1/ Red-Blooded Risk: The Secret History of Wall Street (Aaron Brown)
"From 1982 to 1992, rocket scientists hollowed out Wall Street and rebuilt it. Most people, including most people working on Wall Street, didn’t notice the fundamental change." (p. 6)
amazon.com/Red-Blooded-Ri…
"From 1982 to 1992, rocket scientists hollowed out Wall Street and rebuilt it. Most people, including most people working on Wall Street, didn’t notice the fundamental change." (p. 6)
amazon.com/Red-Blooded-Ri…
2/ "Your bets must be as independent as possible.
"Search hard for new things to bet on, unrelated to prior bets. Avoid falling into habits.
"Size bets properly. Don't lose so much that you’re taken out of the game; but be willing to bet big for the right gambles." (p. 7)
"Search hard for new things to bet on, unrelated to prior bets. Avoid falling into habits.
"Size bets properly. Don't lose so much that you’re taken out of the game; but be willing to bet big for the right gambles." (p. 7)
3/ "Optimizing requires goals and constraints and requires that the two be interchangeable. One way that can happen is if both are measured in money. It also turns out that any time you use the same units for goals and constraints, you create a form of money.
4/ "There is a security that pays $1 tomorrow if life existed on Mars one billion years ago (nothing otherwise). There is some price at which you will buy or sell this security. According to de Finetti, that price is the probability that life existed on Mars a billion years ago.
5/ "It’s subjective; someone else could have a different price. But there is always a definable probability: you can always be forced to name a price at which you would buy/sell. Saying you don’t know the probability of something is saying you don’t know what you think." (p. 10)
6/ "It is not the significance level that tells you the reliability of a frequentist statistical claim; it is the vigor and sincerity of the falsification efforts undertaken in measuring the significance." (p. 16)
7/ "A coin could stick to the ceiling or land on its edge, you could win and not get paid, or you could get paid but dollars might be worthless at the time—or a 100% tax on gambling winnings could be passed while the coin is in the air, or you could die before the coin lands.
8/ "You might assign probabilities to all of these things, but it’s hopeless for practical decisions.
"We made the empirical discovery that the sum of the probabilities you could define reliably was often more valuable than the statistic you were originally estimating." (p. 17)
"We made the empirical discovery that the sum of the probabilities you could define reliably was often more valuable than the statistic you were originally estimating." (p. 17)
9/ "Exponentials work both ways: The smaller it is, the more slowly it grows. Once it starts getting big, it grows so fast that it seems to come out of nowhere.
"By that time it has lost its exponential character: nothing physical can grow forever. It hits up against some limit.
"By that time it has lost its exponential character: nothing physical can grow forever. It hits up against some limit.
10/ "People describe it as a Black Swan and focus on the sudden growth and spectacular collision with its limit. Concentrate on the exponential nature instead. Once the thing becomes obvious, it’s usually too late either to avoid its danger or to exploit its opportunity.
11/ "Once they’re big enough to matter, they never last long. When a CEO targets a compound average growth rate for earnings, she’s trying to build an exponential. She will probably fail, but if she succeeds, the company will soon hit a limit; the fallout will be unpredictable.
12/ "When an economist justifies a policy using projected future growth rates, he’s relying on exponentials to bail out an idea that cannot float on its own.
"Alarmists use exponential growth rates to conjure sky-is-falling scenarios that would otherwise be laughed at." (p. 18)
"Alarmists use exponential growth rates to conjure sky-is-falling scenarios that would otherwise be laughed at." (p. 18)
13/ "Explaining away errors and taking credit for accidental successes leads to more acclaim than making good risk decisions.
"Good decisions appear as alternating complacency and erratic actions. They are hard to defend even after the fact to people who were not involved.
"Good decisions appear as alternating complacency and erratic actions. They are hard to defend even after the fact to people who were not involved.
14/ "Bad risk management is ingrained into social institutions and popular theories.
"Half of good risk management is just eliminating bad risk management. This can be extremely disruptive and generate strong reactions because it challenges a traditional base of power." (p. 20)
"Half of good risk management is just eliminating bad risk management. This can be extremely disruptive and generate strong reactions because it challenges a traditional base of power." (p. 20)
15/ "Prices can move more than they ever have in the past, you may not be able to trade or even determine prices, institutions can suddenly fail, rules can change, assets can be expropriated, intermediaries can turn out to be frauds—and those are only the things you can imagine.
16/ "You have no idea of the probabilities, nor how they will affect you. Nevertheless, the less leverage you use, the better off you’ll be if something unexpected happens. You will also likely benefit with more liquid assets, more solid counterparties, and firmer legal ground.
17/ "The problem is how to balance well-understood, quantifiable factors with unexpected things that you can't describe. Is it worth a 10% increase in leverage to get an extra 1% expected return? If an illiquid bond pays 0.50% more than a liquid one, is it a better deal?" (p. 31)
18/ "For normal events, you have plenty of data to make reliable quantitative conclusions. But long-term success requires at least surviving the abnormal, unexpected, uncommon events you know will also occur. The greatest long-term success requires exploiting them." (p. 32)
19/ "A low-VaR portfolio is not safer than a high-VaR portfolio; it means only that there is a smaller region in which you have good data about risk. And VaR is the least you will lose on the worst 1% of your days, not the most you can lose." (p. 34)
20/ "The biggest losses of a casino one year:
1. The owner’s daughter was kidnapped and held for ransom.
2. A long-time employee, for no discernible reason, stuck mandatory tax reporting forms under his desk for many years instead of submitting them, resulting in a large fine.
