top of a mountain, rather than something that people 20 00:01:21,560 --> 00:01:24,119 Speaker 1: had to sort of come up with on their own. Yeah, 21 00:01:24,200 --> 00:01:27,760 Speaker 1: so today we're going to be talking about the origins 22 00:01:27,880 --> 00:01:32,080 Speaker 1: of a volatility model that is essentially the backbone of 23 00:01:32,319 --> 00:01:34,760 Speaker 1: a lot of Wall Street risk management and a lot 24 00:01:34,800 --> 00:01:38,600 Speaker 1: of the volatility modeling that we've seen in recent years. 25 00:01:38,720 --> 00:01:41,679 Speaker 1: It's something called value at risk, and I think you're 26 00:01:41,720 --> 00:01:44,760 Speaker 1: already familiar with it. I have a vague idea of 27 00:01:44,800 --> 00:01:46,600 Speaker 1: what it is, but I think it's something like, if 28 00:01:46,640 --> 00:01:49,840 Speaker 1: you have a big portfolio, you want to measure what 29 00:01:50,040 --> 00:01:53,280 Speaker 1: is a sort of reasonable amount you might expect to 30 00:01:53,360 --> 00:01:56,720 Speaker 1: lose on any given day over some time period, to 31 00:01:56,800 --> 00:01:59,920 Speaker 1: see how risky your portfolio is. That's pretty good. Actually, 32 00:02:00,320 --> 00:02:05,680 Speaker 1: that's impressive. So just to just to harden, thanks, I'm 33 00:02:05,720 --> 00:02:09,840 Speaker 1: glad my very rudimentary definition was enough to impress you. No, 34 00:02:09,960 --> 00:02:12,519 Speaker 1: but it's true. I mean, that's it. It's the amount 35 00:02:12,520 --> 00:02:14,400 Speaker 1: of money that you might expect to lose at a 36 00:02:14,440 --> 00:02:18,600 Speaker 1: given confidence level over a certain time period. So, for instance, 37 00:02:18,840 --> 00:02:21,840 Speaker 1: if you and I were running odd lots capital, which 38 00:02:21,880 --> 00:02:24,280 Speaker 1: we totally should do at some point, uh, and we 39 00:02:24,320 --> 00:02:27,400 Speaker 1: had a one day value at risk of a million dollars, 40 00:02:27,440 --> 00:02:31,960 Speaker 1: say at the percent confidence level, that would mean we 41 00:02:32,000 --> 00:02:36,040 Speaker 1: would expect to lose more than a million dollars on 42 00:02:36,040 --> 00:02:39,720 Speaker 1: one day out of twenty at a confidence level. So 43 00:02:40,400 --> 00:02:43,200 Speaker 1: this model, you know, it was invented in the nineties 44 00:02:43,440 --> 00:02:47,160 Speaker 1: by JP Morgan, and it spread throughout all the banks 45 00:02:47,360 --> 00:02:50,239 Speaker 1: and it became the backbone. As I said, but it's 46 00:02:50,240 --> 00:02:54,440 Speaker 1: also intensely controversial, and you see lots of criticisms of it. 47 00:02:55,000 --> 00:02:58,680 Speaker 1: Nicholasness seemed to leb is probably the most famous critic 48 00:02:59,160 --> 00:03:02,280 Speaker 1: of the model. But I should say today we're going 49 00:03:02,360 --> 00:03:05,520 Speaker 1: to speak with someone who not only invented value at 50 00:03:05,600 --> 00:03:08,480 Speaker 1: Risk but can also explain what it is that the 51 00:03:08,560 --> 00:03:10,960 Speaker 1: thing does and what it is that it doesn't do 52 00:03:11,280 --> 00:03:13,680 Speaker 1: great well. I I really don't know much more about 53 00:03:13,720 --> 00:03:16,280 Speaker 1: it beyond what I told you, so I am looking 54 00:03:16,360 --> 00:03:26,639 Speaker 1: forward to learning more. So, without further ado, let's bring 55 00:03:26,680 --> 00:03:30,440 Speaker 1: on our guest for this episode. It is Till Goldenman. Till, 56 00:03:30,560 --> 00:03:33,520 Speaker 1: thank you so much for joining us. You're welcome, glad 57 00:03:33,560 --> 00:03:36,800 Speaker 1: to be with you. So Till, maybe just to begin with, 58 00:03:37,080 --> 00:03:40,920 Speaker 1: you could walk us through your early career history. You know, 59 00:03:41,000 --> 00:03:44,840 Speaker 1: I mentioned that var Value at Risk was invented at 60 00:03:44,920 --> 00:03:48,160 Speaker 1: JP Morgan, and you were obviously at the bank when 61 00:03:48,160 --> 00:03:51,119 Speaker 1: you invented it. But how did you end up there? 62 00:03:52,080 --> 00:03:55,880 Speaker 1: So I was a banker at JP Morgan and in 63 00:03:55,920 --> 00:03:58,480 Speaker 1: the banking career in those days, tho you were sent 64 00:03:58,520 --> 00:04:02,720 Speaker 1: around around the world different jobs and different locations. And 65 00:04:03,560 --> 00:04:09,640 Speaker 1: my job at the time before this was trading room 66 00:04:09,680 --> 00:04:12,480 Speaker 1: in Hong Kong, and that was in charge of that. 67 00:04:12,720 --> 00:04:16,479 Speaker 1: And I was approached this from a numerical viewpoint rather 68 00:04:16,520 --> 00:04:19,960 Speaker 1: than from making money viewpoint, and as a consequence, I 69 00:04:19,960 --> 00:04:22,640 Speaker 1: didn't make that much money, but I had collected a 70 00:04:22,680 --> 00:04:24,840 Speaker 1: lot of numbers. So they said, well, perhaps you're better 71 00:04:24,880 --> 00:04:27,720 Speaker 1: off coming back to New York and apply to you 72 00:04:27,880 --> 00:04:33,240 Speaker 1: number skills and we make the money, which I accepted, 73 00:04:33,279 --> 00:04:36,080 Speaker 1: and I became head of Asset Liability Mansion, which at 74 00:04:36,080 --> 00:04:40,000 Speaker 1: the time was being in charge of the balance sheet 75 00:04:40,080 --> 00:04:44,480 Speaker 1: risks of the back and our trading was increasing all 76 00:04:44,520 --> 00:04:48,880 Speaker 1: over and we then decided to use a new methodology 77 00:04:49,279 --> 00:04:54,600 Speaker 1: to look at trading risks in addition to balance sheet risks. 