(Bloomberg View) -- Slippage.
Basically the way I think about U.S. equity market structure is:
- "Real" investors (hedge funds, mutual funds, etc.) do not want to trade with high-frequency traders, because high-frequency traders are evil.
- Real investors do want to be able to buy and sell unlimited quantities of stock immediately.
- The only people who are consistently willing to trade instantly with real investors are high-frequency traders.
- You see the problem.
- The solution is for the real investors to trade on a platform that says it protects them from high-frequency traders, while also quietly having lots of high-frequency traders.
The trick is that the real investors want to be deceived, but not, you know, deceptively. You can't just lie to them. So when Barclays Plc showed users of its dark pool a bubble chart of the different user types, and deleted the bubble for a big high-frequency trader, it got in trouble. When Bank of America told investors that it was filling their orders out of its inventory to shield them from high-frequency traders, but was actually shopping those orders to the standard lineup of high-frequency traders, it got in trouble. You want to create a vague sense of protection from high-frequency traders without coming out and saying "there are no high-frequency traders here," if there are.
This pleasant self-deception is perhaps a symptom of a more general issue in market structure, which is:
- Real investors want to be able to buy and sell unlimited quantities of stock immediately.
- They do not want to move the price.
The way modern high-speed electronic markets work is that there is a price for a stock, and you can buy 100 shares immediately at that price, but if you want to buy 10,000 shares immediately you will pay way, way, way more than that price. The people selling you that stock are thinly capitalized electronic high-frequency traders who are not interested in getting short 10,000 shares while there is a huge buyer in the market; they'll sell you 100 shares cheaply if they think that the next trade will be the other way, but if you are just buying a ton then they assume the stock is going up, and they preemptively raise their prices to avoid being on the wrong side.
If you are a real investor who frequently buys 10,000 shares at a time, this is a frustrating dynamic. You want to buy stock at the market price. You don't want to pay more just because there is a big buyer (you) for the stock. You tend to think that the high-frequency traders' efforts to notice big buyers and raise their prices are unsporting, even "front-running," and you do stuff to avoid tipping them off, like breaking up your orders and timing them carefully and trading in dark pools. And of course the real dream is to find a place to buy 10,000 shares of stock where there are no high-frequency traders and where the price doesn't move. But then you are left with the problem above: The places where there are no high-frequency traders are not the places where you can buy 10,000 shares of stock quickly.
Anyway here's a story about how Steven Cohen's Point72 Asset Management LP's venture-capital arm invested in Imperative Execution Inc., a new dark-pool company founded by a former Point72 trader who "said he waged a daily battle to prevent slippage from hurting his profits":
Many big investors say HFT erodes their profits. That’s because prices often drop when a large player is selling, or rise when the large player is buying—a phenomenon called “slippage” that’s often blamed on speedy traders.
Slippage is “a multibillion-dollar-a-year problem” for the hedge-fund industry, said Matthew Granade, managing partner of Point72 Ventures.
If you conceptualize "high demand causes prices to go up" as a problem, then you will look for solutions. (If you conceptualize it as "economics," then you will be like "umm right sure.") One popular solution is batch auctions, which Imperative will apparently try, except also with "artificial intelligence":
A key feature is that it only executes trades at discrete points in time, rather than continuously, the way exchanges work. The length of the intervals varies randomly, which Imperative says will keep the speedy traders from figuring out a pattern and gaming the platform’s design.
The idea of “non-continuous” trading venues isn’t new. But Imperative says it has added a new element by using artificial intelligence, or AI. Its systems will monitor for whether trades on IntelligentCross are causing stock prices to move, and adjust the length of its intervals to minimize such slippage, using AI-powered software.
It seems to me that slippage is not, in its essence, a technical problem. The problem is that real investors want to buy a lot of stock instantly. If there were exactly offsetting selling interest from other real investors, at exactly the same time, then the real buyer could trade with the real seller in size without moving the price. That would be a delightful, and somewhat strange, coincidence. (If an index fund is buying a stock, it seems unlikely that another index fund is selling it.)
Otherwise, though, you have genuinely competing desires. The real investors want to trade large size without moving the price, to be able to hide their intentions from high-frequency traders. The high-frequency traders want to be able to respond rapidly to information about supply and demand. Most of what you can do to make your venue appealing to real investors -- making it less continuous, less informative, so that less information about their intentions is available -- makes it less appealing to high-frequency traders. Which is fine! Unless you want your real investors to be able to trade.