1. The owner’s daughter was kidnapped and held for ransom.
2. A long-time employee, for no discernible reason, stuck mandatory tax reporting forms under his desk for many years instead of submitting them, resulting in a large fine.
21/ "3. A tiger mauled one of the casino’s star performers.
"Only when we learned to set a VaR limit and make provisions for events outside it could strategies survive long enough to benefit from risk ignition.
"Suddenly we got a slew of hedge fund billionaires.
"Only when we learned to set a VaR limit and make provisions for events outside it could strategies survive long enough to benefit from risk ignition.
"Suddenly we got a slew of hedge fund billionaires.
22/ "These people followed the same well-known, trivially simple strategies. But now they knew how to manage, and to ignite, risk. Financial institutions went from earning 15% of corporate profits to over 40% and doubled their share of GDP.
23/ "They didn’t change their business models or discover new markets; the increase was due to changes in risk management. And, despite popular opinion, the gain did not come at the expense of the rest of the economy but added to overall economic welfare." (p. 38)
24/ "Money measures constraints and goals in the same units, meaning that partial successes in the goal reduce constraints on future activity. The more you win, the more you can bet. That isn’t true if winnings are not exchangeable with the things wagered." (p. 38)
25/ "Add some randomness to your learning and experiences. Religious scholar Greg Barton noted that the more you read, the less certainty you find. The people with the most narrow and rigid views have generally read the least." (p. 47)
26/ "The Prisoner’s Dilemma demonstrates that people following rational strategies in the simplest possible games can arrive at a bad outcome. The best outcome can only be reached by irrational choices. That calls into question the definition of “rational.” " (p. 53)
27/ "We can be precise because we have a good mathematical theory for coin flips and can repeat the experiment. But we have no mathematical theory for predicting World Cup winners. Because the experiment cannot be repeated, it is not even clear what probability means." (p. 58)
28/ "One of the admirable characteristics of the gambling subculture is that you are expected to be willing to bet on any assertion you make. If you back down, you lose status. Anyone who shoots his mouth off about things he doesn’t know about usually ends up disgraced and broke.
29/ "In comparison, the civilian world sounds like a bunch of ignorant loudmouths.
"If you assert a degree of belief, you should be willing to bet on it: in the long run, the frequency of your wins should guarantee you a profit." (p. 61)
More on this:
"If you assert a degree of belief, you should be willing to bet on it: in the long run, the frequency of your wins should guarantee you a profit." (p. 61)
https://twitter.com/ReformedTrader/status/1302337159303540737
30/ On sports betting: "It was easy for quants to identify profitable bet opportunities: spreads were set to equalize money bet on both sides, not to reflect the true probability of winning. (The organization did not want risk; it wanted a guaranteed profit whatever the outcome.)
31/ ""You could make consistent money with strategies as simple as betting against the Los Angeles Lakers National Basketball Association team at home, because Los Angeles was a large, rich, and high-betting city, and the Lakers were a glamorous team." (p. 63)
32/ "You cannot understand the economy without understanding the markets, and you cannot understand the markets without trying to beat them.
33/ "Portfolio managers are salespeople who attract AUM and collect fees rather than generating excess profits with those assets. Most of what you read about finance comes from the salespeople. They also run the firms and shape government policies.
34/ "PMs constitute the majority of guests on financial television. Some are smart and some are not, but the job depends on interpersonal ability, not intelligence. With some exceptions, they haven’t the slightest idea of the economic function of finance." (p. 68)
35/ "There is a small group of people who actually do finance: they make bets.
"An independent financial quant in 1980 had two big problems: getting the data necessary to make good bets and placing the bets on fair terms. Information hoarders had the business sewn up.
"An independent financial quant in 1980 had two big problems: getting the data necessary to make good bets and placing the bets on fair terms. Information hoarders had the business sewn up.
36/ "You had to work for a big bank to get access to data, and it wasn’t electronic. Only starting from a bank trading desk could you get a trading account with the low cost and flexibility necessary for quant trading.
"You could also do it by buying a seat on a public exchange.
"You could also do it by buying a seat on a public exchange.
37/ "A critical mass of computer-literate quants started to mine information and share it on the dial-up bulletin boards.
"We wrote handbooks for opening accounts and protecting yourself from the tricks banks liked to play. The result was superior to anything banks had." (p. 70)
"We wrote handbooks for opening accounts and protecting yourself from the tricks banks liked to play. The result was superior to anything banks had." (p. 70)
38/ "Most people take a few large risks, which dominate their life outcome. If they lose, they blame the Black Swans that they created.
"Had they taken thousands of positive edges and sized bets properly, the law of large numbers would have virtually guaranteed success." (p. 75)
"Had they taken thousands of positive edges and sized bets properly, the law of large numbers would have virtually guaranteed success." (p. 75)
39/ "Kelly allows you to make decisions based on approximate knowledge of what a single asset will do over a single time period with no reference to utility functions at all. Markowitz said that if Kelly’s principle conflicted with utility theory, he would reject utility theory.