78 00:04:56,120 --> 00:04:59,080 Speaker 1: And that's when all the numbers I had collected about 79 00:04:59,120 --> 00:05:02,240 Speaker 1: foreign exchange became handy. I knew how to deal with 80 00:05:02,360 --> 00:05:05,600 Speaker 1: large data sets and how to look at large numbers, 81 00:05:05,800 --> 00:05:10,640 Speaker 1: and that's how we came across this value at risk system. 82 00:05:10,800 --> 00:05:15,039 Speaker 1: Until what years were these because obviously these days it's 83 00:05:15,240 --> 00:05:19,080 Speaker 1: unimaginable to think that there would ever be trading, let 84 00:05:19,120 --> 00:05:24,240 Speaker 1: alone large scale trading without a very significant quantitative or 85 00:05:24,400 --> 00:05:28,560 Speaker 1: numerical bent. So when when did this sort of transition 86 00:05:28,640 --> 00:05:35,200 Speaker 1: start to happen? This and we started with looking at 87 00:05:35,240 --> 00:05:39,560 Speaker 1: foreign exchange, which at the time we had about fifteen 88 00:05:39,600 --> 00:05:44,840 Speaker 1: trading rooms around the world and traded about fifteen twenty currencies, 89 00:05:45,080 --> 00:05:49,160 Speaker 1: probably five or six in volume. And we had limits 90 00:05:49,279 --> 00:05:54,080 Speaker 1: around the world which we set in terms of millions 91 00:05:54,120 --> 00:05:57,200 Speaker 1: of dollars of dollar mark or millions of dollars of yen, 92 00:05:57,400 --> 00:06:01,320 Speaker 1: dollars or pounds stelling you could take as a position, 93 00:06:02,360 --> 00:06:05,279 Speaker 1: and that was not very good thing to do, because 94 00:06:05,279 --> 00:06:07,520 Speaker 1: every time you invent you wanted to give a new 95 00:06:07,560 --> 00:06:10,120 Speaker 1: limit on a new currency, you had to set a 96 00:06:10,120 --> 00:06:14,880 Speaker 1: new limit in amounts of outstandings. We decided we need 97 00:06:14,880 --> 00:06:21,080 Speaker 1: a common measure of these limits, and that measure was volatility. 98 00:06:21,800 --> 00:06:24,200 Speaker 1: So how much of the value at risk model was 99 00:06:24,320 --> 00:06:29,200 Speaker 1: influenced by the events of nine seven when you had 100 00:06:29,560 --> 00:06:33,520 Speaker 1: you know, the Black Shows formula that may have contributed 101 00:06:33,600 --> 00:06:37,360 Speaker 1: to the stock market crash and really the first sort 102 00:06:37,400 --> 00:06:40,800 Speaker 1: of um systematic sell off I guess in the market, 103 00:06:42,320 --> 00:06:45,560 Speaker 1: I would say not at all. At that time, we 104 00:06:45,560 --> 00:06:48,240 Speaker 1: were fairly well on the way of understanding how we 105 00:06:48,279 --> 00:06:53,120 Speaker 1: wanted to do it, and we realized that a good 106 00:06:53,160 --> 00:06:56,599 Speaker 1: part of the market wasn't and we could explain why 107 00:06:56,839 --> 00:07:00,000 Speaker 1: why they weren't doing the right way because they were 108 00:07:00,040 --> 00:07:04,719 Speaker 1: looking at normality of markets, which we knew was a 109 00:07:04,800 --> 00:07:09,440 Speaker 1: reasonably good starting position to take, but not necessarily the 110 00:07:09,600 --> 00:07:15,040 Speaker 1: final answer. So we could explain what happened in that instance, 111 00:07:15,040 --> 00:07:18,840 Speaker 1: in that accident, and we stuck to our guns and said, well, 112 00:07:19,400 --> 00:07:22,080 Speaker 1: in the absence of such extreme events, we will at 113 00:07:22,160 --> 00:07:26,560 Speaker 1: least have something better than looking at notional amounts. So 114 00:07:27,000 --> 00:07:30,760 Speaker 1: can you explain, before we really get into the development 115 00:07:31,040 --> 00:07:33,840 Speaker 1: of value at risk, walk us through a little bit 116 00:07:33,880 --> 00:07:39,040 Speaker 1: more what risk management looked like in the old days, 117 00:07:39,080 --> 00:07:43,280 Speaker 1: before the sort of numerical approach, because obviously risk management 118 00:07:43,360 --> 00:07:46,200 Speaker 1: has been around for a long time, But how did 119 00:07:46,240 --> 00:07:51,160 Speaker 1: you how did people approach the concept before you started 120 00:07:51,280 --> 00:07:53,760 Speaker 1: you started doing your work. I think you have to 121 00:07:53,840 --> 00:07:58,120 Speaker 1: understand that risk management in the old banking days was 122 00:07:58,240 --> 00:08:02,960 Speaker 1: mostly about credit. Can you lend somebody money and how 123 00:08:03,040 --> 00:08:05,720 Speaker 1: probable is it that this person will give you the 124 00:08:05,760 --> 00:08:09,880 Speaker 1: money back? And then as the balance sheets of the 125 00:08:09,920 --> 00:08:13,640 Speaker 1: banks got bigger and the interest rates started to move. 126 00:08:14,400 --> 00:08:16,440 Speaker 1: The new risk came up, and that was called the 127 00:08:16,520 --> 00:08:20,360 Speaker 1: interest rate risk of banks, and that was if you 128 00:08:20,480 --> 00:08:24,720 Speaker 1: were borrowing short term and lending long term, that is, 129 00:08:24,800 --> 00:08:27,520 Speaker 1: you could borrow overnight with a short term rate and 130 00:08:27,600 --> 00:08:32,520 Speaker 1: lending for five years, you get squeezed when interest rates 131 00:08:32,559 --> 00:08:37,480 Speaker 1: go up. And to measure that squeeze or that that risk, 132 00:08:38,160 --> 00:08:42,240 Speaker 1: asset liability management was developed in the sixties and seventies. 