Quant Awareness and Odd-Lot Trades

Matt Levine is a Bloomberg View columnist. He was an editor of Dealbreaker, an investment banker at Goldman Sachs, a mergers and acquisitions lawyer at Wachtell, Lipton, Rosen & Katz and a clerk for the U.S. Court of Appeals for the Third Circuit.
Read More.
( Updated
a | A

Raising awareness.

Two Sigma is holding a contest:

That’s what $38 billion Two Sigma is doing with its new partner Kaggle, a community of data scientists who collaborate and compete in writing machine-learning algorithms. In a contest starting Thursday at noon, Kaggle’s quants will have three months to create a predictive trading model from four gigabytes of financial data provided by Two Sigma. Cash prizes go to the seven best performers. Two Sigma, which competes with Silicon Valley giants for a limited supply of data scientists, aims to raise their awareness of markets and job opportunities in finance.

Are there data scientists who are not aware that there are jobs in finance for them? I should set up a Kaggle contest where I provide a data set that consists of all the articles about machine learning in finance and see if anyone can spot a pattern. Anyway:

The quants, who might not have access to large amounts of financial data without the help of Two Sigma, won’t know what securities or even the asset class they’re evaluating as they hunt for predictive trading signals.

There are good contest-y reasons for that (you don't want people cheating by going and finding out-of-sample data to fit, etc.), but it also has a certain symbolism. Gut instincts, financial intuition, a sense of how the economy works and how asset classes interact: all worthless! You can't use them! You don't know what you're looking at. It's just some numbers. I have written that Two Sigma's wildly successful competitor, Renaissance Technologies, "doesn't really hire people because they understand investing":

It's a little embarrassing, no? That investing is best understood by people who don't understand investing? That it's a trivial application of broader data-science principles, best addressed by people who were trained on harder and more interesting applications? 

Two Sigma is taking it even further, trying to source its investing algorithms from people who (1) don't understand investing, (2) seem to be unaware that finance jobs exist, and (3) don't know what asset classes they're trading. [Update: A Two Sigma spokesman said: "In fact, the algorithms generated in this competition will not be put into production at Two Sigma, will not become part of Two Sigma’s models and will not generate any revenue for Two Sigma."] Eventually the world's most successful hedge fund will be an astrophysics lab that is trading stocks without even knowing it.

People still read Graham and Dodd! People still invest by trying to understand the workings of the stock market, the economic fundamentals of an industry, the competitive position of a company. Doesn't that seem like it should work? And then you have the scientists who don't even know what asset class they're looking at. The scientists seem to be ascendant these days. What will happen to the understanders? Will the astrophysicists end up running the whole financial world? Will they think of Graham-and-Dodd fundamental investing the way they think of astrology now? It is a pretty story that has captured the imagination of many people for so much of history, but perhaps modern science will just expose it as superstition and wipe it away.

Also, I have quoted this before, but here is a famous story about artificial intelligence:

Sussman talked over his program with the man, Marvin Minsky. At one point in the discussion, Sussman told Minsky that he was using a certain randomizing technique in his program because he didn't want the machine to have any preconceived notions. Minsky said, "Well, it has them, it's just that you don't know what they are." 


In 2012, Pacific Investment Management Company launched an exchange-traded-fund version of its flagship Total Return Fund, run (like the TRF itself) by Bill Gross. The ETF version (ticker: BOND) had a good first few months, outperforming the main fund. It did this in part by the traditional method of buying bonds for less than they were worth. So for instance it would find a non-agency mortgage-backed security worth 82.75 cents on the dollar, according to its outside pricing vendor, and buy it for 65 cents on the dollar. Good deal! For a small fund -- as BOND was at the beginning -- even a little bit of this goes a long way:

For example, on March 9, 2012, PIMCO purchased an NA MBS odd lot at $64.9999 with a current face of $0.2 million. PIMCO then valued the position at the Pricing Vendor’s institutional round lot mark of $82.74585 (a 27% increase). This trade alone increased BOND’s NAV by nearly $0.02 per share in one day.