40/ "In the end, however, he decided they were consistent and that Kelly’s work gave a maximum amount of risk any investor should consider (but that most investors would prefer to risk less than Kelly)." (p. 78)
41/ "The real problem with the individual investments was not that they were undiversified; it was that putting 100% in any one was overinvesting. Table 5.2 shows how you could have done with perfect foresight, if you had invested the correct amount in each commodity." (p. 79)
42/ "The optimal risk in the ex-ante max Sharpe portfolio turns $1,000 into $8,145 after inflation. This happened when six of the seven commodities returned less than the risk-free rate and the one positive return was only 10.6% for the entire 40 years." (p. 83)
43/ "Hold investments with different time scales and different kinds of risk: stocks of different types, private equity, venture capital, real assets, high frequency trading, relative value strategies, global macro strategies, high grade bonds, distressed bonds." (p. 85)
44/ "Belief in MPT CAPM helped make the markets more efficient cross-sectionally: returns on asset classes aligned with their risk levels. But arguably, MPT CAPM contributed toward prices diverging from fundamental value. Index investors don’t ask what something is worth." (p.96)
45/ "Slight changes in assumptions can make huge differences in valuations, and the cheaper the commodity, the greater the potential for increases. These facts have been observed time and again whenever people are paying for exponential growth—real or imagined." (p. 111)
46/ "We can explain part of the increase in tulip prices as a decline in the value of silver money. Something used as money becomes monetized and acquires a value above its intrinsic use value. When a substitute comes along, the original substance can fall back to its use value.
47/ "This can explain apparently frenzied trading among people with no knowledge about tulips (as well as the popularity of futures contracts).
"Someone who wanted to borrow money, which meant holding a net short tulip position, had to pay a very high effective interest rate.
"Someone who wanted to borrow money, which meant holding a net short tulip position, had to pay a very high effective interest rate.
48/ "These rates make sense in a country in severe capital shortage. Since usury—charging interest on loans—was technically illegal, a rapidly deflating currency was a handy backdoor way to reward investors.
"I see some form of attempted money creation in all bubbles." (p. 115)
"I see some form of attempted money creation in all bubbles." (p. 115)
49/ "Without good money there will be far less innovation—not because people don’t have the resources to devote to research, but because risk cannot be managed. Money does not just make exchange and specialization easier; it makes constraints and goals the same." (p. 133)
50/ "The potter's gross profit margin will not be a steady 50%. Some pots will be failures—100% loss. Others may fetch 5x cost. There will be interruptions & surges in supply and demand. Without money, the potter cannot manage these risks to take advantage of exponential growth.
51/ "He can only guess what scale is optimal. If he guesses wrong, the entire enterprise collapses. He needs money to “carry over” good luck. If he has a successful period, he can grow at a prudent rate and bank some surplus. In bad times, he can scale back and dip into surplus.
52/ "Without money, he will have to make more violent and inefficient changes in scale and will be unable to compute the proper strategy. The business might not survive, even though it has the potential for 50% growth, or it might be run so conservatively that it never innovates.
53/ "To carry over luck, he needs a store of value. Unsold pots are his only option without money, and they take a lot of room to store and have zero yield. He needs to be able to use that store to buy supplies when demand returns. Pots are not a good medium of exchange.
54/ "But his biggest problem is that without a numeraire, he cannot relate supply to demand, the price of pots to the price of clay, or even the price of one input versus another. So he cannot be a Kelly bettor." (p. 135)
55/ "Traders are the ones who forge links among strangers and are the first to share and the last to fight. Equal-value exchange underlies technological and social progress." (p. 154)
56/ "If the Earth were tiny, it would be smoother than a billiard ball. However, at a human scale, there are mountains and oceans. Similarly, markets are efficient for large portfolios over long periods of time, but there are inefficiencies and disequilibria at smaller scales.
57/ "A hydroelectric dam holds water back, moving the system farther from equilibrium. The value is the work done with the electricity, not the effect on the Earth's smoothness. Similarly, in finance, the main value of transactions is the good done with the profit produced.
58/ "Another social value of finance is in coordinating exploitative activities and designing things that can withstand disasters caused by others." (p. 155)
59/ "Professional investors did no better than random (before fees). They even seemed to do worse: they all bought the same things, driving up prices, and all sold at the same time. Ones who beat the market in one year were no more likely than anyone else to do so the next year.
60/ "The total fees charged for investment management far exceed the maximum potential excess profit from beating the market. Inefficiencies can make a lot of people very rich, but if the profit were spread around equally, it wouldn’t raise the average return by much." (p. 157)
61/ "All of us were wiped out on Oct. 19, 1987.
"And boy, were we excited. This was the game we had come to play. We had pushed the market, and it had pushed back. It wiped us out, but we were ready for round two. There were new secrets to uncover, new opportunities to exploit.
"And boy, were we excited. This was the game we had come to play. We had pushed the market, and it had pushed back. It wiped us out, but we were ready for round two. There were new secrets to uncover, new opportunities to exploit.
62/ "Faux quants ran scurrying for cover, leaving the field open to us. Work to revise our models began immediately. An old presentation I had given at Yale a month after the crash contained almost half of the ideas that would be developed and refined over the next five years.
63/ " I was by no means alone; the bulletin boards were buzzing with this kind of thing. The burning question of the day was how to run quant strategies without blowing up, even when the market did." (p. 177)
64/ "Futures contracts, and derivatives more generally, improve on paper money. Instead of an interest rate and exchange rate for every bank, we have an interest rate and exchange rate for every commodity and everything else of economic interest.
65/ "The key to derivative money is the spread trade. Buy March wheat in Chicago and sell in Kansas City? You’ve just lent transportation services from Chicago to Kansas City, assuming the price is higher in Kansas City (if the price is higher in Chicago, you’ve borrowed them).
66/ "Buy #1 wheat and sell #2 wheat? You’ve just borrowed wheat cleaning services.
"You don’t borrow money; you borrow whatever precise future goods and services you need. And you don’t repay in money; you repay in whatever future goods and services you can provide.
"You don’t borrow money; you borrow whatever precise future goods and services you need. And you don’t repay in money; you repay in whatever future goods and services you can provide.