133 00:08:42,840 --> 00:08:46,439 Speaker 1: Asset liability mansion was something which was long term, but 134 00:08:46,600 --> 00:08:49,080 Speaker 1: if you developed your risks or you looked at the 135 00:08:49,160 --> 00:08:53,080 Speaker 1: risks over years instead of over a shorter period. And 136 00:08:53,120 --> 00:08:57,080 Speaker 1: then in the seventies and eighties trading started to really 137 00:08:57,080 --> 00:09:00,160 Speaker 1: take off. There wasn't much trading in banking before them, 138 00:09:00,920 --> 00:09:05,120 Speaker 1: and the trading was concentrated in foreign exchange, and that 139 00:09:05,280 --> 00:09:09,200 Speaker 1: risk was really an overnight risk because the traders would 140 00:09:09,240 --> 00:09:13,319 Speaker 1: take a position during the day and then some took 141 00:09:13,600 --> 00:09:18,679 Speaker 1: these positions overnight, and overnight you had jumped in currencies 142 00:09:18,800 --> 00:09:22,080 Speaker 1: and you tried to figure out how much could you 143 00:09:22,160 --> 00:09:28,319 Speaker 1: lose overnight. And when these foreign exchange risks became substantial 144 00:09:29,000 --> 00:09:33,480 Speaker 1: in the early eighties, as for an exchange trading took off, 145 00:09:34,120 --> 00:09:38,320 Speaker 1: we needed a measure for that which was in addition 146 00:09:38,360 --> 00:09:42,280 Speaker 1: to the credit risk, in addition to the boundariet risks. 147 00:09:42,320 --> 00:09:46,720 Speaker 1: So as you're developing this model, I mean walk us 148 00:09:46,760 --> 00:09:50,960 Speaker 1: through what it was exactly that was going into it, 149 00:09:51,040 --> 00:09:54,120 Speaker 1: Like what's the data we know that it looks at 150 00:09:54,160 --> 00:09:57,680 Speaker 1: historical data, and what are the parameters um that it 151 00:09:57,720 --> 00:10:00,920 Speaker 1: also involves, and how did you agree on those parameters. 152 00:10:01,800 --> 00:10:05,400 Speaker 1: So it's it's very simple. Assume that you have a 153 00:10:05,440 --> 00:10:11,040 Speaker 1: trader who has a limit of ten million dollars in 154 00:10:11,200 --> 00:10:14,719 Speaker 1: dollar mark exchange risk of foreign exchange risk, of course, 155 00:10:14,800 --> 00:10:19,000 Speaker 1: let's say dollar pounds sterling risk, right, and you would 156 00:10:19,000 --> 00:10:24,200 Speaker 1: simply ask yourself if that trader had that position over 157 00:10:24,240 --> 00:10:26,839 Speaker 1: the last year, over the last two years, or the 158 00:10:26,920 --> 00:10:31,120 Speaker 1: last three years, every night, what is the maximum aunt 159 00:10:31,160 --> 00:10:36,640 Speaker 1: he could have lost overnight on that position. And when 160 00:10:36,640 --> 00:10:40,520 Speaker 1: you looked at the history of these positions over you know, 161 00:10:41,080 --> 00:10:44,160 Speaker 1: or rate changes overnight, you came up with a bell 162 00:10:44,280 --> 00:10:49,520 Speaker 1: curve of distributions of potential gains and losses. And we 163 00:10:49,559 --> 00:10:52,240 Speaker 1: looked at that bell curve and said, well, the best 164 00:10:52,280 --> 00:10:55,640 Speaker 1: way to explain or to put a number value on 165 00:10:55,720 --> 00:11:00,320 Speaker 1: that Bell curve is called volatility or standard v ation 166 00:11:00,520 --> 00:11:05,680 Speaker 1: in mathematical terms, So that was kind of you look 167 00:11:05,720 --> 00:11:08,480 Speaker 1: at the history and see how much you would have 168 00:11:08,559 --> 00:11:10,280 Speaker 1: lost if you had it in the past, and then 169 00:11:10,320 --> 00:11:14,600 Speaker 1: you say, the past is a reasonably good representation of 170 00:11:14,679 --> 00:11:17,720 Speaker 1: what could happen in the future, and if you lost 171 00:11:17,880 --> 00:11:23,120 Speaker 1: no more than a million dollars over of the hundred days, 172 00:11:23,480 --> 00:11:26,600 Speaker 1: then that was a good measure of what you could 173 00:11:26,640 --> 00:11:31,320 Speaker 1: potentially lose. Now, you have to understand that at that time, 174 00:11:32,760 --> 00:11:38,199 Speaker 1: the entire portfolio management theory in asset management, as well 175 00:11:38,240 --> 00:11:46,079 Speaker 1: as the entire options model process, both things which gained 176 00:11:46,440 --> 00:11:51,480 Speaker 1: Nobel prices for their respective inventors, were all based on 177 00:11:51,559 --> 00:11:56,400 Speaker 1: the same concept, that is, measure volatility in terms of 178 00:11:56,520 --> 00:12:00,720 Speaker 1: standard deviation in a normal distribution. So it was a 179 00:12:00,720 --> 00:12:04,000 Speaker 1: reasonable assumption to make that same assumption that was used 180 00:12:04,080 --> 00:12:09,040 Speaker 1: also in asset management and in the options world. So 181 00:12:09,200 --> 00:12:13,800 Speaker 1: then the obvious question is and the criticism that you 182 00:12:13,920 --> 00:12:17,640 Speaker 1: hear now and that you've heard for a while is okay, 183 00:12:17,679 --> 00:12:21,240 Speaker 1: all these measures are based on sort of a normal world, 184 00:12:21,760 --> 00:12:25,280 Speaker 1: but we get fat tales in the world doesn't often 185 00:12:25,480 --> 00:12:28,240 Speaker 1: look like a bell curve. And so then the question 186 00:12:28,360 --> 00:12:31,600 Speaker 1: is what is the value of the model in light 187 00:12:31,679 --> 00:12:35,440 Speaker 1: of what we know about sort of extreme events. I 188 00:12:35,480 --> 00:12:39,000 Speaker 1: think there's two things you have to consider here. Number one, 189 00:12:39,600 --> 00:12:44,720 Speaker 1: you have to understand the context within which the traders 190 00:12:44,800 --> 00:12:48,960 Speaker 1: and management of trading operations who are operating. And the 191 00:12:49,040 --> 00:12:53,760 Speaker 1: context was, we now have a