The Wisest Man on Wall St.
A few weeks ago, I had the opportunity to sit down and speak with Aaron Brown, Risk Manager at AQR about risk management and his new book. There is no doubt that Aaron is highly intelligent, but I think he has even greater wisdom than intelligence. He is entirely unassuming for his talent and he’s extremely funny…we laughed a lot during our chat. I can talk to Aaron all day, and that’s what eventually happened to our scheduled 45-minute conversation about his book: it went on for 2 hours. Talk about risk management…
Here is the transcript of that conversation.
Mike Martin: Hi, everybody. It’s Michael Martin of “MartinKronicle.” Today’s interview is with Aaron Brown, the Risk Manager of AQR and who Nassim Taleb called “the Dean of Wall Street Risk Managers.”
He’s the author of a great new book that I heartily recommend called Red‑Blooded Risk.
Aaron, thanks for being here.
Aaron Brown: Thanks for having me.
Mike: It’s a great honor. I liked all your books and the fact that you have played professional poker. You’re definitely my kind of guy. On Wednesday (November 23) there was a bit of a failed auction, in German bunds. What was your initial impression with that?
Aaron: Well, it was a big surprise. We’ve all been watching Europe very anxiously, because there’s a lot of moving parts and this thing could break in a lot of ways. Ultimately after some ups and downs of the market, I really view it was a positive. I think we’re in better shape Friday than we were on Wednesday morning before this happened.
The big thing is European equities are still doing pretty well. So people are telling us, that the government’s got problems, they don’t know which parts they’re going to pay, they expect problems, but, we’ll continue making things and people will have jobs and we’ll get along, and I think that’s the attitude we need, to get through this thing.
Mike: Now, interesting enough, you wrote an article for Wilmott magazine called Let’s Put the Fun Back into Funded Debt, if I remember correctly? So let’s kind of make the segue from Wednesday auction to the bund offering, and talk about how things could have gone differently, had they employed some of the things that you delineated in that article?
Aaron: Well, I’m not sure that would be the solution through things today. The point of the article was that government debt ought to be marketed aggressively to everyone, that it’s not good when all of your debt is held by foreigners, or all of your debt is held by rich people. It creates divisions in society, conflicts of interests. You really need to get everyone on the same page.
When things are healthy financially, it’s the people, citizens supporting the government by buying bonds. But the Government was looking at the citizens’ welfare, both financial welfare in the future and currently. One of the things that, historically, has made government bonds much more popular is that they had some gambling elements. Public investment in England is something they call premium bonds, which are basically prepaid strips of lottery tickets.
You add a little gambling to things and people will trade a lot more, people buying a lot more, and it really makes no difference to the government. The normal bond, the government gets money today, kind of paves it out later in the future, pro rata, to bond‑holders with the gambling it just pays more money to some bond‑holders, and no money to others.
But individuals will find it much easier to buy these things and hold them and that’s a much broader participation in governments’ auctions and Europe’s kind of turned all of its bonds into gambling bonds, but they didn’t do that on purpose. They just risked the default or this was you know a three dollar bit bonds and expensive bonds but they didn’t do it on purpose. So now, they’re just risking a default or a partial payment it’s a break that we’re getting accidentally.
Mike: Now that would seemed like it would be a little bit politically risqué, here in the United States, to put a gambling element into government issued debt, would it not?
Aaron: I think it might be, although many countries have done this and historically it’s very common. In fact the New York Stock Exchange they held a brokers freeze, it started out as lottery brokers and started adding a disreputable thing called Common Stock, it was only later in the 19th century that Common Stock became respectable and lotteries became disreputable.
Also, I think that given the economics of state’s bonds to gamble the revenue we have already, it’s kind of hypocritical, if you’re going to draw the line at bounces of gambling elements and so that’s…
Mike: Right! Well, we talked about a few things that when I think, understanding risk management and mathematical expectations is the key to good gambling, if you call it that, or risk taking.
In Red Blooded Risk you mentioned something about lottery tickets, and I said something very provocative to you in an email, to prepare you for this ‑ about the pricing of lottery tickets. We can use New York State Lotto as an example: you get two games for the price of a dollar. I thought it was priced at a dollar because it’s basically a throwaway piece of paper at that time. And you had a different opinion about why it was priced at a dollar. Tell everybody what your take was.
Aaron: I am really going to jump all over it, it’s a good way to see how I feel about things. The first thing I tell people, though everyone pretty much knows this, is that the majority of people buy lottery tickets, including the majority of college graduates, and we find gambling in virtually every human culture. It seems to go way back in history.
So people are going to call it irrational when they talk about how it’s got a negative expected value, so you shouldn’t buy tickets. You’re really saying that more than half of the people, including more than half the educated people, are irrational. And that every human culture is doing this thing.
I think that just automatically puts you in the position, like a guy who comes out and says, dancing, for example, dancing’s also ancient in human culture and most people do it. Everybody says, “Oh, it’s irrational, it’s a waste of energy, and it doesn’t produce any productive goods or services.” We can call it introspective or not and we’d say “You better have a darn good theory or a lot of empirical evidence, before we can even entertain this hypothesis.” For some reason, the people who assert this thing about lotteries, when they expect value, as if they should be taken seriously.
So, my first comment is, the burden of proof is on the person who says “These lotteries are irrational, that these people playing it are misinformed or stupid.” Given how many people do it. The next thing I go into, so, “OK, what is the argument that this is irrational?” Now when you look at it, lots of people seem to accept this argument unquestionably and this is very strong assumptions.
There’s something called utility, that how you feel isn’t some complicated combination of lots of things where you’ve got long‑term feelings or short‑term feelings, feelings of security, feelings of love, and all these things kind of merge together to have your state. There’s one number you can put on how you feel.
Not only that, there’s one number that’s related to the amount of money that you have, and if you get one more dollar you’re going to feel better, and if you get two more dollars, that second dollar is also going to make you feel better, but not as much better as the first one. Because the first one you’ll use to buy the stuff you want the most, and the second the next most.
And you can weigh this on a ladder all the way up to $100,000,000 Powerball prize, and each dollar is this neat mathematical ladder. There’s no threshold effect, there’s no amount of money when you start changing who you are, and how you act, and your lifestyle, that’s all. So it’s a continuous function with diminishing flow.
Finally, you have to assume that the proper way when you’re dealing with a gamble and you have to make a choice about it, you’re not supposed to think about the most likely outcome, you’re not supposed to think about the range, what’s the worst that could happen, what’s the best that could happen.
What you’re supposed to do is say, “Look, if we’re going to play this an infinite number of times, what would my average outcome be?” And there’s nothing in rationality that has to do with the way you look at things. So, we have these crazy ideas somebody’s saying that lotteries are foolish or irrational, and we find out that the reason for it is this complicated model. It really makes no sense and nobody would really believe.
Let’s look at the empirical evidence, right? Let’s actually go out and maybe even buy a lottery ticket to see what it feels like. I talked to some people who play the lottery, talked to similar people, one who plays, or one who doesn’t. We talked to people who wouldn’t, and what they do. Maybe going to the lab and do some experiments with lotteries. People have done all this stuff.
We actually know why people buy lottery tickets. To complicate it different ones do it for different reasons. No one has yet identified the irrational one, or a person who’s misguided about the outcomes.
The most common thing, the way to think about it, is the guy who’s going to the store, he buys a six‑pack of beer and he chooses not, “Do I buy a lottery ticket or not buy a lottery ticket,” he chooses to buy a name‑brand beer for a dollar more, or do I buy the generic beer and use the dollar for a lottery ticket. So, if he’s trading off minor short‑term luxuries…and this is where money people spend on lottery for most people, this is the most common situation.
They take it out from other minor luxuries. What we discovered when we talked to them and say “I actually as much or more satisfaction per purchase” So the guy who walks out of the store with a generic beer and a lottery ticket actually feels better even if he loses, he wins no money from the lottery, than the guy who bought the name brand.
The guy who buys the name brand might enjoy his beer a little more the first or second time he drinks it, but after that he gets no more satisfaction out of beer than the guy who buys the generic beer. But if he later has to go back from the name brand to the generic, it hurts him more than the gain he got when he moved up. So then he starts saying which one of these guys is irrational.
Plus, on the lottery package everybody who buys a lottery ticket actually has a chance of winning something. You don’t have much chance on one ticket, but many people buy lots of tickets and play over and over and they re‑invest their intermediate prizes, and then you add that up. The odds are not astronomical against winning over a lifetime.
So if the lottery tickets are kind of a form of saving. It’s both a form of consumption, the actual playing of the ticket is consumption, it’s at least as rational as other minor luxuries, and this is a form of savings. I think the real significance of lotteries is not “Lotteries are a terrible thing and people are irrational, we should get rid of them.”
We can’t design financial products that are as popular as lotteries, and lotteries, when you actually figure out what the state takes out of them, it’s not just the 60% or so they advertise, it’s more like 85 or 90%, because they tax the winnings and the intermediate prizes…people forget that.
So, for every dollar spent on lotteries, the state takes about 90 cents and we can’t design mutual funds, or bank accounts, or ETS, or any kind of financial products that could get anywhere near as many people to voluntarily purchase them. One of the reasons is that we have this tremendous unwillingness to use the insights from retail gambling.
We don’t go into a casino and say, “Ooh, these guys are getting people to voluntarily risk their money, we could learn something from them.” We go, and we say, “We got to shut this place down because it’s evil and bad.”
