Money Stuff

Ethics, Quants and Cold-Calling

Also Fannie Mae and Freddie Mac, worries about worries, and cod.


I used to be a lawyer, and lawyers have a code of ethics. Now I am a journalist, and journalists have a code of ethics. One thing that strikes me about these codes is that they are opposites. Oversimplifying massively, the basic rule for a lawyer is that your obligations are to your client, and you have to act in her best interests, even if that is against the interests of accuracy; legal ethics is then mostly a set of exceptions to this principle. Oversimplifying massively, the basic rule for a journalist is that your obligations are to the public, and you should be accurate even if that is against the interests of the people you talk to; journalistic ethics is then mostly a set of exceptions to this principle. In both cases the exceptions are huge and important: You're not supposed to lie to the public as a lawyer, or mislead your sources as a journalist, etc; none of this is meant to be any sort of ethical advice. But if someone says to you "oh yeah I murdered someone," as a lawyer, your baseline expected response would be not to tell anyone; as a journalist, your baseline expected response would be to tell everyone.

Obviously these opposite rules make sense in their respective contexts; the role of a lawyer is different from that of a journalist, and each profession's ethics are well adapted to doing their jobs usefully. Still it is weird to think of them as "ethics." They are both functional systems adapted to the work of their professions, not absolute moral-ethical rules handed down by a higher power. Keeping a murderer's secret is not absolutely ethical for humans, and disclosing that secret is not absolutely ethical for humans; each is ethical or unethical depending on its social context.

Recently there were some scandals in the foreign exchange market, which led to a group of regulators and market participants getting together in a Foreign Exchange Working Group that today released a code of conduct for the FX market. Oversimplifying massively again, you could point to three controversial points in recent FX practice:

  1. Customers would give banks orders to, say, buy zlotys at a fixing price at a particular time, and the banks would buy zlotys ahead of that fix, which tends to push up the fixing price, resulting in a worse price for the customers.
  2. The banks would also share those customer orders with other banks, who might also trade in ways that disadvantaged the customers.
  3. The banks would use "last look" systems where they'd put out a quote to buy or sell zlotys, and a customer would say "okay we will buy from you at your offer," and the banks would wait a bit to respond and then fill the trade if the price had moved in their favor but say "oh never mind" if the price had moved against them.

Are those practices ethical? Unethical? It is tempting to think that there are absolute answers, that it is absolutely unethical for a human to trade ahead of a customer order, or back away from a bid or offer, or whatever. But of course that's not how it works. Banks got in a lot of trouble for, basically, thing 2; sharing information with their competitors turned out to be pretty unethical. But they also got a lot of grief for thing 1: Trading ahead of customer orders sort of looks bad. But it turns out to be fine! (Dan Davies wrote a good parable explaining it.) After paying all the fines, the banks sent out letters to their customers saying in effect "by the way we are going to continue trading ahead of the fix." Last look is similarly ambiguous: Barclays PLC paid a large fine for its last-look practices, but regulators stopped well short of saying that last look is illegal or unethical, and banks continue to use it.

The new FX Global Code, similarly, is fine with trading ahead of client orders at a fix ("pre-hedging"). One stylized example that the code approves of:

A bank receives a large order from a fund (Client) to sell EUR/PLN at the London 4 p.m. fix. According to their pre-agreed terms and conditions, the bank and the Client have agreed that the bank will act as Principal and may hedge fixing transactions depending on market conditions. The bank hedges some of the order amount before the fixing window since it judges that the five-minute fixing window is too short to clear such a large amount without affecting the market rate to the Client’s disadvantage.

Of course if the client's order is to buy zlotys, the bank's pre-hedging (before or during the fix) will push up the price of zlotys. The bank should do it carefully so as not to push up the price too much -- and it shouldn't tell all its bank buddies that it has a big buyer of zlotys so they can trade too -- but the ethical principle here is "you should buy ahead of a customer trade responsibly" rather than "you shouldn't buy ahead of a customer trade."

Or last look. The idea of last look is that if the price moves against you, you can back away from your market; if it is stable or moves in your favor, you execute the trade. When you just put it nakedly like that it sounds bad, so the FX Global Code puts it more blandly:

If utilised, last look should be a risk control mechanism used in order to verify validity and/or price. The validity check should be intended to confirm that the transaction details contained in the request to trade are appropriate from an operational perspective and there is sufficient available credit to enter into the transaction contemplated by the trade request. The price check should be intended to confirm whether the price at which the trade request was made remains consistent with the current price that would be available to the Client.

You could imagine a different rule, in which markets are firm and you have to execute at the price you bid or offer. (That's how many stock exchanges work.) Or you could have the opposite rule, in which your prices are indicative and you can always consider any relevant information in deciding whether to back away. (Plenty of voice-traded markets have norms more like that.) None of these rules are absolute rules of ethics; which conduct is ethical and which is unethical depends on market practices and expectations and disclosures. If everyone expects you to execute your quotes, you should execute your quotes. If their expectations are softer, your quotes can be softer.