measure, a reasonably good 192 00:12:53,760 --> 00:12:58,680 Speaker 1: measure of risk. Shouldn't we compensate our traders based on 193 00:12:58,760 --> 00:13:02,640 Speaker 1: the profits they make in relation to the risks they take. 194 00:13:04,440 --> 00:13:08,040 Speaker 1: That was a very new concept, and that was really 195 00:13:08,760 --> 00:13:11,640 Speaker 1: very helpful because if you had one trader who was 196 00:13:12,160 --> 00:13:15,679 Speaker 1: trading pork bellies and made a million dollars in profits 197 00:13:15,760 --> 00:13:20,000 Speaker 1: and another trader who was trading dollar mark and made 198 00:13:20,480 --> 00:13:25,520 Speaker 1: a million dollars, which one was the better trader? Well, 199 00:13:25,559 --> 00:13:28,280 Speaker 1: the better trader was the one which had less risk 200 00:13:29,559 --> 00:13:33,280 Speaker 1: compared to the million traders you made. So you now 201 00:13:33,360 --> 00:13:37,880 Speaker 1: had a benchmark to evaluate traders with the same profitability, 202 00:13:38,000 --> 00:13:42,320 Speaker 1: and you increased the limits for the trader which had 203 00:13:42,640 --> 00:13:46,360 Speaker 1: the lower risk number for the same amount of profits. 204 00:13:47,960 --> 00:13:55,080 Speaker 1: That was fundamental in better managing risk activities, and that 205 00:13:55,320 --> 00:14:01,120 Speaker 1: in turn led to distortions because if you now put 206 00:14:01,200 --> 00:14:04,640 Speaker 1: yourself in the position of a trader who says, well, 207 00:14:05,480 --> 00:14:09,320 Speaker 1: if I'm getting paid with a bonus in terms of 208 00:14:09,480 --> 00:14:12,600 Speaker 1: returns un risk. That is, if I make a million 209 00:14:12,600 --> 00:14:16,560 Speaker 1: dollars with ten risks, how about if I make my 210 00:14:16,640 --> 00:14:22,600 Speaker 1: position so they don't look so risky, and now I'm 211 00:14:22,640 --> 00:14:29,920 Speaker 1: motivated very strongly to take positions in of the type 212 00:14:29,960 --> 00:14:35,680 Speaker 1: which has high tail risks, because high tail risks don't 213 00:14:35,720 --> 00:14:42,720 Speaker 1: show up correctly with the standard deviation volatility measurement. So 214 00:14:42,840 --> 00:14:46,120 Speaker 1: there was an and people didn't realize that at the time, 215 00:14:46,240 --> 00:14:50,000 Speaker 1: nor did I. There was a bias created by this 216 00:14:50,360 --> 00:14:58,640 Speaker 1: measurement of risk. Two have traders go against or create 217 00:14:58,720 --> 00:15:02,920 Speaker 1: positions which were not pupperly measured. So that was number 218 00:15:02,960 --> 00:15:09,520 Speaker 1: one very important fact which when in retrospective you see 219 00:15:09,600 --> 00:15:12,920 Speaker 1: it happens all the time. Whenever you measure something and 220 00:15:13,080 --> 00:15:16,320 Speaker 1: you pay people based on that measurement, then they tried 221 00:15:16,360 --> 00:15:23,320 Speaker 1: to gain the system, so that measurement created the bias 222 00:15:23,480 --> 00:15:29,120 Speaker 1: of creating more one sided risks or tail risks. The 223 00:15:29,240 --> 00:15:35,440 Speaker 1: second basic thing that changed was the interconnectivity of the markets. 224 00:15:36,800 --> 00:15:44,080 Speaker 1: The markets in the old days were fairly independent. The 225 00:15:44,160 --> 00:15:47,840 Speaker 1: dollar exchange rate didn't change much when Japanese interest rates 226 00:15:47,880 --> 00:15:52,400 Speaker 1: went up and down, or the interest rates in Uruguay 227 00:15:52,560 --> 00:15:56,240 Speaker 1: didn't change much when the sterling interest rates went down. 228 00:15:57,400 --> 00:16:03,400 Speaker 1: As the financial markets started to link all the markets together, 229 00:16:04,040 --> 00:16:12,040 Speaker 1: the markets became interdependent. And there's a standard theorem in 230 00:16:12,320 --> 00:16:17,920 Speaker 1: engineering which says the more complex and the more interdependent 231 00:16:18,200 --> 00:16:23,360 Speaker 1: the markets system is, the less stable it becomes. And 232 00:16:23,440 --> 00:16:28,880 Speaker 1: stability is basically the Antipoto normal. Right, So the more 233 00:16:29,040 --> 00:16:32,160 Speaker 1: you integrated financial markets around the world, the more you 234 00:16:32,280 --> 00:16:38,360 Speaker 1: made them interdependent and faster, the more you created a 235 00:16:38,520 --> 00:16:44,640 Speaker 1: more normality. So in my perspective, the two things, the 236 00:16:45,000 --> 00:16:50,920 Speaker 1: interdependence and speed of markets and the gaming by the 237 00:16:51,080 --> 00:16:58,000 Speaker 1: profit makers, created more non normality. So that's a really 238 00:16:58,040 --> 00:17:01,640 Speaker 1: interesting argument, and I hadn't actually thought about how this 239 00:17:01,720 --> 00:17:06,440 Speaker 1: basically gave rise to risk adjusted performance for traders. But 240 00:17:06,960 --> 00:17:10,960 Speaker 1: when value at risk is most heavily criticized is usually 241 00:17:11,560 --> 00:17:13,639 Speaker 1: during the financial crisis or in the run up to 242 00:17:13,640 --> 00:17:17,960 Speaker 1: the financial crisis for failing to foresee big trading losses 243 00:17:18,280 --> 00:17:22,359 Speaker 1: on things like mortgage backed securities and subprime bonds and 244 00:17:22,440 --> 00:17:27,520 Speaker 1: stuff like that. Is your argument that because of the models, 245 00:17:27,600 --> 00:17:31,840 Speaker 1: because they were so entrenched, traders were sort of clustering 246 00:17:32,160 --> 00:17:36,680 Speaker 1: into things that didn't look risky, you know, things like 247 00:17:37,080 --> 00:17:41,320 Speaker 1: triple A rated portions of synthetic c d O s, 248 00:17:41,480 --> 00:17:44,960 Speaker 1: but that actually were exposed to significant fat tail risk. 249 00:17:45,840 --> 00:17:49,080 Speaker 1: And also that the financial system was more inter related, 250 00:17:49,160 --> 00:17:54,400 Speaker 1: and so you had correlations that the model wasn't necessarily capturing. 