Mike: That’s very provocative, and I appreciate your insight. I know I’ve bought my share of lottery tickets. From finance class, we’re taught Expected Value and typically, that if something has a negative expected value, the best thing to do is stay away, despite any economic utility.
Or the next best thing would be like in roulette, which also has negative expected value, you take all your money and you make one bet, where you’ll have the best odds, because otherwise, if you go with a fixed‑budget constraint, you’re pretty sure to go bankrupt. You’re saying that if you play over the course of a lifetime, let’s call it 20 years, the odds against you winning are not astronomical.
Aaron: In a lottery, that’s true. Now, in roulette, that’s not true. A roulette player is virtually guaranteed to lose money over his lifetime. The mathematics for a roulette player happens if. It has to be that you enjoy being at a casino and casinos will pay back a significant portion of your losses in the form of discounted goods and services.
So, it’s kind of a different way of paying for your vacation. One of the things I discovered with people who do this, again, I think to understand, you really have to spend some time hanging around casinos, and you really have to do some of it. It really annoys me, the number of people who talk to their kids about risk, but never actually go to places where people are taking risk or try it themselves.
What I’ve found is people will go, to play roulette for a weekend or something, some weekends they’ll lose a couple thousand dollars and get treated to all kinds of good meals, entertainment and rooms. The casinos are very attentive to the losers, because they want to make sure they come back, they don’t want them to go away dissatisfied.
Then some weeks they’ll win, they’ll hit it big, and spend a lot of money. The pleasure they get from this is much greater than if they went and just spent money. When you win money and spend it on luxuries, it’s much sweeter for people. The casinos have captured a lot of the psychological benefits, and figured out how to market it and make it.
When you talk to people, they are quite aware that over their lifetime they have lost money at casino gambling, and they just regard it as an entertainment expense. They feel they get more fun doing it this way then going to some resort.
Mike: That’s a good point. I’ve seen documentaries that talk about casino design, how things are laid out, the height of the ceilings, and the angle of the lights. When you watch these things, Aaron, although they are in Vegas, you’ll say they are the new Hotel California, where you can check in [laughs] but it’s very hard to check out.
Aaron: A few years ago, I was in Las Vegas, and I was giving a talk. I was staying at the Wynn, and I went over to the Bellagio, and got lost getting off the floor, which is very easy to do.
Aaron: I saw a casino employee and asked could you direct me to the exit, and he asked me, “Why do you want to leave?” I said, “I’ve got a room at the Wynn.” He said, “I’ll give you a better room here. Show me your key, and I’ll give you one suite upgrade. Whatever kind of room you have at the Wynn, we’ll give you one better here.”
I said I have a dinner I’m suppose to go to over there, he said, we’ll give you dinner here free. Everything I said, he had an answer. The one thing I didn’t say is my wife is waiting for me back there. I’m sure he would have offered me a better wife.
Aaron: It’s the kind of customer service you don’t get anywhere else. There’s a tremendous desire to please the customer, which is one of the things I think finance is so terrible at. They’re marketing products that should be good for people, and we advertise the statistics, this is a good way to save for retirement, this is a smart tax decision, and so on.
Yet, only a tiny minority of people voluntarily purchases these things. But we don’t take that to heart, and think about what we can do to solve our tremendous marketing hardships. Why don’t we look at the people who market successfully and take some of the lessons they’ve learned. We just go on thinking people are so irrational and we should cast a law make them do stuff.
Mike: You’ve got my creative juices going. For those of you who don’t know, Aaron played professional poker in the 70s and the 80s, and it’s highly likely that this fellow who was trying to talk him into staying actually recognized him, and wanted you to stay for a few other reasons other than great customer service.
So, let’s say, Aaron Brown goes to the Wynn, and he sees Phil Ivey sitting there, and he’s going to play poker. He’s got his betting system down, he actually has several different ways to play poker. As you know you can play passive, you can play aggressive, somehow you have to manage that risk and figure out for the strategy what would be the optimum amount of money to have in that system. I’m leading into the Kelly Formula. How important…
Aaron: Can I stop you for just a second?
Mike: Yes, of course…go ahead.
Aaron: I’ve got a secret for the audience. This is the best way to play Phil Ivey at poker.
Aaron: Five years ago, I had a match with him on the Internet. It’s very easy to get him on the Internet, it’s hard to get him in person. We agreed we do it at a certain time and we’d play over the Internet and it would be a head to head match. He called me up and said, oh well, I just got this confirmation to go to Atlantic City for a tournament.
But it’s OK, we can have our match, I’ll play on my laptop, from the car while I’m driving down. Every four or five hands, he would go into a tunnel, or something, and it would hang up and he would fold his hand, automatically.
Aaron: Best session I ever had against Phil Ivey. Get him on the Internet, while he’s in a car. If you put him at a table, you’re not going to make money. I don’t care if you’re the best poker player in the world, you’re looking to break even at that point. You’re hoping you’re not going to lose, but he’s not a guy you’re going to making a living playing poker from, at the table.
Mike: Probably a great example of how there is an absolute skill, and it’s not luck. For a guy who has won seven bracelets or some ridiculous number.
Aaron: There is skill there. I don’t mean he’s indomitable, I don’t mean he’s going to absolutely win from you. But it’s not a good life strategy to make that your income source.
Mike: Yeah. You’ll get a good education out of it, so there has to be some other kind of thing.
Aaron: You’ll get a good education watching him.
Mike: [laughs] Yeah, and it’s cheaper, exactly. So, we were talking about has anyone here seen Kelly, and the importance of the Kelly Formula in terms of being able to calculate the amount of risk you would allocate to a certain strategy. Why don’t you bring us up to speed on Kelly, and how it’s used on Wall Street, if it’s at all used, and then we’ll take it to the next step.
Aaron: When people talk about the Kelly Formula, they get sort of obsessed with the mathematical formula, which isn’t the point. It’s a key insight about the nature of risk, and Kelly was this follow who worked at Bell Labs and unfortunately died young. He was also a fighter pilot, he was an expert pistol shot, and fast draw pistol champion.
He lived a very exciting life, a Texan. He started thinking about the right way to bet on horses. He came up with this insight, which is actually kind of trivial mathematically, but I won’t go into that right now. Most people think if you keep more risk, you increase the chance of very good or very bad outcomes. What Kelly showed, was beyond a certain point, increasing risk just increases the chance of very bad outcomes.
There are some levels of risk that are just stupid to take. You can just calculate the amount in certain casino settings, where you know all the variables, and the odds, and so forth. In real life you can’t calculate that precisely. If you take your risk up to the Kelly point, up to the maximum point of risk, you’re guaranteed to do better than someone who bets any other way, in the long run.
The key to success is sizing your bets correctly. People are upping their bets and risk all the time, that’s life. They either make financial bets, like trades, gambling games, decisions like do I quit my job and start a company, do I go back to school, do I marry this person. If you can calculate the right size, you get carried along, you’re going to win some and lose some, but you can have faith that in the end you’re going to end up better off than any other way.
If you don’t think in terms of good bets and bad bets, the question you’re always asking, is what’s the right size. This insight can be incredible valuable to trading, I think. A lot of traders come up to you with good ideas and bad ideas just don’t think about it that way. You say “OK, I’ve heard your idea, here’s the right size.” The right size might be zero or a penny or a dollar, or the right size might be everything you can beg, borrow and steal, but there is just a number that is the right size.
What people tend to do without Kelly, is they put a very high standard on any risk they are going to deliberately take. They just lose out, because 99% of the bets they should be making they’re just not making. Maybe they should be making them smaller, but they are not making them at all. The positive expected value of those bets, over 99 bets, you really eliminate most of your risk.
Because if you flip enough coins, you’re going to get the expected number of heads. When something comes along that they should bet – this is assuming that people are correctly identifying the good bets, which not everybody can do – they bet way too much on it. If they win, they talk about how smart and bold they are.
These are many of the most successful people in the world follow that strategy and people look at them and try to imitate that strategy. They don’t realize that it was a fool’s strategy that just happened to work out. If the person loses their big bet, they blame it on black swarm, an unexpected event, it’s not my fault, I couldn’t control it, but in fact they could.
They could have made a hundred smaller bets, they would have been guaranteed to be in a better place. That’s what Kelly really teaches us how to do. It teaches us the questions we should be asking. In any practical real life bet, you can’t calculate all the odds, exactly.
You don’t exactly know all the payoffs. It’s not like you can take a piece of paper and say, this is exactly the size. You can get pretty good at estimating the approximate size, if the size is approximately right, you’ll do pretty well. If you’re getting your sizes wildly wrong, which most people do, you can’t win.
Mike: Say the number I get from Kelly if I plug in the numbers is 32. What would that 32% mean to me if I have $100 million under management?
Aaron: Kelly asks you, what’s your expected edge, and what’s your risk, and what’s your bankroll. The things that really matter for Kelly are the worst case and your true bankroll.
You don’t really care about your average loss, where you’re not going to lose more than a $1 million on this trade. What that means of course, once you go over the million, you’re going to try to get out, and once you get out you may have loss $1.2 million or more sometimes. You really want to get a plausible worse case through this trade.
When you do that, you find that it’s bigger than your average loss. Which means your actual Kelly number, the theoretical one was 32%, but it’s actually more like 20%, when you really capture it with what the worst case could be. You don’t really have to worry about the average loss, or the “99% of the time losses.”