67/ "Transactions are not limited to things people can imagine. I don’t know how to turn flour back into wheat, or move a crop back in time from August to March, but the Chicago Board of Trade has been quoting prices on those services for 160 years.
68/ "You may not be able to find contracts for exactly what you need as inputs and exactly what you expect to deliver as outputs. That’s okay as long as you can find derivatives with a high enough correlation to those things." (p. 184)
69/ "An oil refiner can use the price of refined products minus the price of crude oil as a numeraire; a shipping company can use the price of commodities at port of consumption minus the price at port of production.
"This is the key to the economic function of futures markets.
"This is the key to the economic function of futures markets.
70/ "When your numeraire is your net economic product, you have no risk. When your numeraire is highly correlated to your net economic product, you have small risk.
"Global corporations borrow, spend, and sell in whatever currencies are most advantageous.
"Global corporations borrow, spend, and sell in whatever currencies are most advantageous.
71/ "Generally that means borrowing in low interest rate currencies, contracting future fixed costs in currencies that are inflating, and contracting fixed future revenues in currencies with low inflation rates.
72/ "Borrowing is no bargain if forward exchange rate premiums are high enough to offset the interest rate differential. But except in extreme cases, market expectation undercompensates. The effects of currency choices are large relative to the net income of businesses." (p. 193)
73/ "Since each user has a different numeraire, each one can count a net profit in economic value. Insisting markets are zero-sum is a symptom of not understanding numeraires.
"If speculators went away, the markets would collapse and take vast economic value with them." (p. 195)
"If speculators went away, the markets would collapse and take vast economic value with them." (p. 195)
74/ "Clearinghouses seldom or never go bankrupt; they bankrupt their customers instead. (The first major example was the Paris Sugar Bourse in 1905.)
"When the clearing members get on the wrong side of a trade and lose more money than they want to lose, they change the rules.
"When the clearing members get on the wrong side of a trade and lose more money than they want to lose, they change the rules.
75/ "They accuse winners of trying to manipulate the market, but that’s misdirection. When pushed to an extreme, they set an off-market price for settlement (off-market in the members’ favor, of course). Victims include the Hunt brothers and the German company Metallgesellschaft.
76/ "In 1962, Tino De Angelis floated a few inches of soybean oil on tanks of seawater and used them to collateralize gigantic futures positions in soybean oil and cottonseed oil. When he was discovered, his positions had to be liquidated, and the prices of the oils crashed.
77/ "Members of the New York Produce Exchange declared contracts would be settled at the old high price, changing their losses into customer losses.
"The London Metals Exchange did the same thing twice in tin contracts & is still “the world’s premier non-ferrous metals market.”
"The London Metals Exchange did the same thing twice in tin contracts & is still “the world’s premier non-ferrous metals market.”
78/ "The New York Mercantile Exchange did it in 1976, with Maine potato contracts of all things, and boasts of being “the world’s largest physical commodity futures exchange” today." (p. 196)
79/ "The entire financial system has switched over to derivative money. Cash, in the sense of paper money issued by the government, is meaningless. The cash Lehman held belonged to others and could not have paid its liabilities anyway.
80/ "What mattered was whether it could make its MTM payments and deliver enough assets in repo transactions to get cash to buy back assets it had repoed at other firms. That was a function of asset liquidity and confidence in Lehman rather than green pieces of paper." (p. 199)
81/ "If you don’t know whether your brokerage account is a cash account, it probably isn’t. The securities you think you own aren’t there; they’re lent out to other investors. The cash you think is yours is on the brokerage firm’s balance sheet as if it were its cash." (p. 200)
82/ "Transactions are done on the Internet through bidding and future promises.
"Airline tickets are a natural early transition to derivative trading: the value of seats is unpredictable. There may be empty seats, or demand might be high enough to support 3x the normal price.
"Airline tickets are a natural early transition to derivative trading: the value of seats is unpredictable. There may be empty seats, or demand might be high enough to support 3x the normal price.
83/ "Some people change plans on short notice, whereas others can be flexible for a lower cost. Some want a big seat and TLC, whereas others would sling a hammock next to rabid porcupines to save money. Paper money pricing cannot allocate efficiently under these conditions.
84/ "People virtually build their own computers, and many other manufacturing processes can be done more efficiently this way, satisfying consumer demand precisely and allocating production tasks efficiently.
85/ "Specialist companies and individuals replace work done by big companies. They go to the exchange and see where their services add the most value. No one needs lots of capital, complex plans, or long supply/distribution networks." (p.203)
For example,
thumbtack.com
For example,
thumbtack.com
86/ "Going to medical school next year? Why not sell half the median income of a cohort of medical school entrants similar to you and reduce your exposure to future changes in doctors’ wages?
"The prices for these contracts would be powerful signals for career choices.
"The prices for these contracts would be powerful signals for career choices.
87/ "Also, they allow people to sell future human capital to finance all kinds of ideas.
"The amount of human capital in the world exceeds the value of all other assets. Without derivatives, it remains sterile capital.
"The amount of human capital in the world exceeds the value of all other assets. Without derivatives, it remains sterile capital.
88/ "Someday, you may be able to hedge your life revenues and expenses by buying a contract that pays half the living expenses of a Chicago resident who is in the 95th percentile for income. Then you can stop caring about the value of money and inflation." (p. 203)
89/ "No one could keep track of what went where; no one could know the source of the data. Some transactions were double-counted; others, missed. Front-office people, investors, and regulators never noticed: computers are good at making things look as if they add up.