This is a perpetual problem in finance, particularly when people outside of the financial industry -- or outside of one little corner of the industry -- look in on financial practices and are appalled. "It's just wrong to do ____," they say. "Right is right and wrong is wrong, even in financial markets. 'Everyone does it' is no excuse." But that seems to have ethics exactly wrong. Ethics, in financial markets, are not inherent facts of the world that are discoverable by introspection. They are social practices, descriptions of how a community functions and how people in that community expect to interact with each other. What everyone does -- what everyone agrees everyone should do -- is what is ethical. It's helpful to write it down, though, and make sure everyone agrees.

Quants quants quants quants.

There are two massive areas of job opportunity for data scientists: They can build models that help hedge funds trade stocks and bonds, or they can build models that help internet companies sell advertisements on web pages. Oh or they can build models that help cure cancer or whatever, but compared to financial trading and internet advertising that is a small and unprofitable niche. One of the most incredible feats of marketing of our century is that the internet companies have convinced a lot of people that selling advertisements on web pages is basically the same as curing cancer, while buying stocks and bonds is evil:

“At tech companies, the permeating value is that they’re about trying to make the world a better place, whereas at hedge funds it’s about making more money,” Mr. Epstein said.

That's from a Wall Street Journal article -- in its series on quants -- about the talent battle between Wall Street and Silicon Valley. As far as I can tell, the pitch for data scientists from Silicon Valley is: "Come work here, you can build advertising models and pretend that you're saving the world," while the pitch for data scientists from Wall Street is: "Come work here, you can build trading models and not have to pretend that you're saving the world." I actually think that is a useful sorting metric, and I know which one I would take.

I know which one this guy would take, too!

“We had an option to leave, but we couldn’t leave the position at the time," the 39-year-old former Goldman Sachs Group Inc. quant said, recalling the March 2011 quake and hours of aftershocks that he and some trader colleagues braved from the bank’s offices on the 48th floor. My “life is important, but protecting the P&L is more important."

That's Makoto Yamada, a former Goldman quant who left for SMBC Nikko Securities Inc. I feel like Goldman should add that to its list of 14 business principles. "Your life is important, but your P&L is more important" has a nice ring to it, and will definitely weed out lukewarm job applicants.

Elsewhere in the Journal's quant series, here's a story about Magnetar Capital LLC, a traditional hedge fund that "is going quant." Sort of:

Magnetar’s quantitative push is different from many of its high-speed quant competitors, who parse reams of data in the hunt for an edge—even one they can’t explain. Magnetar is taking a more manual approach. Discoveries such as the types of corporate acquirers are more likely to see through to the end of a tough deal, are translated into computer algorithms that trade automatically.

“We start with intuitions and then go see if the data backs it up,” says Mr. Litowitz. “Most people reverse it. They go looking for data and then go find a signal that explains it.”

A basic idea in quantitative investing is that computers are better at making investment decisions -- at spotting patterns, at risk management, at coldly rational decision-making -- than humans are. Separately, humans who make investment decisions like to use computers as tools. That has been true as long as there have been computers: You wonder how much company XYZ is worth, so you build a spreadsheet model to value it. (This article starts with an anecdote about Magnetar founder Alec Litowitz asking a team of analysts to figure out what percentage of mergers fall apart, which is the sort of question analysts have been answering for years using computer databases.) But as the computers learn to do more things, there will be more blurring of the lines. Even where humans keep making the investment decisions themselves, more of the subsidiary decisions -- the analysis that goes into the ultimate investment decision, or the trading decisions that carry out the investment thesis -- will be made by computers.

Cold calling.

Here's a sentence that I wouldn't be surprised to find engraved over the gates of hell:

Former employees say they spent most of their time cold-calling and sending messages on LinkedIn.

It's from this Bloomberg article about the DeVere Group, a financial advisory firm that is being investigated by the Securities and Exchange Commission and that seems to have taken an old-school approach to financial advising. "Three of the former employees, all of whom asked for anonymity out of fear of retaliation, said some salesmen had cocaine and other drugs delivered to fuel their high-pressure cold-calling." "Three of the former employees said they would drink booze out of paper cups during the day." "At the 2015 event at the Grosvenor Hotel, Green, DeVere’s founder, descended to the stage on a zip line amid fireworks." That sort of thing.

We live in a nostalgic age, and I wonder sometimes if there is purely nostalgia-driven demand for old-school brokerage stereotypes. Are there investors who have rejected efficient low-cost robo-advisers but decided that really they just want their money managed by a coke-fueled cold-caller in suspenders, just as they have rejected mass-produced macrobrews in favor of cocktails produced by artisan bartenders in, also, suspenders? Are there aspiring traders with data-science Ph.D.s who have turned down job offers at Two Sigma for scruffier cold-calling jobs, because drinking booze out of paper cups during the workday just feels more authentic? Is there a marketing opportunity for unabashed boiler rooms? "Wolf Securities: How Wall Street Used to Be." I bet that would sell.