You really have to think about what if everything goes wrong, like getting out of the worse time or a terrible execution or whatever, if we’re talking about a trade. The second thing is, your true bankroll is usually not just what you have under management. I might have a $100 million under management, but all of my investors are going to pull if I lose more than $5 million dollars.
In that case, I only have $5 million to lose. In a different situation, I may earn a lot of money doing other things, and if I lose $100 million, I can just get more anytime I want. In which case I should be betting bigger. I have learned over the years, a good rule of thumb, how you really should advise somebody, is going over and find out what the true numbers are.
Some my guess is for the adviser who has $100 million under management, is he can’t lose the whole thing. They (the clients) would get out long before they lost the whole thing. So we would talk and we could put up about 20% of the bankroll at risk, and the real Kelly number is 20%, so we should be betting $4 million dollars. Not the $32 million, if we took the actual Kelly number of 32% out of $100 million.
You don’t have to get it exactly right, chances are if you are between $1and $10 million, you’ll do OK. You don’t have to hit the number exactly. The 32% would be certainly over betting, and if you don’t do your Kelly calculations you get people who have this kind of opportunity come up and then they put $50,000 on it, or something and you just let the whole opportunity slip away.
Mike: That’s really great insight because most of the stuff you see on the Internet that’s not written with Wilmott as a backdrop, they really just look at the formula. They drop some numbers in a spreadsheet and derive a Kelly number, which, if that’s all you do, you’re probably a little bit ahead of everybody else.
Aaron: Yeah, you’re thinking about stuff, which is good. But what happens is you think about things until you get the variables and then you turn off your brain and stick it in the formula. You got to understand the formula. Having the formula is worthless. You’ve got to understand why the number is what it is.
Mike: No truer words were spoken during the subprime morass. So you’ve got the Kelly formula. You figure with $100 million you have the mandatory puke point at 50% where you have to give the money back. You really don’t want to get anywhere near that.
Plus you’ve got the emotional duress of being in a protracted drawdown of say 25%, where you’re not going to get any incentive fees because you’ve got to earn that back. And it’s lethal to try to play catch up. So say you set the Kelly at $10 million of your $100 million. Your Kelly number, taking in a worst‑case scenario, is more like 20%.
And then the idea, though, is that that’s just one system with $100 million of assets under management but you might be running several systems. So how does it work if you’re going to trade several systems with that money knowing that any combination of losses that equals say, X is greater than $10 million, is going to give my business a fair amount of duress?
Aaron: Well, I’ll tell you the great thing about Kelly is, people object to it on a lot of grounds. And they usually have something that they find, expected utility theory or Markov’s Optimization or something like that. But the thing about Kelly is Kelly tells you approximately the right bet without having to guess all those things you can’t possibly know.
A lot of these other systems make you specify all kinds of things that you just don’t know. It tends to be true, unless you’re making highly correlated bets, so if you got a bet oil prices are going up and you got a bet natural gas prices are going up, this doesn’t apply. When you talk about a system you’ve got to link everything that’s tightly correlated.
But assuming that things aren’t so…that you got some mean-reversion strategy running on oil, maybe you’ve got some stock, a value stock thing, and then you’ve got something else in some other market with a different kind of velocity behind it, when Kelly gets to be reasonably independent, the right size for the bet is the right size for the bet.
And so you can do each one in isolation, because the one thing you never know much about is the correlation of your bets. It does make sense and you can actually do a full calculation and make some assumptions or whatever. You might decide your Kelly bet should be 20% smaller because you’re running three systems instead of one.
But again it’s pretty easy to just do a judgmental thing and say “OK, I know that before I decided it was supposed to be for a company with my size, I’m running three systems, so I think I’ll just run it at three.” The exact number just doesn’t matter that much as long as you get the sizes approximately right.
And the relative sizes are very important, which is something people are very, very bad at doing.
In fact, even professional traders in my experience will typically tend to bet the most on their worst ideas. I don’t know why that is. I actually have this theory for it. And it’s just this farfetched speculation. I think of it as you got these little men running around in your head. Some of them have great ideas and they’ve really thought through carefully and they’re absolutely right based on true, rational analysis.
And these guys are talking in normal tones of voice explaining what they should do. But then you got all these nut cases and crackpots who are running around screaming ideas. So you’re going to pick and choose among them and listen to these little men and what they’re going to do. The one that you actually think is best, the one that is shouting the loudest and that you happen to focus on among all the others that are shouting is probably going to be one of the crackpot ones.
And the ideas that pass muster, they’re good enough that you listen to them and are really right are typically quieter voices. So hearing a trader is just rock solid convinced on the ones to bet the house on, I’d like to tell them make that your smallest trade. I can’t do that. But I can say “Look, don’t make that any bigger than these other ones. You’ve got these four ideas that are really just as good. Personally, I think they’re probably better.”
And if you bet equally on all of them you’re going to do fine. But if you bet the house on the one you love the most, you’re really making only token bets on the others you’re going to be in big trouble. I can’t defend this. I’m not a psychologist. I haven’t done any experiments for that.
But I guess that a long professional experience tells me that if you bet equally on all your ideas that you’re willing to move forward on instead of trying to guess which one is best you’re going to be better off. You can actually train yourself to figure out which one you want to bet the most on and bet the least on it you’ll be even better. But that’s pretty hard to do.
Mike: You strike a nerve close to my heart because I think being that we’re humans first, traders and risk managers second, it’s oftentimes very hard to not just go with the emotion of it even though we’re both trained. I’m not in your league but we’re trained to know better.
Now when you were saying what you just said about the empty barrel makes the most noise and they also put the biggest trades on the worst ideas, immediately I saw Dick Fuld’s face out of in a scene from Larry McDonald’s awesome book, “A Colossal Failure of Common Sense.”
Towards the end of the book Fuld’s just saying “we need growth at any cost, keep adding leverage, keep buying more real estate regardless of the price.” I have no understanding how someone at the C‑suite could just be so blind to risk management, so egomaniacal, and have no sense of self-awareness.
Had I behaved the same way Fuld did, there would be three people knocking on my door: the “F”, the “B” and the “I”. And I’d be talking to you right now from Leavenworth, Kansas.
Aaron: No, no, you are right and you could put John Corzine in that club. If you look at Lehman’s businesses you can see lots of really profitable, great businesses which got smashed because they wanted to bet the house on something they didn’t seem to have any real fabric in.
If you had told me Lehman was the leading house of commercial real estate for 30 years and that that was the backbone of their profits and so when times got tough they decided to bet everything in the firm that their commercial real estate would pull through, I would say “OK, I can understand that bet.” When times get tough, you go to your best performers.
But they chose something that had been a sideline that they never really made much money at. They had other things that they were really, really good at and had been for many years. And the tragedy in Lehman, there’s a lot of tragedies, but one of them is the number of good businesses that got smashed because they were yoked with bad ones.
But the other thing that is going on there, I think, is quite possibly that there’s a certain double or nothing, deer in the headlights bet. And you feel like “OK, if I pull out at this stage and take my loss, I can’t bear the thought of that. I just can’t.” And that’s usually very stupid. Usually when you pull out, you take the loss and five minutes later you’re just feeling great. You got it off your back…
Mike: No kidding.
Aaron: You’re embarrassed but it’s done. Given that I can’t pull out, I might as well double up because if this thing keeps going south I’m going to get killed anyway. So I might as well make myself a hero if it does work out. I don’t know to what extent that was part of the psychology but that’s very, very easy to do, especially in a group.
You get a group of people together and left to their own devices each one would quietly just pull the plug and take the loss and live to fight another day. But nobody wants to be the first person to say that [in a group]. Nobody wants to be the guy who went down before the real battle.
So sometimes you get organizations that can’t do it even if individuals can. There’s a great book I’d really recommend, I think it’s one of the great books of the 20th century, which was General Eisenhower’s book about World War II called “Crusade in Europe.”
And one of the thing he says, he says he had tremendous respect for the German officer corps, especially on the offensive. And it really surprised people who did incredible things during the war. But he said once they got on the defensive they had this tremendous difficulty. They were unable to go into any battle without committing every possible resource to it. They never held back.
Everything they did they did with 100% effort, which was a tremendous strategic advantage for him because he always knew the full measure of their strength. He never had to worry that there was anything more. So he could wait and see what they did, know their strength and then he could just send a little bit more strength and always get it right.
These people weren’t stupid. They had been running a big army. They had been very successful for five years. They were experienced, whatever, but they were institutionally incapable of changing their risk posture. They had been very successful with the posture of throw everything at it, blitzkrieg. Drive through France in two weeks, drive through Poland in a week.
And when they had to switch to play defense, they just were totally incapable of doing that. And I think this is also true. You got a big run‑up on Wall Street from say 2003 to 2007, where everybody is making a profit doing everything. Like the German army in ’39,’40, ’41, you’re winning all the battles and you’re doing things people have said were impossible.
You get to be institutionally incapable of switching gears and saying “OK, now there’s a bear market, the credit crunch. It’s time to trim exposures and look to minimize losses instead of trying for maximum gains and be you’re really honest and transparent with everybody because we’re going to need their trust. And if we start shading the truth a little bit now they’re not going to believe us when we need the credibility.” And it’s just very, very difficult to make that switch.
Mike: Yeah, again while you’re saying that I’m thinking in terms of Wall Street when you’ve got your back to the wall. And I remember reading somewhere, I can’t remember who said it, but it was very pithy and something I’ve meditated on for 20 years and that is “When you have a large position, you have to get out when the market will let you out, not when you want to get out.”