251 00:17:54,600 --> 00:18:01,720 Speaker 1: Is that the criticism that's almost correct. When you say correlations, 252 00:18:03,160 --> 00:18:06,959 Speaker 1: it means that the correlations were not normal. The correlation 253 00:18:07,080 --> 00:18:10,360 Speaker 1: is just a statistical measure, and again there you assume 254 00:18:10,920 --> 00:18:15,919 Speaker 1: normality in the correlation distribution. But the more interdependent and 255 00:18:15,960 --> 00:18:20,720 Speaker 1: the more non normal the positions are, the less the 256 00:18:20,960 --> 00:18:28,359 Speaker 1: standard measurements of volatility and correlations are correct. So is 257 00:18:28,359 --> 00:18:31,920 Speaker 1: the idea. Let's say I were a foreign exchange trader, 258 00:18:31,960 --> 00:18:35,439 Speaker 1: and I was trading the Euro, and I was trading 259 00:18:35,520 --> 00:18:39,159 Speaker 1: the end and the Korean wan and so forth, And 260 00:18:39,240 --> 00:18:42,280 Speaker 1: you might look at the past hundred days or three 261 00:18:42,359 --> 00:18:46,200 Speaker 1: years or five years of performance of each of those 262 00:18:46,320 --> 00:18:50,480 Speaker 1: and come up with some sort of reasonable expectation that, Okay, 263 00:18:50,720 --> 00:18:53,240 Speaker 1: the one falls this much and the end falls this much, 264 00:18:53,320 --> 00:18:57,399 Speaker 1: and stuff like that. But the idea being that what 265 00:18:57,520 --> 00:19:00,919 Speaker 1: happens now, or what happens in inter correlated market is 266 00:19:00,920 --> 00:19:04,040 Speaker 1: that they all fall into they all move dramatically on 267 00:19:04,080 --> 00:19:06,400 Speaker 1: the same day at the same time, and so they're 268 00:19:06,400 --> 00:19:09,959 Speaker 1: not just sort of random and normal, but the performance 269 00:19:10,000 --> 00:19:12,600 Speaker 1: of all the different positions in the books go to 270 00:19:12,680 --> 00:19:15,119 Speaker 1: extremes at the same time. Is that sort of the 271 00:19:15,200 --> 00:19:19,720 Speaker 1: idea of like how the model breaks down under more 272 00:19:19,720 --> 00:19:24,080 Speaker 1: extreme correlations or more extreme interconnectedness. That's correct, it's the 273 00:19:24,119 --> 00:19:28,320 Speaker 1: interconnectedness and the complexity of the system which makes these 274 00:19:28,720 --> 00:19:33,840 Speaker 1: movements erratic. But at the same time you have to 275 00:19:34,080 --> 00:19:39,000 Speaker 1: invent positions which don't look so risky. And the first 276 00:19:39,080 --> 00:19:43,000 Speaker 1: type of positions that were taken were called long dated 277 00:19:43,119 --> 00:19:49,440 Speaker 1: forwards options. That is, a trader said, well, dollar N 278 00:19:50,359 --> 00:19:56,520 Speaker 1: usually moves within ten percent within how about if I 279 00:19:56,640 --> 00:20:00,320 Speaker 1: make a contract but which says I'll sell you an 280 00:20:00,359 --> 00:20:06,879 Speaker 1: option that the end doesn't move more than over the 281 00:20:06,960 --> 00:20:12,800 Speaker 1: next ten years. And if you take that position, you 282 00:20:12,880 --> 00:20:17,520 Speaker 1: make money every day except when the head's the fan, 283 00:20:18,200 --> 00:20:21,560 Speaker 1: and that's when you lose really big. So this is 284 00:20:21,600 --> 00:20:27,640 Speaker 1: a typical highly high tail risk position. And that kind 285 00:20:27,680 --> 00:20:32,560 Speaker 1: of position doesn't get picked up correctly with the war measurements, 286 00:20:32,600 --> 00:20:36,400 Speaker 1: and the traders were biased to take these positions because 287 00:20:36,920 --> 00:20:41,360 Speaker 1: they got easier limits for those and could make more 288 00:20:41,440 --> 00:20:46,720 Speaker 1: profits on a risk adjusting basis. Why doesn't var pick 289 00:20:46,800 --> 00:20:49,960 Speaker 1: up that risk because VAR is a you know, is 290 00:20:50,000 --> 00:20:55,800 Speaker 1: this simple measurement which depends on the normal malaity of 291 00:20:55,960 --> 00:21:01,040 Speaker 1: the distribution. And if you didn't have that simple measure 292 00:21:01,280 --> 00:21:05,159 Speaker 1: of the bell curve, you couldn't do the math. It 293 00:21:05,320 --> 00:21:09,440 Speaker 1: was simply the math becomes so complex at the very 294 00:21:09,480 --> 00:21:13,840 Speaker 1: moment you move away from normality that it was just 295 00:21:13,960 --> 00:21:18,280 Speaker 1: not feasible. So I have a related question. You walked 296 00:21:18,359 --> 00:21:22,880 Speaker 1: us through how how the traders would game the VAR system. 