90/ "Neat reports went out, usually on time. But the numbers were one part reality and nine parts fantasy. Some places were better than others. Investment bank internal numbers were a joke. Commercial bank numbers were merely bad.
91/ "Systems were so bad, even at JPMorgan, that the reports were always wrong and had multiple errors that couldn’t be disentangled and assigned to one responsible party. The front-office people knew they wouldn’t be disciplined for blowing off the back office." (p. 223)
92/ "Financial statements of large, diversified financial firms have little meaning. The SEC in 1998 ordered financial firms to include risk information in their reports, but I’ve never met an equity analyst, credit analyst, or investor who uses the reported VaR data." (p. 227)
93/ "Looking at the accounting statements of a solid (vs. shaky) company tells you little about the chance it will default, nor about what recovery you can expect.
"The balance sheet beforehand will not look anything like today’s, whether because of fraud or unfortunate events.
"The balance sheet beforehand will not look anything like today’s, whether because of fraud or unfortunate events.
94/ "The key insight of quants in the late 1980s was to use precise short-term prediction to decide what to do and the robust long-term prediction for what to do when the first method is wrong. VaR is one of the aggressive, precise predictions.
95/ "It is small harm if you assign a nonzero probability to a scenario that is impossible. It can be fatal to assign a zero probability to a scenario that is actually possible. That’s why you include anything you think is even remotely possible in your cautious predictions.
96/ "If it’s likely, exploit it; if it’s plausible, hedge it; if it’s possible, prepare for it; if it’s impossible, you just bet everything you have on that assessment.
"Allow employees to bet on VaR breaks. All kinds of new information is revealed this way." (p. 229)
"Allow employees to bet on VaR breaks. All kinds of new information is revealed this way." (p. 229)
97/ "In every large, complex system, there are enough input/processing errors that the results would be meaningless except if outputs are checked against reality. The output comes from selective error correction until the result is within bounds acceptable to the system owner.
98/ "To know whether you should rely on a number from a complex system, don’t ask about the inputs or the processes; ask what it is checked against. If the answer is that the number is not checked against anything, or only against other outputs of the same system, it’s worthless.
99/ "If the number is checked against independent systems, it may have some value, but you have to be careful. If the number is checked rigorously against objective reality, then you can trust it." (p. 235)
100/ "People and companies do sloppy jobs sometimes—and usually get away with it.
"There were a lot of bad numbers given out and a very small number of people who cared. Even the people working on projects with me were mostly able to ignore the meaninglessness of what they did.
"There were a lot of bad numbers given out and a very small number of people who cared. Even the people working on projects with me were mostly able to ignore the meaninglessness of what they did.
101/ "One guy used bad data or made-up data but figured it didn’t make any difference to the project. Another guy ran code with bugs but lied rather than admit the error. Someone else misrepresented results to the customer without knowing whether they were accurate.
102/ "And everyone went home at 5:00 p.m. and cashed their paychecks. They say if you work in a kitchen you’ll never eat at a restaurant. Well, I never worked in a restaurant kitchen, but I’ll never believe a number unless it’s something I can validate." (p. 238)
103/ "95% of work done is wasted. Most people don’t notice: they consider only waste at their level. If a truck driver carries a load of goods from a warehouse to a store 400 miles away, he figures he did work. After all, he got paid $300. But how much of that is useful work?
104/ "Some items are defective, others will never be sold, and others don't provide satisfaction to the user. Some will be used in businesses that fail. Some will be moved other places, perhaps back to the warehouse. Some could have been sold for more at the original location.
105/ "By accounting rules, the numbers add up as if the truck driver did $300 worth of work. But it’s fantasy.
"The reason the fraction of useful work is as high as 5% is finance, social infrastructure, & technology. Those can let everyone live twice as well with half the work.
"The reason the fraction of useful work is as high as 5% is finance, social infrastructure, & technology. Those can let everyone live twice as well with half the work.
106/ "One of the spurious appeals of socialism is noticing the amount of waste and thinking that it could be eliminated by top-down decree.
"Refusing to accept waste in one area leads to much greater waste in others." (p. 238)
"Refusing to accept waste in one area leads to much greater waste in others." (p. 238)
107/ "If a local property tax collector says there are 2,000 residences in a town with a total assessed valuation of $200 million, that bears some relation to reality. You can predict the effect of changes in the property tax rate or estimate the cost to install smoke detectors.
108/ "When the federal government says there are 100 million residences in the county with a total assessed valuation of $20 trillion, there are so many definitional issues and measurement problems that prediction based on them might as well be a coin flip.
109/ "Say the federal government imposes a tax on carbon emissions. The actual rules and tax collection will go through dozens of layers of regulators with unpredictable consequences at each stage. This will also induce changes in people's behavior.
110/ "The effects are unpredictable for money collected or for emissions. It is a myth to think of a large institution as a rational actor using reliable inputs to take actions with predictable effects." (p.240)
The Law of Unintended Consequences (thread)
The Law of Unintended Consequences (thread)
https://twitter.com/ReformedTrader/status/1257849272231317504
111/ "Academics took garbage data without validating it. Not one would have bet real money—the very question was offensive.
"There was an extensive literature in economics on how the number of firms in an industry was positively correlated to the frequency of price changes.
"There was an extensive literature in economics on how the number of firms in an industry was positively correlated to the frequency of price changes.
112/ "This was alleged to prove that firms used sticky pricing to fix prices.
"In the government sample, the more firms in an industry, the more firms were sampled. If any firm changed prices, the average changed; surveying more firms produced more apparent price changes.