And I always look at liquidity as one of the heuristics to measure when I’m trading because the last thing I want to do is be involved in some thinly, thinly traded security subject to sharp moves where the slippage and skid will be a fantastic expense to both my nerves and to the portfolio.
And so when I think about the real estate portfolio and how the whole world was kind of levered in these levered portfolios of debt, of mortgages, as well as holding property which at least on a spreadsheet never had anything worse than a soft landing – whatever that means.
I also get concerned for the risk managers out there who are trading these types of dealer‑driven markets where I have to pick up the phone, call “Aaron” and talk you into buying something as opposed to seeing the liquidity or the depth of market on the screen.
I think that can be very, very perilous. And for the analyst who comes into trading and not from a poker standpoint of knowing how to take good risks and knowing the difference between good risks and bad risks, it’s very easy for the human being to rationalize a bad position or a bad market and kind of get into this hope that it will come back and it will be different.
Aaron: That is absolutely true and the dealer market thing is a very good thing, too. Especially if you might be trading something that in good times you can do electronically or maybe you deal with a dealer but it’s just pro forma, phone call, price, done.
And it’s something you get in time, yeah, where you’re talking people into stuff. It’s a totally different game. And you understand they know their position and they’re offering you a bad price today, but if you don’t take it they’re going to be more desperate tomorrow and they’re going to offer you a worse price.
I just can’t stress enough how important it is to cut positions when they’re too big. And what that gives you is that gives you the ability to stick into it. If you cut early and you take some losses and you have terrible execution and you had a big bid‑ask spread, you’re bleeding money on commissions or bid‑ask spread, whatever, and you feel bad about it. But now you have a position you can hold. And you can hold for the rebound.
Aaron: The best thing in the market is for being the strongest we can be. So you’re the guy who’s strong enough to hold on through all the pain, all the pain, all the pain but finally have to just get out. Wiped out all your positions, blow it up right before things come back.
So as soon as things start going in the can you’ve got to say to yourself “OK, am I sure that I can hold this position through anything that might happen?” And if you’re not, get rid of 20%, or 50%, or 80%, or 100%. Stick with the size you can hold through anything. Not only can you survive, which is the main thing, but you actually end up with more profit than the person who waited and sold later at a worse price.
Mike: It’s kind of like having a pair of fives or a presto in position against Phil Ivey. You might as well get in and see the next card, see if you can flop a trip and take it to the next hand. If not, you can puke it out and risk your blinds or whatever. It seems like another good analogy on how to take a good risk versus a bad risk.
For me, and I’m saying this without any preparation or knowing what you might say, I look at winning 3 times my bet size, 40% of the time and losing 1 times my bet size 60% of the time as emotionally equal. Because for me, I’m never going to get to be Cliff Asness or Aaron Brown if I do something really stupid and say to myself that “it’s different this time” because I am involved in the trade.
The feeling comes from the discipline of taking both of those trades, of being able to take profits when you have to, and there is a time when you have to, as well as taking those consistent small losses and dealing with your attrition of capital during those periods of time, which of course is a drag.
But there’s really no other way to do it. When I say that to my students, they’re like “I don’t understand how you can feel equally happy about losing 30 basis points 60% of the time as you do winning 70 basis points 40% of the time.”
I say “Well, the goal is to survive. The goal is longevity. I’m not emotionally invested in the outcome of any one instance or any one roll of the dice.” Am I onto something or am I such a rookie that I should just quit and cash in my chips now?
Aaron: No actually, and by the way his comes through beautifully in your book and it comes through to people who might not think about gamma man, gamma quant. What you’re really saying in another way is the Kelly insight.
Anybody who worries about a gain or a loss, anytime one trade is going to make or break you, then you’ve already lost, you’re in a bad station. What you got to do is say “OK, I don’t really care about the next coin flip because I’ve got 1,000 coin flips to go. And I know that if I win more when I win than I lose when I lose and I win higher than 50%,” now you’re actually getting a slightly different thing.
You’re actually losing more than you win but your wins are enough bigger to make up for it that you’re going to be good. You’re going to be fine. You’re going to make money on your personal money and your business is going to grow and your vocation will grow. And you’ll have job offers. You can play it a lot of different ways.
If you don’t do that, let’s say you bet everything on one roll of the dice and you win. Well, OK, all you’ve done is you’ve made some money but you haven’t really gotten anywhere ahead in what you want to do. And of course, you might lose.
So you have to enjoy losses, like you have to go out and all cheery like “I’m happy I lost that money.” But this is something you say in your book that really got to me, you just really got to have this feeling. You learn you can put in a good bet and you lost but you didn’t lose much. You didn’t lose more than you can afford and you’ve got opportunities to make bets again. And if you make enough of these, you’ll win.
The poker analogy that I use is you get all your money in when you got the best of it. Then the other player draws the one card on the river that can for them, and takes all your money, you’re happy. And the reason you’re happy is “OK, fine, if that’s how this guy is going to play, I’m going to win a lot of money in the long run.”
“I don’t care about losing this pot because I’m going to win everything he’s got. I might just be a little slower maybe because this pot that I got my money in was the best of it.” In the long run that’s what matters because the cards are going to even out.
Mike: One of the best books I ever read in my life, and I know you know this book and you probably know the guy, a buddy of mine was a market maker in S&P 500 Futures, the big daddies, before the E‑minis came out and he was at Susquehanna.
He recommended a book to me by David Sklansky called “Getting the Best of It,” probably one of the best books that you could read about knowing when to hammer when you have a high expected value and knowing when to muck your cards.
Aaron: David is a great guy and I know him. And actually what’s kind of funny about that is we were members of the same generation. We were a bunch of college kids in the mid to late ’70s and early ’80s who were playing a totally different kind of poker than anybody had ever played before and we were winning at it.
It took us a while to really believe that because we’d go in with these guys. You’d sit down with this Texas road gambler who had been doing this for 40 years and knew all the tricks and was shrewd and had great psychological insight. And could laser his eyes through you and make you think he was reading your mind.
But we’d just say “Well, I know a little game theory and a little math and if I keep making this bet over and over I’m going to win.” And we did. We changed the game of poker. People today who look at the game don’t realize how fundamentally non‑quantitative and non‑theoretic it was back in those days. People played a totally different game. The only books on poker were written by bridge players who had no concept of the game.
If a poker player wrote a book it wouldn’t make any sense to anybody because they didn’t have any coherent theory. They didn’t have anything to say other than some aphorisms. The only good book about poker written by a poker player was “The Education of a Poker Player” by Herbert Yardley. And he actually was the first. He wrote that in the ’50s.
He was actually the head of the U.S. cryptanalysis effort in World War I, coincidentally. So he was a math guy who played serious poker and that was the first book where somebody really put some math into it and started really thinking about it theoretically.
But it wasn’t a very good strategy book. He didn’t actually get very deep into the game, mostly because the people who were playing it were so bad that he could get away with stuff that wasn’t very deep. Getting back in the ’70s and ’80s we really had to play against some good people and use some theories.
It was tremendously influential on us in the sense that we said “OK, we’re inexperienced, wet behind the ear college kids who took a little bit of math and we won.” And we started saying “Well gee we can do this anywhere. We can do this in the stock market. We can do this in other life arenas.” And it changed a lot of things.
Mike: I think that David Sklansky and Mason Malmuth probably wrote the best books on poker, as well as your first book The Poker Face of Wall St., which was very interesting, too, about Wall Street and poker.
Before we completely run out of time I want to segue into value at risk. And you go from Kelly to value at risk, what people call VaR, for those of you not listening not familiar with it.
And how folks like AQR or Caxton or Tudor, folks who have a substantial amount of money under management are using value at risk. Your take on it is one thing, Nassim Taleb says it’s useless or dangerous. Explain folks how you go from Kelly to value at risk and then…
Aaron: Nassim is another guy. He’s a friend of mine. He’s right, it is very dangerous. And it probably has been misused more than it has been used correctly. But it is essential. It’s really a key to modern risk. In fact, value at risk and Kelly together are the pillars.
When value at risk started out, it’s a long story. I won’t go into all the historical detail. But what it is is you try to guess each day and then a loss number such that one day in 100 you’re going to lose that much or more the next day. So you say “OK, I’ve looked at all the numbers. I’ve looked at my position, this is what the market’s like” and so forth. “And I think there’s one chance in 100 I’ll lose more than $1 million dollars tomorrow.”
Then you check. Are you right? Do you in fact lose more than that amount one day in a 100. And you also want to know that the days are completely unrelated in time. You don’t want a “Well, sure if I lose it one day then I’m going to lose more than that in the next two or three days as well.” You want it unrelated to the level you don’t want “Gee, I always lose more than the VaR amount when I set the VaR amount low.”
Basically what you want is like a point spread. You want a number so that nobody could make money betting against you. So every day you really have that 1 in a 100 chance. And the first thing you do when you start doing this is you realize you didn’t know anything at all about your center risk. You didn’t know anything at all about your risk. Your systems were terrible. You didn’t have the data you needed for things. You were just really terrible at predicting how much this might lose.
It gets you humble. You work and you work and you work, and finally you get pretty good at it, where you can set a good point spread and you’re right more than you’re not. I’d compare it to you might think football. You hear some loudmouth on the talk radio “The third string left defensive tackle ought to do this and that and then we’d win all our games.”
They sound really knowledgeable because they have the names of all the players, the positions and all the terminology. But they’re not. The guy who really knows something is the guy who is setting the point spread in Las Vegas. He has to know something because the world is betting against him, if he’s wrong he’ll get found out.