That's from the Securities and Exchange Commission order against Pimco, because the SEC did not approve of Pimco's particular method of buying low and selling high here; Pimco settled the SEC's charges for almost $20 million. The issue is that those bonds were worth 82.75 only in "institutional round lot" size -- call it $1 million of face amount -- but Pimco had bought an odd lot of $200,000. The odd lot was worth less. You can tell because Pimco bought it for 65 cents on the dollar. You can't really just go around buying things worth 82.75 for 65. That's not how this works. Pimco bought this stuff cheaper than its pricing marks because it was worth less than the pricing marks, because it was smaller and less liquid, and then pretended that it was worth as much as the pricing marks.

Except, nope! Pimco's response to the SEC settlement notes that, "since 2014, BOND has sold all but $400,000 of the 43 smaller-sized non-agency MBS positions -- at prices averaging 99% of the levels where they were valued." Basically its marks were right. The prices it paid were wrong. It is sort of magical, honestly? One way or another, Bill Gross was able to buy at 65, mark at 82.75, and sell at 82.75. He just created value. Perhaps he sometimes did this by assembling round lots -- buy $200,000 at 65, $400,000 more at 65, $400,000 more at 65, and boom, you've got $1 million at 82.75. Perhaps the price gap converged over time. Perhaps he and his traders were very clever, or very persuasive. Perhaps they bought odd lots from desks who were desperate to get rid of them, and sold to people desperate to acquire them (to make up a round lot?). Or perhaps desks were not super price-sensitive in trading tiny odd lots with Pimco. You can lose a few points on a $200,000 trade when you're dealing with the guy who runs -- at the time of the BOND launch -- about $245 billion in his main fund. 

So the SEC's issue is not really that Pimco marked these bonds wrong, since after all Pimco's marks turned out to be right. (I mean, it is not hard to argue that the initial marks were wrong, and that the value was realized later, but the SEC doesn't really build out that argument.) Instead, the issue is that they weren't sustainable: You might be able to buy $200,000 of bonds at a 17-point discount because you're Bill Gross, but even Bill Gross can't do that on $200 million of bonds. And those odd lots made a big difference when BOND was a tiny new ETF, but now that it has $2.3 billion under management, they're not going to move the needle.

And so BOND's early reported performance, though kind of true, was misleading:

“PIMCO misled investors about the true long-term impact of its odd lot strategy and denied them the opportunity to make fully informed investment decisions about the Total Return ETF,” said Andrew J. Ceresney, Director of the SEC’s Division of Enforcement.  “Investment advisers must accurately describe the significant sources of performance and the strategies being used.”

Ehh! You can see why funds would try to get some performance by driving hard bargains, and why they might not want to boast about it in their disclosures.

Anyway the SEC order also contains some random funny stuff about how Pimco operates, or at least how it operated under Bill Gross. Would you not find this a bit condescending?

At the time, PIMCO employed a “gold star” program to incentivize traders to maximize performance. Gold stars were financial rewards given to PIMCO traders at $1,000 per award.

Or there's the time that "BOND’s portfolio manager" -- Gross -- "sent a handwritten note to PIMCO’s trading desks, including NA MBS traders, giving instructions to purchase positions below Pricing Vendor Marks." The SEC cites that as indication that Gross was intentionally gaming the odd-lot strategy, but it is also evidence that Gross communicated weirdly. Who sends handwritten notes? Oh, right, a guy who "doesn't like employees speaking with him or making eye contact." 

Proxy access.

There are two main ways to nominate someone to serve on the board of a public company:

  1. Be the board of directors of that company, and put your nominees in the company's annual proxy statement; or
  2. Run a proxy fight in which you put up your own slate, print your own proxies, and wage an expensive bitter campaign against the incumbent board.

For a long time, people have wanted an in-between approach in which shareholders could just nominate a director, without all the misery of a proxy fight. Like, why not let a big shareholder just nominate someone, and include that nomination in the company's own proxy statement? This is called "proxy access," and it is viewed as a form of good governance that will give big thoughtful non-activist institutional shareholders more of a say in how their companies are run, and it is finally slowly happening.