297 00:21:22,920 --> 00:21:27,119 Speaker 1: How did value at risk actually fit into risk management 298 00:21:27,240 --> 00:21:30,080 Speaker 1: at the banks, Like, how were the senior risk managers 299 00:21:30,359 --> 00:21:34,119 Speaker 1: using it, how valuable did they find it as a 300 00:21:34,240 --> 00:21:39,080 Speaker 1: risk management instrument, and what actually happened to people if 301 00:21:39,160 --> 00:21:42,199 Speaker 1: they breached their VAR limits. I've never been able to 302 00:21:42,240 --> 00:21:44,159 Speaker 1: get to the bottom of that. Everyone seems to have 303 00:21:44,200 --> 00:21:48,399 Speaker 1: a different answer. Well, we had something which was called 304 00:21:48,440 --> 00:21:54,000 Speaker 1: the four fifteen report, which was every day at quarter 305 00:21:54,080 --> 00:21:57,960 Speaker 1: past four in the afternoon, we would have assembled a 306 00:21:58,040 --> 00:22:02,760 Speaker 1: report which would oh the bar that they did the 307 00:22:02,880 --> 00:22:07,399 Speaker 1: risk around the globe in all our trading positions. And 308 00:22:07,480 --> 00:22:11,240 Speaker 1: with that report we went to the chairman's office and said, well, 309 00:22:11,280 --> 00:22:14,240 Speaker 1: here is your risk. And the chairman at that point 310 00:22:14,280 --> 00:22:18,560 Speaker 1: in time was Mr Weatherstone, and he said, yeah, that's 311 00:22:18,640 --> 00:22:20,800 Speaker 1: like a nice number, but you know, I don't really 312 00:22:20,800 --> 00:22:25,960 Speaker 1: believe so much into it. Perhaps it's not right, and 313 00:22:26,040 --> 00:22:33,000 Speaker 1: he was rightfully skeptical. But the number proved out to 314 00:22:33,000 --> 00:22:37,400 Speaker 1: be reasonably good assumption for most of the days, so 315 00:22:37,600 --> 00:22:44,480 Speaker 1: everybody became more reliable. You could show that the actual 316 00:22:44,720 --> 00:22:50,600 Speaker 1: loss in the global positions of JP Morgan was only 317 00:22:51,119 --> 00:22:54,160 Speaker 1: five in five out of a hundred days, more than 318 00:22:54,240 --> 00:22:59,040 Speaker 1: that number showed, so that was the verification of the calculations. 319 00:23:00,680 --> 00:23:06,080 Speaker 1: There was always a understanding that this is not the 320 00:23:06,160 --> 00:23:10,600 Speaker 1: maximum you could lose. It's the minimum you could lose 321 00:23:11,080 --> 00:23:16,320 Speaker 1: in a bad situation. So in ent a time you 322 00:23:16,320 --> 00:23:19,359 Speaker 1: would not lose more, but in the other five percent 323 00:23:19,560 --> 00:23:22,680 Speaker 1: of the time you would actually lose more. So it's 324 00:23:22,760 --> 00:23:28,760 Speaker 1: part of the criticism of our sort of mischaracterization of 325 00:23:28,800 --> 00:23:32,200 Speaker 1: how it was used, or perhaps present, in your view, 326 00:23:32,240 --> 00:23:37,400 Speaker 1: a cartoonish view or a naivete that did never actually 327 00:23:37,400 --> 00:23:42,280 Speaker 1: really existed. I think that naivete didn't exist in large 328 00:23:42,359 --> 00:23:47,800 Speaker 1: sophisticated institutions, and the Morgan was one of them, but 329 00:23:47,840 --> 00:23:52,360 Speaker 1: there were many others. But then as the var became 330 00:23:52,520 --> 00:23:58,120 Speaker 1: kind of a government regulated standard or a central bank 331 00:23:58,200 --> 00:24:01,360 Speaker 1: regulated standard of or so a lot of people started 332 00:24:01,400 --> 00:24:09,440 Speaker 1: to use it without really understanding what the limits were. 333 00:24:10,680 --> 00:24:16,359 Speaker 1: It's very much like you drive on a highway above 334 00:24:16,400 --> 00:24:20,320 Speaker 1: the speed limit and you feel happy because you're faster, 335 00:24:21,000 --> 00:24:23,359 Speaker 1: and then you see a big accident and you slow down, 336 00:24:24,160 --> 00:24:27,000 Speaker 1: and then an hour later you're back to your old speed. 337 00:24:28,119 --> 00:24:34,119 Speaker 1: People take risks as they perceive it, and when somebody 338 00:24:34,160 --> 00:24:37,199 Speaker 1: says sixty is the right number or bar is the 339 00:24:37,320 --> 00:24:42,720 Speaker 1: right number, then they behave accordingly until nothing but bad 340 00:24:42,840 --> 00:24:46,120 Speaker 1: happens much worse, and then when something happens then they 341 00:24:46,200 --> 00:24:49,600 Speaker 1: kind of rain themselves in and then go relax again 342 00:24:49,640 --> 00:24:54,359 Speaker 1: over time, so until in the aftermath of the financial crisis, 343 00:24:54,400 --> 00:24:58,400 Speaker 1: we did see some banking regulators who try to alter 344 00:24:59,000 --> 00:25:01,720 Speaker 1: var models. They try to make them more robust. You 345 00:25:01,800 --> 00:25:05,520 Speaker 1: had things like stressed far, which was supposed to be 346 00:25:05,520 --> 00:25:09,840 Speaker 1: better at measuring the fat tails in the probability distribution. 347 00:25:10,160 --> 00:25:14,880 Speaker 1: You also had other models like expected shortfall. How useful 348 00:25:14,920 --> 00:25:19,440 Speaker 1: do you think those changes actually are to value at risk? 349 00:25:20,760 --> 00:25:24,280 Speaker 1: I think they're very useful. Vari at risk is the 350 00:25:24,359 --> 00:25:28,480 Speaker 1: starting point to get a feel of kind of where 351 00:25:28,520 --> 00:25:30,679 Speaker 1: the risks are about, and then you have to do 352 00:25:30,800 --> 00:25:36,120 Speaker 1: stress modeling and you have to do simulation under extreme circumstances. 