The minute you start doing that, trying to set a point spread in sports, you realize you didn’t know anything about sports before you tried to do this. You didn’t have any of the data or the statistics you needed. The first and greatest advantage of VaR is you get your systems in order and you learn something about what you’re trading.
But it has a second advantage as well. It’s virtually impossible to lose a lot of money unless your VaR goes up or you lose more than your VaR amount. So you always get some kind of early warning for things. You do get a lot of false alarms. But that’s OK as long as you get the warnings.
Where VaR gets dangerous is that people start thinking of VaR as a risk measure. It’s the worst risk, the most you could lose. And of course it isn’t. One day in 100 you expect to lose more than that amount. It’s only telling you what happens 99 days out of 100. It doesn’t tell you what happens on the 100th day and that’s where your risk is. It doesn’t tell you what’s going to happen on your worst two or three days of the year, it only tells you what happens on all the others.
So when people start thinking of it as a risk measure they start doing foolish things. And they start thinking “My risk has been quantified. So it’s safe, it’s under control.” But it isn’t, 99% of your risk has been quantified, but that’s not the risk you care about.
That’s not the thing that’s going to blow you up. What VaR really does is VaR splits everything into two regimes. You say “OK, 99 days out of 100 the markets are going to be normal and I’m going to lose less than I expect to lose and I know this going to happen. I can get a lot of data and optimize things and run my business according to statistical principles.”
“And one day in 100 I’ve got no idea in the world what’s going to happen. And on those one day in 100 I’ve got to fall back on some more robust things to survive. I’ve got to start using stop losses. I have to have big capital reserves. I’ve got to do all kinds of things that are going to get me through those days.”
The reason this fits into Kelly as I mentioned earlier is that what’s really important to Kelly is your worst loss. So a lot of times people look at Kelly and they’ll try and estimate their average loss or something like that. They’re really doing these calculations within the VaR, the days where they’ve got plenty of data where they don’t have the really bad losses.
But things that really determine your Kelly amount are the things that happen on the 1% of the worst days. Because like I said, the Kelly bet is very sensitive to worst outcomes. So what you’ve got to do is you’ve got to say “OK, I think there’s one chance in 100 I’m going to lose more than $1,000,000 dollars tomorrow.”
Now, and these things are very hard, “If I lose more than $1,000,000, what does that mean?” Well, maybe it means that everybody who’s trading my strategy is going to be blown up and selling it out. Maybe it means that there’s a major disruption in some market. Maybe it means that my margin requirements can get doubled and going to have to get optimism.
Maybe it means my biggest investor is going to redeem. You just think for all the things that might happen on that day, conditional on you losing more than $1 million. If I think, OK, if I really pull the sales and trim the risk as fast as I can, once I get to this point, here’s the point I can lose money to. I’m just going to run my business normally until I lose this much.
Once I lose this much, I’m going to cut back half the risk. Once I get back to this point, I’m going to cut the other half, and going to make some very conservative assumptions that things go badly. I’m going to say, OK, I think I can really wrestle it down, so I won’t lose more than $5,000,000 in any one bet. You got to be 100% sure, you got to kind of pick a number.
That $5,000,000 loss is what you got to put in the Kelly number. A lot of people would do the VaR, put the $1,000,000 VaR and say that’s my worst loss, and that’s just not true. That means you’re going to bet five times as much as you should bet because your real loss that you could get it down to is $5,000,000. It also tells you where you got to start cutting.
You can’t start cutting when you lost $5,000,000. You might not even have to be able to wait until you lost a $1,000,000. You might have to start cutting when you lose $500,000 if you want to keep the loss to $5,000,000.
These are things you can reason out in calm times, come to a decision. You can communicate to your investors or to your bosses or your risk managers if you’re an institution and get everybody on the same page and have any arguments you’re going to have then. The worst thing you can do, get to the midst of a crisis, being having arguments about the philosophy of risk management.
Aaron: You got to get it done beforehand. You got to say, OK, fine, you believe you never cut losses. That’s great, but we had that argument and you lost. Now get on the team and start doing your part. People will do that, because they were heard out.
You got another guy who said you always have a stop loss. Maybe he lost the argument. OK, maybe there are times we’re going to go through our stops. There’s no right or wrong answer to this stuff, but you got to decide in advance. You’re always going to make the wrong decision in a crisis.
Mike: It would seem to me that miscalculation of Kelly can be absolutely catastrophic when you go back to the dealer driven markets that we were talking about. Because here you’ve underestimated your risk in the first place and now you’re in a market where you’ve got Don Corleone with Luca Brasi who is going to come and give you a deal the first time.
And you’re going to say “no” and then you’re going to come back and he’s going to guarantee “that either your brains or your signature is going to be on the contract” the second time.
By that point, you’ve put your whole entity at risk and you could likely lose your job because it shows that you don’t know what you’re doing. All from a simple overestimation of both your confidence emotionally, and underestimation of your risk, which seems like it could happen very easily.
Aaron: Yeah. One thing I tell, this is not so much for people trading for themselves, although the same thing is true, it’s even more true in institution., I say “you’re a young trader, 28‑year‑old guy, you’re making tons of money, you got a net expected present value lifetime earnings of some astronomical figure. That’s what you got to protect.”
If you take a $1,000,000 loss here, you could get fired. As long as people believe you’re under control, you’re not going to have a job, you’re going to do fine. People will give you money to invest and raise money. The one thing people won’t tolerate is a guy who’s out of control.
If you say well, yeah, I blew up, all my positions went wrong. I got out, my investors got 80% of their money back. I’m not saying that’s good. You’re better off making than losing money. People say, OK, there’s a guy who is in control. He was tested and he didn’t double up all his bets, he didn’t lie about his position, he didn’t try and go for broke.
On the other hand, you get a guy who does, maybe he goes for broke and he wins. He ignores all the stops, he holds onto his position, things turn around and he makes twice as much money as anybody else, then he finds someone that will invest with him or he’s a trader at a bank he gets fired. He can’t make a lot of money.
People really, really want to know you’re under control. As a trader you’ve got a tremendous lifetime to come to terms, if you’re any good at it at all. You can make all the money you ever possibly spend in your life as long as you don’t blow it off. The way you blow it all is by proving you’re out of control.
You’ll lose money a lot of times if you’re under control.
Mike: I think, again because you’re Aaron Brown, and I’m not, that is pretty much a much more eloquent answer to the folks who are out there that are trying to trade and want to understand why I’m equally happy taking 60% small losses and 40%, because I’m looking for an allocation.
Allocators are going to ask to see your daily equity runs for a whole bunch of reasons. Like Aaron says, on some level when you’re running money with a 2 and 20, or a 1 and 20, or any kind of incentive fee, you have a free call option for the most part to make an ungodly some of money.
Protecting your franchise is really the goal in certain ways.
Mike: OK. You set up a $5,000,000 VaR, and then something happens and the bad bund offering Wednesday in Germany, people back off, market sells off, and you have what’s called a “VaR break.” First of all, explain to everybody what a “VaR break” is. Then, I want to know, because you’re doing this, what does Aaron Brown do during a VaR break, or at a VaR break?
Aaron: OK. You have your bund position, you said they’re going to have us offering in the bund and everybody in Europe is fleeing to get quality and the bund is the last quality in Europe so they’re going to have a good auction and the bond will sell for a lot. There’s one chance in a hundred I’m going to lose more than $1,000,000 on my bet. I’m OK with that.
Well you have to be OK with a VaR loss because it’s going to happen a lot of times, two or three times a year. Actually you have to be OK with a much bigger loss. You do that and all of a sudden the auction fails, the price drops and you’re down $2,000,000 on your position. Now, totally in defense of the VaR, you also presumably had some kind of a stop loss on this position.
Let’s say you didn’t hit, let’s say you set your stop at a $3,000,000 loss, or something like that. Now you’ve lost $2,000,000. What you figured something is OK, a VaR break’s a warning. A VaR break is telling me that maybe it doesn’t really matter what happened yesterday, because yesterday we hadn’t had the VaR break.
I got to think about what this means for the market and I got to think, OK there are a lot of banks out there that sold all their Greek and Italian debt and then put it in German debt because they thought it was safe and suddenly they’re going to think it’s not safe and they’re going to sell it like treasuries.
You got to start thinking about what could happen and you’re going to say, all right, I think that the way this plays out, the Euro could drop to 120 and German interest rates could rise to 4%, whatever you think. You’re just saying that the world has changed. I got to get a handle on this new environment.
Now you might say, I have no idea what this means, and then the answer is easy. You solve for positions, you go flat. Usually what happens is you say, OK, now I got this new situation and now I think the VaR’s are probably much bigger. If I keep my position, I thought I would lose $1,000,000 yesterday, I was wrong, I lost $2,000,000.
Tomorrow, I probably think I could lose $5,000,000 because the nature of the world is just a lot more uncertain. Now I might say, OK I want to trim my position because it’s just a lot more volatile position. I can make or lose a lot more money, so at a smaller position I can actually get the exposure I wanted before.
I guess the other thing you do is look really hard at your VaR model. What is this really just 10 heads in a row in coins, or is the coin rigged? Do I really just get a one in 10,000 draw from the distribution or did I misunderstand what the distribution was. Was there some risk that was missing somewhere?
I think one thing just for this particular example, I started thinking really hard about how many bunds are actually being bought by Bundesbank? This actually had a lot of implications for the Repo market in Germany and these things were not big factors before.