"In the government sample, the more firms in an industry, the more firms were sampled. If any firm changed prices, the average changed; surveying more firms produced more apparent price changes.
113/ "Adjusting for this made the effect go away. But publications were not rescinded; PhDs were not taken away. People ignored the problem. They had never believed their results in the first place." (p.240)
More on the replication crisis:
More on the replication crisis:
https://twitter.com/ReformedTrader/status/1336398442403307521
114/ "Ford's economist told me 75% of tractors ran on diesel and gave me supporting detail.
"The Dept. of Agriculture guy insisted it was 25% diesel, snorting that Ford was only talking about new sales, not the existing stock. That wasn’t true; I had the numbers from Ford.
"The Dept. of Agriculture guy insisted it was 25% diesel, snorting that Ford was only talking about new sales, not the existing stock. That wasn’t true; I had the numbers from Ford.
115/ "He had answers to all the points I made, giving me his own supporting detail. I checked those with Ford. That guy said Agriculture was counting old tractors rusting in barns or that had been sold to Mexico in the 1950s (not true).
116/ "The more I learned, the more I realized that no rationing plan could work. The amount and types of fuel and the times they were needed varied dramatically from place to place, crop to crop, & season to season. Moreover, people needed to know about supplies well in advance.
117/ "If you really wanted to deliver the fuel necessary to service agriculture in the United States, you would need a nationwide organization with a large degree of local autonomy and massive databases, and you would have to partner with tens of thousands of businesses.
118/ "Of course, nothing like this was contemplated. The idea was that farmers would be issued ration coupons.
"The whole U.S. government standby plan to ration fuel was based on wild oversimplifications and terrible data." (p. 242)
"The whole U.S. government standby plan to ration fuel was based on wild oversimplifications and terrible data." (p. 242)
119/ "The subject had a dial to control refrigeration to keep the temperature within a band such that meat would not spoil. Simple mechanical thermostats have done an excellent job since 1660. Almost all people, even with lots of explanation and practice, do a terrible job.
120/ "Instead of waiting 5 minutes to see the effect of a change, everyone keeps moving the dial because he doesn’t see an immediate decline in temperature.
The other near-universal mistake is to wait until the temperature is at the right level before turning refrigeration down.
The other near-universal mistake is to wait until the temperature is at the right level before turning refrigeration down.
121/ "At that point you’re going to wildly overshoot on the cold side, then overreact on the warm side. The only people who can master this extremely simple task are ones who apply explicit quantitative reasoning and have the faith even when it seems not to be working." (p. 243)
122/ "People managing a virtual village devote resources to health care, irrigation, house building, and other projects. Almost everyone creates disasters, regardless of training and replay time. They try to correct short-term problems without considering long-term consequences.
123/ "Everyone in the village dies; the area becomes an environmental wasteland. The refusal to tolerate waste leads to total waste. The refusal to let anyone suffer leads to total suffering. Top-down thinking leads to bottom-up failure." (p. 243)
124/ "In my experiment, each subject controlled a country and could allocate to investment, production, & military spending and could trade or fight with other countries. The subjects were professors and grad students using quantitative modeling in economics or political science.
125/ "The simulation was not designed to be difficult; with moderate actions, the world chugged along. Everyone had complete information (accurate data and knowledge of the equations that generated the outputs). In every run, people quickly generated global thermonuclear war.
126/ "Tellingly, no one was interested in exploring how rational decisions of well-intentioned people led to disaster. Instead, people claimed the simulation was rigged, without pointing to any inconsistency between their inputs and the computed outputs.
127/ "Humans cannot manage simple virtual systems with complete information and power plus lots of training and practice. It would be idiotic to expect them to manage complex real systems with noisy/out-of-date information, limited power, and no training or practice." (p. 243)
128/ "If anyone bought a car, he would look it over carefully and shop for the best price. But that same person was willing to take money to write a plan on which every American’s food supply and economy depended without checking any data or doing any experiments.
129/ "No one even was interested in reading about the experiences of agricultural planners in communist countries. Worse, no one had any interest in contrary opinions. The guy in the Agriculture Department and the guy at Ford just wanted to explain away each other’s results.
130/ "They certainly weren’t going to try to falsify their results. Their interest started and ended at being the expert. The same seemed true of everyone I met working on quantitative projects. Expertise was a fortress to be defended, not a tool to explore reality." (p. 244)
131/ "Congressmen, cabinet members, senior civil servants, generals, and top diplomats were happy to sit down in poker games with me. Some were smart, some honest, and some good-hearted. But none seemed to have much clue about how to do the jobs people thought they were doing.
132/ "None of them had the data to do things right. None of them had any objective evidence that what they were doing had the effects they claimed. None were interested in falsification any more than the quants were.
133/ "Anything bigger than a small town can be run efficiently only with quantitative methods, and those methods require constant, rigorous validation and openness to challenge. Most of all, they require people who are willing to bet on the results.
134/ "In the 1970s, more than half the world was in the grip of totalitarian horror regimes, and no country had ever emerged from communism into prosperity and freedom. None admitted mistakes, even after tens of millions of people died for them." (p. 245)
135/ "Measure an average of things that may be of only marginal interest... don’t tell anyone what those things are, lest they change their behavior.
"Express each number as σ from its long-term mean.
"VaR can be useful measured on things other than profit and loss." (p. 247)
"Express each number as σ from its long-term mean.
"VaR can be useful measured on things other than profit and loss." (p. 247)
136/ "A risk measure has to surprise you if it is to be of any use at all. In finance you are always going to have surprises, and VaR surprises are cheaper than waiting and letting market price movements surprise you instead." (p. 268)
137/ "VaR numbers are used to regulate and risk-manage trillions of dollars of decisions and are computed by some of the smartest and best-paid quants in history. Yet the most popular methods clearly fail on the simplest task, estimating VaR for a constant investment the S&P 500.