353 00:25:36,760 --> 00:25:40,800 Speaker 1: So these are all very good further developments, particularly if 354 00:25:40,840 --> 00:25:44,000 Speaker 1: you have a bias in the markets that runs against you. 355 00:25:45,000 --> 00:25:48,199 Speaker 1: The bias are as I mentioned before, the complexity and 356 00:25:48,280 --> 00:25:53,000 Speaker 1: the trader motivation. So till the other thing that we've 357 00:25:53,040 --> 00:25:57,560 Speaker 1: seen happen since the financial crisis is that volatility trading 358 00:25:57,840 --> 00:26:01,760 Speaker 1: has sort of become a thing in and of itself, 359 00:26:01,880 --> 00:26:04,240 Speaker 1: or at least on a scale that we didn't necessarily 360 00:26:04,320 --> 00:26:09,919 Speaker 1: see before two two thousand nine. Given your background in 361 00:26:10,200 --> 00:26:14,119 Speaker 1: modeling volatility, what do you think about the explosion in 362 00:26:14,200 --> 00:26:19,760 Speaker 1: volatility trading? You know, retail investors buying and selling things 363 00:26:19,840 --> 00:26:23,240 Speaker 1: like exchange traded products tied to the VIX. Is that 364 00:26:23,320 --> 00:26:28,240 Speaker 1: inherently risky or does volatility not necessarily equal risk? No? 365 00:26:28,400 --> 00:26:32,800 Speaker 1: I think volatility trading is very much like all other trading, 366 00:26:32,880 --> 00:26:36,440 Speaker 1: particularly if it's done in retail, is done by UH 367 00:26:36,640 --> 00:26:39,760 Speaker 1: income poops who aren't done, understand and they just want 368 00:26:39,760 --> 00:26:43,800 Speaker 1: to go to the casino. And the financial institutions are 369 00:26:43,880 --> 00:26:47,480 Speaker 1: very happy to provide the casino environment because they can 370 00:26:47,480 --> 00:26:52,320 Speaker 1: make money of it. So volatility trading is like any 371 00:26:52,359 --> 00:26:54,680 Speaker 1: other kind of trading. It's a little bit more complex 372 00:26:54,720 --> 00:26:59,240 Speaker 1: and more it has a better story to it, but 373 00:27:00,200 --> 00:27:06,199 Speaker 1: it's just another way of speculation for the retail for 374 00:27:06,480 --> 00:27:11,480 Speaker 1: the professional, it's a more sophisticated way of hedging things. 375 00:27:12,200 --> 00:27:15,919 Speaker 1: So until clearly you're you're not still working at JP Morgan. 376 00:27:16,440 --> 00:27:18,919 Speaker 1: When did you get out of finance and what are 377 00:27:18,920 --> 00:27:22,800 Speaker 1: you up to now? Well? I got out of finance 378 00:27:22,840 --> 00:27:29,040 Speaker 1: in the early nineties, mid nineties when my career I 379 00:27:29,080 --> 00:27:33,040 Speaker 1: think JP Morgan was stalling and I didn't think that 380 00:27:33,240 --> 00:27:37,840 Speaker 1: was right. So I moved out and came to the 381 00:27:37,920 --> 00:27:44,159 Speaker 1: Silicon Valley to work in a startup in financial technology, 382 00:27:45,000 --> 00:27:47,600 Speaker 1: and I was very lucky. That start up turned out 383 00:27:47,680 --> 00:27:52,159 Speaker 1: very well, and it allowed me then to and do 384 00:27:52,280 --> 00:28:00,000 Speaker 1: other things like building a vineyard, growing wines and making wine, 385 00:28:00,520 --> 00:28:05,640 Speaker 1: and now by now I have fully transitioned into agriculture. 386 00:28:06,359 --> 00:28:11,240 Speaker 1: That does sound like a pretty awesome turn of events there, Oh, 387 00:28:11,240 --> 00:28:13,639 Speaker 1: it's wonderful, I can tell you. I would I wish 388 00:28:13,680 --> 00:28:17,680 Speaker 1: I had seen how nice it was a long time ago, 389 00:28:18,600 --> 00:28:20,719 Speaker 1: and would have become a farmer in the first place. 390 00:28:21,800 --> 00:28:25,280 Speaker 1: Are there any similarities between modeling risk at a large 391 00:28:25,280 --> 00:28:30,520 Speaker 1: investment bank and growing grapes in California? Yeah? I think 392 00:28:31,680 --> 00:28:36,200 Speaker 1: we understand an equal amount of both. We certainly, I 393 00:28:36,200 --> 00:28:39,120 Speaker 1: am surprised how little we understand about how to grow 394 00:28:39,280 --> 00:28:43,280 Speaker 1: good grapes and make good wine. It's all in art, 395 00:28:44,120 --> 00:28:46,960 Speaker 1: and the more numbers I applied to it, the less 396 00:28:47,520 --> 00:28:51,000 Speaker 1: certain I am I understand what's going on. So it's 397 00:28:51,200 --> 00:28:54,880 Speaker 1: very much like risk management. Collect a lot of numbers 398 00:28:54,920 --> 00:28:57,960 Speaker 1: and you hope you'll get to some insight, and by 399 00:28:58,000 --> 00:29:01,120 Speaker 1: the time you think you have some insight, uh, something 400 00:29:01,120 --> 00:29:04,840 Speaker 1: else happens and you start all over again. I love 401 00:29:04,880 --> 00:29:08,040 Speaker 1: that answer because it sort of blows up literally everything 402 00:29:08,120 --> 00:29:10,120 Speaker 1: we think we know about the modern world, which is 403 00:29:10,280 --> 00:29:12,920 Speaker 1: they had all these old practices that people do. We 404 00:29:12,960 --> 00:29:15,520 Speaker 1: could just perfect them more if we really apply some 405 00:29:16,160 --> 00:29:20,000 Speaker 1: data or AI and machine learning. And that is your 406 00:29:20,040 --> 00:29:23,480 Speaker 1: answer sort of just undercuts the entire thing. It's not 407 00:29:23,520 --> 00:29:27,479 Speaker 1: on the cutting. It's just that as we industrialize the world, 408 00:29:28,400 --> 00:29:30,640 Speaker 1: you know, we have to put numbers on things because 409 00:29:30,720 --> 00:29:34,720 Speaker 1: you can only manage my numbers, and there's limits to 410 00:29:34,920 --> 00:29:39,959 Speaker 1: how much you can industrialize. But in the wine industry, 411 00:29:40,120 --> 00:29:43,360 Speaker 1: the world has become industrialized. It's like coke. You know, 412 00:29:43,440 --> 00:29:46,480 Speaker 1: a lot of large producers produced the bulk of all 413 00:29:46,480 --> 00:29:51,560 Speaker 1: the wine that's drunk and very small number of small 414 00:29:51,600 --> 00:29:56,440 Speaker 1: producers getting all the stories. But they don't do it 415 00:29:56,480 --> 00:30:01,880 Speaker 1: that way, so most of the wine is produced by 416 00:30:04,720 --> 00:30:07,840 Speaker 1: of all the producers, and it was exactly the