Use the model so you can go back and say, OK, these things are now factors I want to account for, probably do some research and sometimes you find out, gee, if I had really understood these factors before, I would’ve set the VaR limits to $3,000,000 or something like that. Now that I know about these factors and can put them in, I now believe I’ve got a more accurate model. You don’t always have time to do that one unfortunately.
Sometimes these retrospectives, you have to wait until after the trading is done and everything’s calmed down a little bit. But what you do is important. I think of this two ways, you have to learn a lot to put a VaR model together. I’ve said before, you got to have systems, you got to really understand your risk. Once you do that, you learn a lot.
Because somebody who didn’t compute a VaR might think I just lost $2,000,000 cause I didn’t make the prediction in advance, they’re not going to have anything to say that my prediction was wrong, so I don’t learn.
Making predictions that could be wrong, let me go back and fix things, and you’re always fixing things. The goal isn’t that you have the perfect VaR system that wins an award, the goal is to have a system that tells you when you got to learn something. It hits you over the head and makes you improve things.
Mike: That brings us to scenario testing. I asked you, it would seem to me, when you’re getting into these scenarios and testing “what could happen if…” you have to really start thinking about the outlier events.
What are the things, I don’t mean outlier, like Nassim means outlier in the tales, but I do mean things that are unexpected that would just catch a lot of people off guard. They might not turn into 20% hits to your portfolio, but it would surprise people.
Now maybe that’s a black swan specifically, but I think there’s a difference between a 9/11 black swan and something like a bad bund offering on a Wednesday before Thanksgiving. Maybe it’s just me, but talk about scenario testing and how important creativity comes into that when you are a front, middle, or back office risk manager.
Aaron: My favorite story about scenario analysis is, this is actually not in a financial firm, but this was a management exercise. They take a managers off away to this off site. This was a company that did this kind of thing for training. They said you’re going to have this scenario.
The scenario was, they divided the people into two groups, and they said you guys are the plant, and you guys are the corporate management in another city. He stops and he’s mature in explaining this problem and we got different information because one’s in the plant and one’s in the corporate office.
The corporate office, in order to get information from the plant or to send instructions to the plant, is only allowed to send one person. They can have one emissary who can go over and talk to the other group and back. The guys from the corporate office were two hour exercise.
The guys from the corporate office got the first 45 minutes going over there and discussing what to do. They came up with a plan and some information and they elected one guy and they sent him over to their guys in the plant where he was supposed to gather the information and give them instructions.
They wait and they wait and they wait, the whole two hours goes by and the guy never comes back. What had happened was the guys at the plant got really pissed off that it was 45 minutes and no one had come to talk to them. When the guy showed up, they held him in the room, they blocked the door, and they wouldn’t let him out.
Aaron: This is the kind of thing when you’re doing a scenario analysis. Now, nobody would ever imagine that in advance. Nobody would ever say, gee we have to worry about confrontation. This was a much stronger..the guys who set up the scenario, they’re a consulting firm, they said that had never happened before.
Nothing like that had ever happened. Everybody involved in that exercise learned a lesson far more important than any kind of corporate business management stuff about how you handle plant problems. If you don’t communicate with people things get really royally screwed up. I find when I’m doing a scenario analysis test, I really try to put in a lot of detail, a lot of seemingly random stuff, because you never know what triggers things for people.
Give people their room and get them chatting a little bit and then I’ll try and make it as realistic as possible. Suddenly this news has come and I’ll do where one guy gets one version and another guy gets a different version. Maybe one of our systems is broken and we can’t get some information that was supposed to be available at the time.
You’ve got to get people reel into it. They really got to be imaginative and playing in it, doing the arguments they’re going to have at the time. You just find out tremendous amounts of things about people. It’s not just…I mean, you’re also working through the scenario and you’re sort of saying, “OK, would we be able to do something you’ll always find?”
I have never run one of these things where you didn’t find out some unexpected detail that no one had thought through that turned out to be important, that’s key. You can’t do same day settlements for this kind of security. We used all of our plans and it’s sort of something that you’ve got to assume that you’re going to do same day settlement to get the cash for this.
Or, yeah, we have this but we couldn’t make the foreign currency translation that it took to make the transaction work. Or, it would be over the exchange limit for concentration in this contract if we did what our contractor said they did. Just finding them and working through these details is so important. I can’t tell you I never had a situation where you found a detail and then sent it over to analysis and it later saved us.
But what is does do is really impress on people how uncertain things are, that when you cut the crisis it’s not like what you thought it was going to be. You never had all the information you thought you were going to have. There’s always this one important decision maker who’s not in touch. There’s always some wrinkles in the thing you didn’t understand, that you can’t do.
That just gets people the respect for, “OK, we’ve got to have extra reserves. We’ve got to build ourselves an extra time. We can’t run this thing so everything works perfectly. We’ve got to assume that some of the time things don’t work.” I find these scenarios very, very useful.
The least useful part or it is the number you come up with at the end because that’s totally arbitrary, it depends on the scenario. But people having been through a scenario, it’s the next best thing to having actually been through a disaster toward getting people to think right before the facts and to really get people to understand you’ve got to get things right now.
The risk manager is always there nagging people and complaining, “This report is late. This number is unreliable.” But the thing is the one day you need the report it’s got to be right. If you don’t get it right every day it’s not going to right on the day you need it.
Mike: Amen. This answers a question I get asked a lot from people who are soliciting me for their clearing and execution business. They say that they can get me 30 milliseconds faster and that competes with the high frequency traders and “dah, dah, dah, dah, dah.” I just realized, in speaking with you, you talk about the things that come up when you’re doing the scenario testing.
It occurred to me that the reason I pay three bucks a round turn is because I’ve got a live person on the phone who I’ve been dealing with for 18 years, going all the way back to the days of ED & F Man. I realize that when I’m putting my orders in a couple things happen. One, I’ll be like, “I want to buy 50, five‑zero, December Gold at 1800 on a stop.” You know, he’ll read it back to me.
Right then and there we are able to make sure that there’s no errors in the order itself. You’d be surprised. You put in thousands of orders and I’m likely to put on at least one boneheaded order in, like one that I delineated in my book, where you’re short and you’re trying to cover and you go short and you double down. So, boneheaded things happen.
I also realize that I’m able to do a little scenario testing. It’s rude and it’s crude compared to what you probably do at AQR, for example, but it is a form of scenario testing. When you’re a one or two person shop and you have people who have 20-something years of execution behind their back they bring a dearth of information to you that you can’t get from Trading Technologies or from some depth of market machine on CQG or Bloomberg.
You need that other person there to kind of say, “Well, don’t forget this and don’t forget that and First Notice is over here.” Like you said, it really helps if you’re humble enough to listen.
Aaron: Yeah, I agree. I certainly go back and forth on the human versus computer trading because you’ve got good and bad points both ways. Humans make a lot of errors but they tend not to make ridiculous errors. Machines don’t make as many but they’re perfectly happy to do ten million instead of ten thousand.
Aaron: You build checks into the computer but sometimes the checks are what caused the problem. I totally agree with you on the experience of just saying the trade out loud to an experienced person. It also matters…you’ve got a really crazy market and you’ve got a flash crash kind of situation.
If you’re on a computer and you put in an order you’ve got no idea what’s going to happen to it. With a person you might get some feedback on that. On the other hand, I remember and have very strong memories of the crash of ’87 where it was just impossible to get any dealer on the phone.
So it was the only way you could execute was electronically because the whole system was so clogged up. It’s something you’ve got to think about is if you’re in a market crisis am I going to be able to get this guy on the phone.
Mike: So now you’ve done your Kelly. You’ve done your VaR testing. You’ve done your scenario testing. And now someone has to go to your star trader, who’s up 200% year to date, and tell that guy or gal that “we’re cutting risks in half because of the nature of the marketplace and where we are and also that we want to play a little poker with our track record because we want to clean up the year.
It’s been a tough year in 2011, all the trend followers are getting hammered and if we can finish the year up. We’re up 800 basis points net the year, net or fees expenses, slippage, skid, and the whole gambit. We’re cutting risks by 40% across the board. We’ve got to go to this top trader who undoubtedly is a strong personality.
What’s that conversation like?
Aaron: It is quite different. Right now, at AQR, we’re a quant shop. With quants it’s really easy because the great thing about quads is quants think in terms of return. They don’t think in terms of dollars. Yet, compared to you they made 5% last month. I’m really happy. It doesn’t matter whether they made it on a dollar or a billion dollars. They just don’t think in those terms. When you cut the risks you don’t even know this.
If the return doesn’t change they just want to run their model. We told them they couldn’t run their model, that they were getting closed down. That was hurt.
Qualitative traders aren’t like that. Qualitative traders all think in terms of dollars. So cutting their limit or cutting their positions is cutting their lifeblood. It’s like you’re taking a knife to them.
I guess there’s two ways it works. The best situation is, you’ve won their respect and you’ve been in there with them every day advising them and sizing every position but helping them. They feel like, “OK, I made 50 million dollars this year but it would have only been 40 and it would have been a lot more volatile along the way if my risk manager hadn’t been there coaching me, helping me, pointing things out, keeping things going.”
When he comes in and says, “Look, it’s time to pull in your claws a little bit and let the market flow. You can put it back on later,” you’ve got that respect. He’s going to do it because he trusts you. Sometimes it’s not that. Sometimes it’s either because of you or because of him or the combination. You’ve never really won that respect and it just got a rule.