138/ "Nobody cares. Bad numbers go out, bad decisions get made, and life goes on.
"The simplest validation exercises would have revealed the underlying sloppy work.
"You need a VaR estimate that reacts violently to any unusual event, and settles down afterward." (p. 272)
"The simplest validation exercises would have revealed the underlying sloppy work.
"You need a VaR estimate that reacts violently to any unusual event, and settles down afterward." (p. 272)
139/ "Conflicts of interest between a firm and its customers are incompatible with long-term success of a financial firm. Investment bankers get fees for advising companies and underwriting their securities. I think this is a business that has to be independent.
140/ "Combining it with investment management, equity analysis, and principal investing means some parts of the firm are competing with customers, both for the firm’s benefit and for other customers’ benefit. This causes dysfunctions in a crisis (such as 2007-8)." (p. 299)
141/ "Risk managers give people permission to fail.
"Before the advent of modern quantitative risk management, failure was usually evaluated after the fact. The failed risk taker would argue that the idea was sound and had been implemented skillfully.
"Before the advent of modern quantitative risk management, failure was usually evaluated after the fact. The failed risk taker would argue that the idea was sound and had been implemented skillfully.
142/ "The prosecution would argue that the initial idea or the implementation was flawed (finger-pointing). Risk managers change the process.
"Now failure is evaluated before the fact. If the risk manager approved the risk, then failure is not charged to the risk taker." (p.308)
"Now failure is evaluated before the fact. If the risk manager approved the risk, then failure is not charged to the risk taker." (p.308)
143/ "Large organizations try to standardize everything. Standardization increases correlation, which creates dangers. Doing things a bit more randomly on the micro level can improve macro-level predictability. Standardization is easy to exploit by rogue employees & competitors.
144/ "Organizations with one fixed system can be slow to recognize the need for change and can find it expensive to make the changes. People tend to assume that standardization results in fewer errors, which is probably true, but the errors can have far more impact." (p. 311)
145/ "To make degree of belief correspond to frequency, measure things in the right units. Reversing that logic, the thing you choose for your numeraire alters your interpretation of statistical evidence. That’s why the nature of money is so crucial to an economy." (p. 332)
146/ "Statistics cannot settle arguments unless people agree on a relative set of actionable prices; on then can probabilities be defined objectively. Statisticians err in not seeing that these are questions of relative preferences, of prices, not of statistical methodology.
147/ "Practical statistical analysis must specify a range of application—what the numeraire is and what it can buy. Otherwise there is no logic to connect frequency with degree of belief.
148/ "And the only range of application you can trust is one that is validated by back-testing. Putting it another way, the only numeraire you can trust is one that can be measured in observable transactions.
149/ "With all of the difficulties and adjustments, no one would take seriously a theoretical computation of error rate. The only reason to believe a poll result is if the researcher or organization that administered it has been right in the past.
150/ "If the pollster says there is a 60% chance of the Democrat winning, am I willing to put up $59 to $41 on the Democrat winning? If not, I don't believe his claim.
"The gold standard of statistical validation for me is opening up predictions for public betting." (p. 334)
"The gold standard of statistical validation for me is opening up predictions for public betting." (p. 334)
151/ "Traditional intermediaries want to keep their customers far away from their suppliers: it let the intermediaries get away with higher fees. The more complex intermediate layers serviced by anonymous entities with mysterious duties, the more money could leak out unnoticed.
152/ "Thirty years of academia failed to persuade individuals to learn enough to get a better deal. It took headlines to do that. The language of public discussion has changed ("junk bonds," "corporate raids") and today is closer to the language of politics and sports." (p. 346)
153/ "Quants know how to create true capital, and you don’t need a printing press or sovereign powers. The keys are derivatives and securitization.
"Persuading people to rent out unused land or put unused cash in the stock market is part of what the finance industry does.
"Persuading people to rent out unused land or put unused cash in the stock market is part of what the finance industry does.
154/ "Capital allows businesses to form and grow, and—hopefully—allows the provider of capital to achieve financial security.
"Securitization takes the institution out of finance—disintermediation is the technical term. Providers and users of capital are linked directly.
"Securitization takes the institution out of finance—disintermediation is the technical term. Providers and users of capital are linked directly.
155/ "That requires financial engineering: most providers of capital want to provide a different type of capital (diversified, liquid, predictable) than most users need (funding for illiquid projects with unpredictable cash flows).
"Bank lending is not an efficient solution.
"Bank lending is not an efficient solution.
156/ "There is no true liquidity; the bank just hopes not too many people want their money back at once. There is some diversification, but only in a small geographic area with correlated housing values and incomes.
157/ "Banks were expensive, consuming half or more of the interest paid on the mortgage before paying it out to depositors, and socking the taxpayers for about $100 billion more—which used to be considered a big bailout—when interest rates went up.
158/ "It’s far better to put together a national pool of mortgages in a structure that directs cash flows to different investors at predictable times. It can be standardized for liquid trading to make rebalancing and cashing out cheap and easy.
159/ "Borrowers pay less on mortgages, and investors earn more on mortgage securities than in savings accounts.
"This can create an agency problem (the institution making decisions/doing underwriting doesn’t bear the risk). However, any intermediation creates agency problems.
"This can create an agency problem (the institution making decisions/doing underwriting doesn’t bear the risk). However, any intermediation creates agency problems.