opposite 417 00:30:08,080 --> 00:30:11,000 Speaker 1: a hundred years ago. We'll have to do an odd 418 00:30:11,040 --> 00:30:14,880 Speaker 1: lots of episode on the changing wine market at some point, 419 00:30:14,960 --> 00:30:18,000 Speaker 1: but I think we'll have to leave it there for 420 00:30:18,160 --> 00:30:21,120 Speaker 1: now till Goldenman. Thank you so much for joining us. 421 00:30:21,200 --> 00:30:24,560 Speaker 1: Really fascinating conversation. Thank you for having me. It was 422 00:30:24,600 --> 00:30:27,480 Speaker 1: a pleasure. Oh Jill, wait before you go, what's the 423 00:30:27,520 --> 00:30:29,200 Speaker 1: name of your winery so people can look it up. 424 00:30:29,440 --> 00:30:34,200 Speaker 1: It's called shadow heads Chive, which comes from Swiss German 425 00:30:34,400 --> 00:30:43,840 Speaker 1: and means there is no chapeau here. Perfect. Thanks Jolly 426 00:30:54,320 --> 00:30:58,240 Speaker 1: so Joe. I found that conversation really really interesting. Again. 427 00:30:58,600 --> 00:31:01,360 Speaker 1: I think it's great to actually go to one of 428 00:31:01,360 --> 00:31:06,360 Speaker 1: the foundations of volatility as we currently understand it and 429 00:31:06,400 --> 00:31:09,120 Speaker 1: talk about it today. But there are also things in 430 00:31:09,160 --> 00:31:11,920 Speaker 1: there that I hadn't really thought about before, like the 431 00:31:11,920 --> 00:31:14,880 Speaker 1: notion that because you had this model, you had a 432 00:31:14,920 --> 00:31:17,920 Speaker 1: bunch of traders who essentially tried to game it by 433 00:31:18,240 --> 00:31:21,920 Speaker 1: clustering into things that they didn't think the model would 434 00:31:21,920 --> 00:31:24,360 Speaker 1: pick up. I thought that was fascinating. I had not 435 00:31:24,520 --> 00:31:27,000 Speaker 1: thought about that either, and so now that's a whole 436 00:31:27,040 --> 00:31:29,600 Speaker 1: new sort of avenue of the thing I want to 437 00:31:29,640 --> 00:31:34,080 Speaker 1: think about and explore. Also, the idea that we look 438 00:31:34,160 --> 00:31:37,200 Speaker 1: at VAR as this sort of regulatory measure, but it 439 00:31:37,240 --> 00:31:40,040 Speaker 1: didn't start off that way. And so the idea that 440 00:31:40,360 --> 00:31:43,760 Speaker 1: maybe at one point this was a thing that sophisticated 441 00:31:43,760 --> 00:31:48,400 Speaker 1: people understood have limitations eventually becomes this thing that becomes 442 00:31:48,440 --> 00:31:51,520 Speaker 1: a sort of de facto measure of bank health inappropriately 443 00:31:52,360 --> 00:31:54,920 Speaker 1: is a concept that also I had never really thought 444 00:31:54,960 --> 00:31:58,240 Speaker 1: about before. And also I think sort of vindicates its 445 00:31:58,360 --> 00:32:03,320 Speaker 1: usefulness even with the well known limitations that it has. Yeah, 446 00:32:03,400 --> 00:32:06,440 Speaker 1: it kind of makes me think that the criticisms that 447 00:32:06,520 --> 00:32:09,480 Speaker 1: you've seen of value at risk that they can't anticipate 448 00:32:09,520 --> 00:32:13,080 Speaker 1: tail risks. You know, it's not really it's not a 449 00:32:13,120 --> 00:32:15,440 Speaker 1: great criticism of the model. Or what I mean is 450 00:32:15,520 --> 00:32:18,000 Speaker 1: we we shouldn't be criticizing the model. Maybe the thing 451 00:32:18,040 --> 00:32:20,000 Speaker 1: we should be criticizing is the fact that we have 452 00:32:20,120 --> 00:32:24,040 Speaker 1: these huge institutions that are so complex that you can't 453 00:32:24,040 --> 00:32:27,200 Speaker 1: actually come up with any model that's able to accurately 454 00:32:27,440 --> 00:32:30,280 Speaker 1: capture everything. It is that they do and all the 455 00:32:30,360 --> 00:32:34,160 Speaker 1: risks that that entails. Absolutely, and I think it's funny, uh, 456 00:32:34,320 --> 00:32:36,160 Speaker 1: you know, even though it's sort of a have joking 457 00:32:36,240 --> 00:32:40,600 Speaker 1: question maybe about the connection between wine growing and risk management. 458 00:32:41,200 --> 00:32:44,600 Speaker 1: I do think there is a common threat of just humility, Like, yes, 459 00:32:44,680 --> 00:32:48,120 Speaker 1: there's only so much we can know about things that 460 00:32:48,200 --> 00:32:51,480 Speaker 1: will happen in the future, and that sort of commands 461 00:32:51,640 --> 00:32:54,400 Speaker 1: us or requires us to not try to get too 462 00:32:54,440 --> 00:32:58,480 Speaker 1: scientific about what could go wrong. Yeah. Absolutely, I'm going 463 00:32:58,520 --> 00:33:01,280 Speaker 1: to go think about the usefulness of financial models and 464 00:33:01,320 --> 00:33:04,600 Speaker 1: how much we actually know about finance over a glass 465 00:33:04,600 --> 00:33:07,680 Speaker 1: of wine right now, and I am going to go 466 00:33:07,760 --> 00:33:12,040 Speaker 1: back to my desk to normal work. All right, fair enough. 467 00:33:12,520 --> 00:33:15,680 Speaker 1: This has been another episode of the Odd Thoughts podcast. 468 00:33:15,760 --> 00:33:18,480 Speaker 1: I'm Tracy Alloway. You can follow me on Twitter at 469 00:33:18,520 --> 00:33:21,760 Speaker 1: Tracy Alloway. And I'm Joe Wisenthal. You can follow me 470 00:33:21,840 --> 00:33:24,800 Speaker 1: on Twitter at the Stalwart, and you should follow our 471 00:33:24,880 --> 00:33:28,440 Speaker 1: producer on Twitter to for Foreheads. He's at foreheads T, 472 00:33:29,080 --> 00:33:32,960 Speaker 1: as well as the Bloomberg head of podcast, Francesco Levy 473 00:33:33,400 --> 00:33:35,959 Speaker 1: at Francesco Today. Thanks for listening.