Barely. Last month GAMCO Asset Management Inc. nominated a director candidate at National Fuel Gas Company, "the first known use of proxy access bylaws to make a nomination." It did not go well:

NFG delivered a letter to GAMCO rejecting its proxy access nominee for inclusion in NFG’s proxy statement on the basis that GAMCO “has not complied, and is not able to comply, with the terms and conditions set forth in the By-Laws to submit a Stockholder Nominee.” [emphasis added] More specifically, NFG reminded GAMCO that a shareholder seeking to use proxy access must, under the terms of NFG’s proxy access bylaw, make certain representations and warranties, including a representation that the shareholder acquired the shares used to satisfy the proxy access eligibility threshold “in the ordinary course of business and not with the intent to change or influence control of the Corporation, and does not presently have such intent.” 

There is some legalistic analysis about what GAMCO's intent was -- it had given some indications of activism -- and GAMCO soon withdrew the nominee. This is a typically wonderful piece of shareholder-democracy logic: You can nominate someone for the board of directors, but only if you don't have any "intent to change or influence control" of the company. If you don't want to influence the company, why would you nominate a director? 

Is Uber bad?

How should you think about Uber? There is a straightforward case that it offers cheap convenient rides by exploiting workers, avoiding employee-protection rules by treating its employees as independent contractors. It gives you a great cheap service, but at the expense of the workers providing it. Here is Izabella Kaminska on Uber's philosophy:

What they want is for the law to bow to “consumer demand” for cheaper taxi services by granting Uber the right to ignore collective regulations on worker rights and conditions. But all this equates to is an economic transfer from the working class over to urban metropolitan elites, which benefits one particular corporation over others.

But that's toward the end of a post that begins by talking about how much money that one particular corporation is losing. Kaminska quotes Hubert Horan at Naked Capitalism, who calculates:

For the year ending September 2015, Uber had GAAP losses of $2 billion on revenue of $1.4 billion, a negative 143% profit margin. Thus Uber’s current operations depend on $2 billion in subsidies, funded out of the $13 billion in cash its investors have provided.

Uber passengers were paying only 41% of the actual cost of their trips; Uber was using these massive subsidies to undercut the fares and provide more capacity than the competitors who had to cover 100% of their costs out of passenger fares.

You can quibble with the numbers, I suppose (here are Part Two and Part Three of Horan's post). And you could be more bullish than Horan and Kaminska are; self-driving cars could do wonders for Uber's margins! But if you take this at face value, then it doesn't quite read like a transfer from the working class to the urban metropolitan elites, does it? The drivers get economics that are roughly comparable to, maybe a bit worse than, what they'd get from regular taxi companies. But 59 percent of the cost comes from investors, specifically, the earnings-insensitive high-net-worth investors in Uber's recent private funding rounds. It's a transfer from the investor class to the urban metropolitan elites. Annie Lowrey has it right: Uber looks more like "progressive taxation from the filthy rich to the somewhat less rich." 

I suppose it can be both. Matt Yglesias famously called Amazon "a charitable organization being run by elements of the investment community for the benefit of consumers," but it is not hard to find articles about how Amazon's low costs and high customer satisfaction come in part at the expense of its workers. Perhaps the trick is that consumers don't feel too bad about exploiting workers if they are exploiting capitalists more. It is not always clear what's in it for the capitalists, though.

Tax reform.

Ugh, yesterday we talked a little about Steven Mnuchin's claim that Donald Trump's tax plan would not represent a net tax cut on the wealthy, and I wrote:

There is an intellectual basis for tax cuts on the wealthy: Tax cuts stimulate growth, the wealthy create jobs, whatever. You can disagree, but there's a theory there. What is the theory for cutting taxes on the wealthy in a way that doesn't actually cut their taxes? Is that also stimulative, because they think their taxes are lower even though they really aren't?