160/ "Mortgage and credit securities were a big part of the 2007-09 crisis. The root failing was not models. The primary cause of every securitization disaster, and there were many prior to 2007, is failure to sell the entire deal in true arm’s-length transactions." (p. 361)
161/ "People see the quant risk management only when it fails. It’s dramatic because the type of failure is new and different. But there were more and bigger failures as a fraction of value created before the quants reengineered everything." (p. 362)
162/ "If a long-term investor trades every five years, it helps to see monthly prices. He doesn’t have to wait five years for another long-term investor to trade. The monthly trader wants daily prices for the same reason.
"The highest-frequency trader has a rough life.
"The highest-frequency trader has a rough life.
163/ "The same is true of size. The 100-share investor needs someone willing to take 1,000-share positions in order to be executed quickly and efficiently. The biggest trader has a problem. If she wants to make a trade, she has to do it in small batches over a longer time.
164/ "The fastest and biggest traders demand extra profit to make up for their unfavorable positions. Competition keeps the liquidity up. HFTs' profits are a fee paid by long-term investors for liquidity, one far smaller than what market makers charged in the old system." (p.364)
165/ "The good from financial innovations exceeds the damage. There were scandals and bankruptcies before quants. Overall, the past 30 years have been the best economic time for the globe in history. In contrast, the worst times are decades of stagnation or totalitarianism.
166/ "The new techniques were empowering and liberating. They lifted more people out of poverty than any previous innovations in human history, and they also created more billionaires. That they didn’t help everybody is just saying they weren’t perfect.
167/ "There will never be a shortage of mistakes, especially in innovative times. As long as people fail fast, losses are acceptable and progress is rapid. But when people instead choose to fail slowly—and big—it’s a different story." (p. 365)
168/ "We designed the system to be run by people with experience in risk taking. We believed in finding niche opportunities, failing often and fast, milking successes, then looking for new opportunities. We built structures to support this using quantitative risk management.
169/ "But things got far bigger than we ever imagined. People with no risk-taking backgrounds came to Wall Street to get rich safely. Risk takers shrank in proportion every year.
"The risk avoiders had better credentials (one way to avoid risk is to amass credentials).
"The risk avoiders had better credentials (one way to avoid risk is to amass credentials).
170/ "They had fewer failures in their background. No risk, no failure. They dressed better and had better interpersonal skills. Most of all, they all agreed with each other.
"They wanted to see something solid with a stable long-term track record that was endorsed by experts.
"They wanted to see something solid with a stable long-term track record that was endorsed by experts.
171/ "Unfortunately, a business like this was usually near or at its limit. Risk avoiders love steadily growing revenue and earnings. You can’t get those taking risks. You can appear to get those by forcing exponential growth on an idea that has already hit its limit.
172/ "You can keep up the pretense if everyone agrees: accountants, economists, and regulators. But reality has an ugly way of bursting through the strongest consensus of the most respected people." (p. 367)
173/ "Risk avoiders are good at explaining after the fact. Risk takers don’t explain well before or after the fact (and don’t care). When someone had to leave, it was always the risk taker. Also, they couldn’t get support for good ideas and saw bad ideas inflated beyond reason.
174/ "They were passed over for promotion in favor of better-credentialed people.
"But even if the risk takers had been in the majority, it would have been hard for them to get their disparate and unexplainable views into what had become a formal decision-making process.
"But even if the risk takers had been in the majority, it would have been hard for them to get their disparate and unexplainable views into what had become a formal decision-making process.
175/ "I would like to see smaller, diverse, innovative businesses—including some not considered financial businesses today—offer competing ways to accomplish financial tasks. These might fail a lot, but none would endanger anyone but their investors and their employees." (p. 368)
176/ "Conventional subprime loans did better than expected given the housing decline and recession.
"No-income-verification loans worked for undocumented immigrants and tax evaders who had cash income that could not be backed up by documents. These people had assets and income.
"No-income-verification loans worked for undocumented immigrants and tax evaders who had cash income that could not be backed up by documents. These people had assets and income.
177/ "Young people with education debt had no down payment but plenty of income to service the loan. 100% LTV made sense for them. People with poor credit histories due to divorce, illness, or business failure often had enough assets and income to be decent credit risks anyway.
178/ "The problem was mainly in loans with multiple underwriting flaws. The worst were NINJA loans. Overlooking one flaw in an application that has strength in another area makes sense, but when an application consists entirely of flaws, there is no reason to overlook anything.
179/ "In some cases, fraudulent prime loans were more of an issue than subprime loans. Losing 100% on a pool expected to have 0.1% losses is a bigger shock than losing 30% on a pool expected to have 10% losses." (p. 382)
Thread on this subject:
Thread on this subject:
https://twitter.com/ReformedTrader/status/1361457559735992321
180/ "Immediately after losing a job, people feel bad. But looking back, a lot of job losses turn out to be blessings in disguise. A bad job is lost; a good career is found as a result. Sometimes you have to shake things up to get them to fall into the right places." (p. 384)
182/ Aaron Brown on the Phil Bak podcast
* Poker
* Factor construction in academia (over-simplified metrics and p-hacking) vs. practice (accepting that some factors will work and some won't)
* Bitcoin
* Incentives and why no one wants to pull the goalie
player.fm/series/the-phi…
* Poker
* Factor construction in academia (over-simplified metrics and p-hacking) vs. practice (accepting that some factors will work and some won't)
* Bitcoin
* Incentives and why no one wants to pull the goalie
player.fm/series/the-phi…
• • •
Missing some Tweet in this thread? You can try to
force a refresh