I have worked for organizations where there’s no thought of debate. At a well run place the risk manager comes and says, “You’ve got to cut your decision path.”
Well, it’s like the police behind you flashing their lights. You’ve got to pull over. You may not agree with it. You might not have thought you were speeding. But you’ve got to pull over. And the guy just knows that if he blows you off he can get fired. I can tell you about Paul. He may holler at you and scream but it’s just fear because he knows that I can just, I don’t know. What’s going to happen now?
If you worked at a place where the traders don’t respect you and you don’t have the backing and a guy can blow you off and get away with it then it’s hopeless. Then I say, “Why were you working there in the first place? Why were you working at a place that didn’t respect you or the guys who were risk managing or the management behind you?” You just quit and go do something useful with your life.
Mike: Wow. That’s very insightful. I don’t think a lot of folks would fully understand what that dynamic is like. You talk about these relationships, about scenario testing, and risk management or whatever. It’s almost like when you’re inside a big firm, whether it’s qualitative traders or quantitative traders, it is like a big family.
Communication is key. Respect for one another is key. It starts there even though you’re dealing with some fairly intense mathematics in stochastic calculus or Ito’s Lemma and whatever, that’s just the language that you speak. At the end of the day it’s, do you get along with people, can you listen, and can you do as you’re told…
Aaron: I’ll tell you though. I tell you, the problem I did have in my career is, what I do is, and this happened to me a few times in different ways. I’m in a room and we’ve got the head of systems and trading and we’ve got the CEO of the firm and we’ve got the executive committee or whatever. We’re talking about some really big position that’s big enough that it’s got to go to the executive committee.
I’ll say, “OK, I’ve been over this,” and dah, dah, dah, dah, dah. I’ll go through my analysis and I’ll say, “OK, we’ve really got to put a billion dollar limit on this thing. That’s the most we could do.” Then the head trader will come up and say, “Well, Aaron, I know that makes a lot of sense but I’ve looked at this a little bit I got a market judgment here and I think we can go to two billion on this one.”
Then the head of the desk will go, “Yeah, I’ve done my numbers here and I think it’s about two billion too.” What tends to happen is I find that those get kind of equally. You get respect. I’ve never been in a situation, if I was I would quit, where people say, “Hey, I’m not going to look at your number. I don’t even care about mine.” People really do listen.
I don’t mean they’re just paying lip service because you’re the risk manager and they’ve got to do it. They understand it and they look at your calculation and they know you’re right and they have to pay attention to it. But you end up with three or four people voting for the two billion and you’re voting for the one billion. It really is only two votes there.
There’s like the quantitative mathematical analysis and any quantitative mathematical analysis would come up somewhere where I am, it might be a little higher or lower. And you got kind of a market intelligence feel number. But because there’s only one quantitative and four or five qualitative traders there, and it’s not that they’re even, these are not people who always say, “We’ve met everything in our pathways.”
These are people who have made a successful career out of correctly sizing bets. You do tend to get outvoted just for that reason. I’ve always felt that we could improve things a little bit if we just said, “OK, if that’s the case we’re going to split. The quantitative says this. The qualitative says this.
We’re going to cut the difference and do one and a half billion on this thing.” I don’t think I’m right more than they are but it gets me when we end up doing $1.8 billion or something because there are four of them and one of me.
Mike: I see. In other words, there almost has to be a model for how much weighting do we give the risk manager versus how much weighting do we give the actual trader or the head of the desk. Then, what kind of haircut are we also going to take into account for the market tone or other extemporaneous things. It’s very interesting, Aaron. I appreciate your candor there.
Again, those are conversations that not a lot of people, even if they are traders, are going to be in, in those kinds of conversations. You kind of give us a peek of what it’s like to be in those meetings. Let’s talk about something in your book…Red Blooded Risk is illustrated with some great comics by Eric Kim (link to video with comics) that are very, very clever.
In one of them one of the refrains is, “The combination of obvious technical expertise and psychotic overconfidence was, as always, irresistible to the bureaucratic mind.” Where did that come from?
Aaron: Yeah, the characters in the comics were composites of people I had met over the years and there is a certain kind of quantity rather. I think this is a stereotype a lot of people have, and I haven’t seen the movie Margin Call yet. I want to see it but I suspect that’s the kind of quantity that they have in those movies.
A guy who is icy cold, really good at math, can compute things right. It’s impressive because the guy can really do a lot of math and do wizard things with computers and do all kinds of magic things, feel like a trade, he can make, for some period of time, make absolutely steady profits, his hedges were perfect and all that. He was just duped and when people like that get wronged, they get wronged magnificently and gigantically.
It is very, very hard to reduce that kind of person. People talk a lot about if you’re a professorial kind of risk manager who is used to people being polite, keep voices soft. You got strangers screaming obscenities at you and they’re at their desk throwing food at you, it could be very hard to deal with. That is true, you know a lot of risk managers who are just a little too polite and timid for the job.
There’s the opposite, there is a certain kind of quant that can be just very, very difficult for people who aren’t mathematically inclined to hold back or to disagree with, because the guy seems to know so much math, he’s so sure. That can be really disastrous, where an overconfident quant is a real danger in the world.
Mike: I started trading commodity futures through what was then known as ED & F Man. They had been around since the 1780’s. You would never think that having your assets custodied in such a place would actually be a risk that you’d have to take into account, either with the Kelly formula or VaR or through scenario testing.
But now it seems if you run, whether a qualitative fund, a quant fund, or a combination of both, who your clearing member is either as a broker‑dealer, a Forex dealer, or an FCM, is actually something you have to scenario test.
We are in a world that revolves around hypothecation, pledged assets for collateral, which are levered up and traded as swaps and turned into CDS’s which no one is paying on now.
I say this on the heels of Oliver Sarkozy’s comments which you may have seen or not, where he said “the problems in Europe are 10 times what we had here in the United States.”
Are we better off buying GE and Disney and holding our assets now in our own name as opposed to street name?
Aaron: Well, one of the things I sort of remember ruefully that you had asked me five years ago, would you ever trade with a broker‑dealer whose CDS spreads were in the triple digits. I would’ve said you’re crazy. I can’t imagine a world, it’d be impossible. Can you imagine a world where the spread between your overnight borrowing rates for banks and LIBOR was 100 basis points … the whole financial system would collapse.
Those things all happened and things didn’t collapse and we were trying to deal with them. It is a radically different world from a risk manager’s standpoint. Often, especially for lower risk transactions, if you’re buying treasury bills or something, then there’s far more risk that they’ll be some operational problem or a problem with the custodian or even someone got repo-ed out and not returned. Then there is the actual risk of the instrument itself.
A lot of things we used to assume, there’s something people use in finance all the time, they called a risk‑free rate of interest and it underlies a lot of financial theory. There is no risk‑free rate of interest. Dollars might blow up in hyperinflation, the U.S. Government might default, and Euros are clearly risky, Chinese currency is currently risky.
It’s your goal to is not to make or lose a lot of money in that. There really is no risk‑free that changes things. I think that in the long run, hypothecation and general sort of derivativization of the economy is good. It’s a more efficient way to run things and it frees up a lot of capital and it’s much more rational.
Our systems have not kept pace, both our legal system and our computer systems have not kept pace with the complexity. If you really had good systems, a lot of these problems wouldn’t happen. MF Global is, of course, a perfect example of something where the records are screwed up, at least at this point.
We don’t know this for sure, at least at this point it looks like nobody stole any money, it’s just that they didn’t keep it segregated like they were supposed to. It’s going to take years to sort out, and they’ll lose money as a result. Things weren’t segregated and they were going to the wrong people.
If we had better systems and a better legal regime, I think we could make this work. On the other hand, I totally sympathize with somebody who wants to get physical assets, or at least physical cash securities held in his own name in, even a bank vault or safe deposit box, where he knows where the actual security is because you really don’t know what’s happening with your assets out in the world.
I think reading the Examiner’s report, the bankruptcy examiners report on Lehman, if you weren’t a risk manager, you’d just be shocked at how much, I don’t want to say fiction, how much virtual reality there was in reading these financial statements. How little physical cash, physical securities, physical assets were there, I mean, you look at this balance sheet and all this nice, neat, adds up, you got hundreds of millions of dollars of assets.
When you look at the reality you find a lot of those assets are actually owned by other people. Possibly with some exceptions, but I’m not talking about accounting fraud here, I’m talking the rules. They have things that are only controlled by other people on their balance sheet and their liabilities are actually controlled by other people.
A lot of it was virtual stuff, a lot of the stuff they owned or been endowed, a lot of stuff that they had actually physically in their possession or stuff they had borrowed from other people on repo.
The kind of person who’s got their idea of a bank because there’s a vault and there’s assets in the vault, and securities, and cash, and there’s liabilities that are like bonds that we can look at and say “OK, this is a deposit, this is a bond.”
That’s just so far from the modern financial reality. Unfortunately a lot of our legal system and regulations seem to be predicated as if that was the way things are.
Mike: I’ve been watching the MF Global thing closely because I’ve always tried, even though I trade commodity futures and spreads, I always think of somebody else’s money and the trust therewith. It’s kind of like there trusting me with a newborn child as a babysitter. There are certain things you have to take into consideration when you run somebody else’s money and it has nothing to do with the gains and losses. It’s the fact that they trust you in areas that they probably have no clue about.