A Fannie/Freddie endgame?

I'll believe it when I see it, but Aaron Back thinks that there is "a promising consensus that is beginning to form about the ultimate future of Fannie Mae and Freddie Mac":

These plans would dramatically restructure Fannie and Freddie, and put in place a federal guarantee on certain mortgage securities. This protection would kick in after private investors have taken initial losses through the new “credit risk transfer” market.

Under most of these plans Fannie and Freddie would still exist, but with a more limited role, perhaps as utility-like mortgage guarantors.

One thing that bugs me about Fannie and Freddie is that this question -- the question of how to structure the mortgage market and the government's role in it -- is completely analytically distinct from the question of what to do about Fannie and Freddie's existing shareholders. You could have a totally nationalized mortgage market, or a totally private mortgage market, or some hybrid system in which the government provides a backstop for a fee and private capital explicitly takes the first losses. And you could give the Fannie/Freddie shareholders a massive pile of cash (or valuable shares in whatever the new entities are), or you could give them nothing, or you could give them a pile of cash of some intermediate size.

The questions about how to structure the mortgage market are policy questions: What are our national purposes in thinking about housing and mortgages, and what are the best ways to serve those purposes? The questions about the investors are mostly legal and political: Were those shareholders treated fairly or unfairly, and will their support be helpful in whatever the new mortgage-market structure is? But people always run these questions together, asking questions like "will a restructured mortgage market with a government backstop be good or bad for Fannie and Freddie shareholders?" They have nothing to do with each other! You could have a government backstop and zero the shareholders, or richly reward them, or do anything in between. 

People are worried that people aren't worried enough.

Here's "Explaining the puzzle of high policy uncertainty and low market volatility," from Lubos Pastor and Pietro Veronesi. 

We argue that the contradictory nature of the political signals in 2017 has reduced their informativeness. Political news is noisier than it used to be. Investors are becoming skeptical that politicians’ pronouncements have much to do with their future policy actions. Markets continue listening to politicians, but they pay less attention than they used to. The result has been a weaker link between uncertainty and volatility in early 2017.

I suggested a related theory last week:

It is hard to resist the idea that there is a news-fatigue element to what is going on. Stocks, traditionally, have overreacted to news. A thing happened, and traders thought they were smart and over-interpreted it, and the next day something else happened and people over-interpreted it the other way. But there is just too much news now. If you over-interpreted all of it, you'd get whiplash. So a retreat into under-interpreting -- from panicking at minor news to shrugging off huge news -- seems like a natural reaction.

Pastor and Veronesi put the whiplash in different terms: Markets shrug off huge news because, if it's likely to be contradicted by opposite huge news the next day, maybe it's not actually news?

People are worried about unicorns.

Nah, it's all great in the Enchanted Forest, this quarter:

Investors injected $14.5 billion into U.S. venture-backed startups in the first quarter, up 37% from the previous period, according to data from Dow Jones VentureSource.

Elsewhere, here's a 40-minute video of a man taking apart a Juicero machine, if that's of interest to you. "I ain't much of one for having a robot masticate my produce for me," he says, before tearing it apart.

Torsken kommer nå!

This month the Norwegian central bank introduced a new 200-kroner note with a picture of a cod on it using this music video, which I insist that you watch. Robin Wigglesworth has an unofficial translation, though officially "there are a number of rhymes and references to Norwegian popular culture that are impossible to convey to an international audience." That ... seems ... true.

Me yesterday.

I did an "Ask Me Anything" on Reddit yesterday afternoon. It was fun! You can read it here. We talked about Nicholson Baker and Aby Warburg and bond market liquidity. One excellent question that I failed to answer, from "tsquaredolivegarden," was: "To improve bond market liquidity, have you considered just squeezing the bonds with your bare hands?" 

Things happen.

Democrats Ask Deutsche Bank to Produce Documents on Trump Family Loans. Dan Loeb to Call for Changes to Dow-DuPont Post-Merger Plan. Winners and Losers in a Post-Fiduciary World. Robo-Advisers Battle Wall Street for Rich Investors. Blackstone’s $100 Billion Bridge-and-Tunnel Man. Judges Examine Legitimacy of SEC’s In-House Courts. Dual-Class Stock and Private Ordering: A System That Works. Hedge Funds Squeezed by World's Highest Rents Are Moving Out. Teenagers Everywhere Don’t Understand Money. Bear rescue. The Exquisitely English (and Amazingly Lucrative) World of London Clerks

If you'd like to get Money Stuff in handy email form, right in your inbox, please subscribe at this link. Thanks! 

(Corrects name of Norway's currency in seventh item.)

This column does not necessarily reflect the opinion of the editorial board or Bloomberg LP and its owners.

    To contact the author of this story:
    Matt Levine at

    To contact the editor responsible for this story:
    James Greiff at

    Before it's here, it's on the Bloomberg Terminal.