As about a dozen people pointed out, this was dumb. Lowering tax rates and broadening the base by eliminating deductions has a perfectly good intellectual basis, better than just cutting tax rates and hoping the wealth will trickle down. It reduces the inefficiencies of deductions, and lowers marginal rates, improving incentives. I was just thinking about Mnuchin's comments sloppily as an odd twist on supply-side tax-cutting, but on their own merits they make perfect sense. I'm sorry. On the other hand no one seems to believe that this is actually Trump's plan, in part because it's not what he said during the campaign, and in part because you probably can't eliminate enough deductions to pay for the rate cut.

People are worried about unicorns.

The holidays are coming up, but you don't need to get me anything. Just the fact that you read Money Stuff each day is the best present I could ask for. But if you absolutely insist on getting me a present, I have had my eye on this "Unicorn Sparkle Ugly Christmas Sweater" at rue21. That name does not begin to do justice to this amazing article of clothing, which features a unicorn who looks like he has been hitting the sparkle a little too hard. If you work at a billion-dollar tech startup and you don't wear this to your office holiday party, what are you even doing?

Elsewhere in unicorns, Zenefits has disrupted the insurance market by (selling insurance without the necessary licenses and also) giving away free human-resources software to small businesses to induce them to use it as an insurance broker. It turns out that this is illegal, at least in some states. It's called "inducement." There is a fix:

Kara Klotz, a spokeswoman for the Washington insurance commissioner, said that to continue operating in the state Zenefits has agreed to start charging for certain software features on Jan. 1. “The inducement law in Washington is clear,” Mr. Kreidler said in a statement. “Everyone has to play by the same rules.”

Sometimes the libertarian Silicon Valley startup ethos can look arrogant and unattractive when applied to regulated industries. (Zenefits probably should have gotten those insurance licenses before selling insurance!) Other times, though, it mostly makes the regulation look bad. 

Elsewhere: "I would love to get involved, whether it's in a couple little startups here and there, take a little risk, have some fun and see where it goes," says Michael Phelps about tech investing. 

People are worried about stock buybacks.

"Is the American Public Corporation in Trouble?," asks this paper by Kathleen Kahle and René Stulz:

There are fewer public corporations now than forty years ago, but they are much older and larger. They invest differently, as the importance of R&D investments has grown relative to capital expenditures. On average, public firms have record high cash holdings and in most recent years they have more cash than long-term debt. They are less profitable than they used to be and profits are more concentrated, as the top 100 firms now account for most of the net income of American public firms. Accounting statements are less informative about the performance and the value of firms because firms increasingly invest in intangible assets that do not appear on their balance sheets. Firms’ total payouts to shareholders as a percent of net income are at record levels, suggesting that firms either lack opportunities to invest or have poor incentives to invest.

Things happen.

Abu Dhabi Sovereign-Wealth Fund Gets Entangled in Global 1MDB Scandal. Credit Suisse Said to Cut 175 London Jobs Ahead of Investor Day. Should Your Employer Go Full Robo on Your Retirement Savings? Highland Capital seeks to unmask author of 'defamatory' post on Dealbreaker. Russia Says Foreign Spy Agencies Preparing Cyberattacks on Banks. OPEC Deal Hinged on 2 a.m. Phone Call and It Nearly Failed. Giant Aluminum Stockpile Was Shipped From Mexico to Vietnam. Italians Stay Calm as Traders Fret Over Risk of Renzi Ouster. Italy’s referendum casts shadow over banks’ bad loans. Aixtron Tumbles as Obama Said Poised to Block Chinese Takeover. Singapore to ban ex-Goldman Sachs star banker amid 1MDB probe. Igor Oystacher is accused of spoofing in French futures. Contagion in the CDS Market. The Carrier deal shows what Trump understands about the politics of jobs. Will Ivanka Trump's China-Made Clothing Line Survive Her Dad's Trade War? Buzz Aldrin recovering after evacuation from South Pole. The SEC has an app. Beaver walks into Md. store, finds only artificial Christmas trees, and proceeds to trash it. Media and social class

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

(Updates first section with comment from Two Sigma.)

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 mlevine51@bloomberg.net

To contact the editor responsible for this story:
James Greiff at jgreiff@bloomberg.net