I’m a pro and I’m a professional reader, and I’m still learning things both from your book and in today’s conversation. I can only imagine what the business owner down the street who’s running the Mailboxes Etc., pulling down $200 grand a year who wants to make an allocation into a structured product of something that would be in the realm of what you and I do for a living.
They’re completely clueless on hypothecation. They don’t even know what they see, hypothecation, what does that mean? Is that a drug store? They don’t know, so I really take it not as a personal affront, but it really boils my blood when I see these types of things happen because MF Global had the resources to have good books.
My own take on the thing is that if they were not segregating client funds, and they co‑mingled funds, the chances are that the money wasn’t stolen. It may have been deliberately misappropriated or not. But if it had to be put up as collateral inadvertently, or not, and everything was then marked to the market, to me, I don’t think the money’s going to be found anywhere.
I think it was put up as collateral, it was marked to the market and it was transferred to whatever broker‑dealer, Prime Broker or FCM that was on the other side of the trade. It’s not going to really come back, but I guess more will be revealed.
Aaron: Yeah. Usually fake money doesn’t come back although the Madoff trustee has managed to come up with more than I thought, so maybe things are different. I will say one of the changes in finance, I’d like to go back to the very beginning of the 1980s, people think of financial people as being dishonest. There is a certain kind of honesty you really do find in finance. I’m serious about that.
Anybody who’s in this for the living professionally, I’m not talking about people who get in for two years and scam somebody or people who are pretending to be in finance and stealing people’s money. When I started out there was a very, very strong appreciation for, this is other people’s money. There is a strong idea that funds were supposed to be figured, they had to be segregated, a real understanding. You didn’t manipulate the market, you didn’t lie, you didn’t defraud people.
Now, you did a lot of things that most people might consider dishonest, you sold things that you knew were overpriced, you bought things that you knew were under priced, if somebody came to you in trouble you took advantage of it. There was a certain kind of honesty you had, whereas I think a lot of professions have. There’s ways you’re honest and there’s ways maybe you’re not.
But this kind of thing was really bred in the bone. It was something people really understood. It wasn’t something you had to teach them to do. It wasn’t something that you had to “Here are the rules, you got to do it.” It was just naturally the way you thought about things and the way you acted. Anyone who didn’t accept that just wouldn’t be trusted, would never be successful.
I think what’s happened is first of all we’ve gotten a lot of people, a tremendous increase in hiring, including people who 20, 30 years ago would never have been interested in finance at all. They just kind of think about the world differently.
And the other thing it has gotten lots and lots of rules. Unfortunately, I think a lot of the rules killed the basic honesty because you can’t just tell someone “Listen to your gut and do what is right.” You got to say “OK, here’s 15 rules about what is and what isn’t a legal trade, when you have material non‑public information, when can you short something, when can you sell or trade this or that instrument.”
I think all the rules killed a lot of the intuition and feeling so people don’t have this second nature that you just don’t do things. They say “There’s a rule you’re supposed to follow,” and you got to find the rule, not “Whose money is this.” I think that’s a problem that’s really got to be solved. It really has to get back to people thinking in basic, honest terms. And I agree it really makes me angry in a way that other kinds of things don’t because segregating something is not difficult, it’s not complicated.
People who are in trouble with money are calling and screaming at you for money and whatever. But it should no more occur to you to send segregated funds as collateral for a different entity than it would to steal your grandmother’s social security check. It should be something just totally beyond the pale that you just never do.
Mike: Yeah, exactly. So let me do a little scenario testing with you. If I was one of the regulators or the accounting firms brought on by the trustee to go over the books, yeah, you can say that you’re lost in a deluge of trade confirmations and ACAT transfers and wire transfers. I think at the end of the day you have to take into account what was going on over that weekend around Halloween.
I remember a scene from “A Few Good Men,” and I’m completely speculating here, where Tom Cruise wants to come in and change the plea and actually assign new counsel because he can’t get the two Marines to take the plea bargain so they “can go home in six months.” Everyone expects him to show up at court the next day and basically resign.
Instead he enters a plea of not guilty and then rhetorically he says, looking up to the sky, “Why is it that a lieutenant junior grade who has zero experience in a courtroom gets assigned a case? Is it because he’s not supposed to see the inside of a courtroom? It’s supposed to get settled and quietly hushed away.”
So when I see a guy who’s 35 years old as the CFO of a company my guess is he was there to not really challenge John Corzine. John Corzine having been a fairly prominent guy at Goldman and then an elected official, my guess is that he wanted to run the thing with an iron fist and pretty much have a CFO who was going to do as he’s told.
Now I can’t say that he did anything illegal because I don’t know. But there seems to be something rotten in the state of Denmark with this.
Aaron: I will say is that what would be the worst outcome here is a billion‑two disappears and nobody goes to jail.
If it’s the case that one guy was powerful and so he appointed a weak and inexperienced person to a position and the weak and inexperienced person made a mistake, you can find an excuse for any one person, but you got to somehow find accountability somewhere. It can’t be one of these things where it’s just so squishy that you can’t really find out who to blame. That kind of money can’t disappear without somebody committing a crime. It just shouldn’t happen.
But I suspect when they do investigate, hopefully they’ll find the money. If they don’t, when they do investigate I suspect they’ll find 20 people contributed in different ways, nobody was clearly criminally liable, nobody had a smoking gun crime, but this person was a little sloppy, that person made a mistake, that person was inexperienced, that person looked the other way. And that is the kind of thing you can’t tolerate.
Mike: Not in this environment.
Aaron: It would be better if they just found somebody who stole the money. They find out it’s in somebody’s shoebox. He just ran off with it. That’s a clear crime, the guy goes to jail, the world goes on.
I am reasonably confident, I would guess, that these firms are in contact with regulators and have explained and perhaps a trustee and have explained the situation, showing what their records show. Even if MF Global had no records at all, you should be able to reconstruct it from people who transferred money in and out. But I would not expect one of them to go public and say ‘Hey, we might have $100 million of this money in here,” because that would be a touchy thing to do.
Mike: Right. All right, last question. You teach, I teach, you’re also involved with Paul Wilmott and the CQF. I’ve seen lectures you’ve done. I’ve read articles in the magazine. And I want to compare it to preparing people to manage risk on Wall Street. In the early ’80s the way you differentiated yourself was to get an MBA. That was going to help your resume show up in a different pile than those who just had baccalaureate degrees.
Then people started going to night school getting MBAs. They added the CFA if they wanted to be an analyst or some type of fund manager. And recently, and I say just near term recent, recent history, Paul [Wilmott] created the designation of a CQF, Certification in Quantitative Finance.
Help people understand, as a teacher or as a mentor to the next breed, how you would compare and contrast getting an MBA, a degree in Quantitative Finance, versus what you can do with a CQF.
Aaron: I’ve given some talks for the CQF program. I’m not actually a professor in the CQF curriculum. We have a lot of quantitative financial degrees now and actually degrees in quantitative risk management. The CQF is, I would say, a more elite program, it’s taught by really some of the top practitioners in the world. It’s a very small group.
And the people who have completed it, they’re held to a very high rigorous mathematical and practical standard. And one of the things I’ve heard from them is the connections you make in the program are really invaluable. And then they also have sort of a more mass market, to take hundreds of people, some of which are very good, some of which are not so good.
I think that the MBA has never gotten a lot of respect. I actually have one myself but I didn’t earn it. I got it as sort of a booby prize for being in a PhD program for two years. It’s got a lot of very soft courses, some of which don’t have a lot of relevance in any business situation. It was always a way to show you were serious about business and to pick up a few buzzwords and things.
I think there are some valuable courses in it and you can actually learn some things in an MBA program if you want to. But you can certainly get an MBA without learning anything useful, for finance anyway, possibly for any kind of business.
The more quantitative financial degrees are primarily offered by people who are already very good at quantitative things, and want to apply it to finance. That can be kind of dangerous. I feel that I would rather have somebody who knew something about taking risk, who had a feel for finance and money and was willing to learn a little math to do it than someone who is really, really great at math but you had to explain what money was and what taking a bet was and how it felt doing it with money and so on.
I think most of the people in these programs are driven by their desire to get a job. I’m good at math. I hear finance duties a lot. I don’t think that’s a good career choice for people.
On the other hand, if you really love finance and you want to get into it and you can handle the math in one of these programs, it can be. It certainly will get your resume pulled up higher. It will get you opportunities to do things, the opportunity to trade, for example.
Traditionally, and I think this true for you personally and people of my age in general, the way you got a trading job was you either started trading for yourself, you got a seat on an exchange, you leased it, you found someone who you could be their apprentice. And you just did it purely on your own or maybe you went to a bank and you got a job hanging around the trading desk fetching meals and filling out trade tickets and adding things up afterward. Then eventually they let you trade a little bit.
And nobody asks for a resume, certainly they never asked for any kind of quantitative trading. And what it meant was the people trading, some of them were good at math, some of them weren’t, but all of them loved trading and all of us understood it. Today it’s much harder to get in that way. Certainly at a large bank they’re going to ask you what is your degree, how much math do you know, things like that.
It is possible to find somebody who will apprentice you. It is possible to lease a seat on an exchange. But I think those ways are getting harder and harder and less and less open. So you kind of do need some sort of entre. If you don’t have a personal contact, the quantitative route may be the second best way to do it.
Mike: Well, guy to guy I just want to say I’m grateful for you. Thanks for speaking with me today for so long. You write great books. And I’m humbled by how good this new book, Red Blooded Risk is.
Aaron: Thank you very much. Talk soon.