Here is a story in the Financial Times about funded collars, and who could resist a story about funded collars? "'Oligarchs use this product; a lot of Russians and Ukrainians,' said one senior banker," so there is that. (Disclosure: I used to work on a structured-equity-derivative desk that did this sort of thing, though I don't think I ever actually did a real funded collar, and certainly not for an oligarch.) But a funded collar is actually a fairly simple corporate-equity-derivative product, as these things go, that nonetheless manages to throw off a lot of magic. A funded collar is just:
- A client owns a lot of stock in a company, worth say $100 per share.
- The bank sells that client a put option on the stock: If the stock falls below, say, $80, then the client can give the stock to the bank and the bank will pay $80 for it.
- The client sells the bank a call option on the stock: If the stock goes above, say, $120, then the client must give the stock to the bank and the bank will pay $120 for it.
- The option prices offset, so no (or relatively little) money changes hands.
- Now instead of just owning stock, the client owns "collared" stock: If the stock falls below the put strike ($80 or whatever), she can't lose any more money; if the stock rises above the call strike ($120 or whatever), she doesn't make any more money.
- The bank lends the client money against the position.
Step 6 is the first piece of magic. If you own stock, and you come to a bank for a loan, it will give you a margin loan. But your stock is volatile, so lending against it is risky, so the bank won't lend you the full value of the stock:
In margin loans, banks are typically willing to lend up to half the value of the borrower’s shares. The borrower then has to pay “margin calls” in cash if the share price drops significantly, with banks seizing the stock if the price drops too heavily for the borrower to keep up with the payments.
But if you own collared stock, and you go to the bank that wrote you the collar for a loan, it will happily lend you more money. (Generally it will lend you the full amount of the put strike, e.g. $80, perhaps minus some time value.) And the reason for this is that your collared position is absolutely guaranteed to be worth at least $80. It is risk-free, so the bank can lend against it. The magic is that the bank has created the risk-free exposure: By writing the put option itself, the bank has guaranteed the price, and so the bank is happy to lend against that guaranteed price. It has transformed lending risk into equity option risk, which tends to be an easier risk to understand and manage. But on the lending side, the risk has simply disappeared.
I submit to you that that is a nifty trick. But here's another one:
While the structure has long existed as a way for investors to raise money from their existing holdings, it has more recently become a popular way of amassing a fresh stake in a company. Investment banks typically hedge their collar position by shorting the stock — borrowing it and selling it on — so they can sell the shares directly to the client to help them build their stake.
When I say that equity option risk is easy to understand and manage, I mean that there is a well-understood and widely accepted formulaic method for pricing and hedging equity options. Basically, if you are short a put (as the bank is here), or long a call (as the bank also is here), your hedge is to sell short a specified percentage of the underlying shares. So if the client has bought a collar on a million shares of stock, the bank might have to hedge that by selling short 700,000 shares of stock. (It will then adjust the hedge by selling more stock, or buying some back, as the stock price moves.) The bank can just sell those shares into the market, though that creates slippage and might also have some legal issues depending on who exactly is getting the funded collar.
But there is an alternative: The bank can sell the shares to the client. The client has 1 million shares worth $100 million. She collars those shares and borrows $80 million against them from the bank. As a result of the collar, the bank happens to have 700,000 shares to sell. Those shares sort of came from nowhere: I mean, it borrows them from some stock lender, but in a sense they were created to offset the collar position that the bank entered into. (In a very deep sense the shares were sold by the client, synthetically, in the form of the collar.) And then it sells them to the client for $70 million. The client has used her existing stock position to create (1) a new stock position and also (2) the money to pay for it. It's pretty neat!
We have talked a few times about the Hovnanian Enterprises Inc. trade, where Blackstone Group LP's credit unit, GSO, offered attractive funding to the company in exchange for its agreement to do a quickie default on some of its debt to trigger credit-default swaps that GSO owns, and to issue weird new bonds to create a large payout on those CDS. We have talked, among other things, about the aesthetic appeal of this transaction, and about the questionable legal arguments against it, but also about how it might be kind of bad for the future of the CDS market if CDS buyers can engineer nonsense like that. But the implications are even bigger: This trade might make for awkward lunches for Lloyd Blankfein.
Lloyd Blankfein and Jon Gray, two of the most powerful figures on Wall Street, sat down for lunch last month, just as tensions between their firms were ratcheting up.
It was a friendly chat between Blankfein, head of Goldman Sachs Group Inc., and Gray, heir apparent of Blackstone Group LP, that veered toward a thorny matter. At issue: a controversial trade involving credit-default swaps that has riveted players in the vast derivatives market -- and set Goldman and Blackstone on a collision course.
Gray questioned Goldman Sachs’s opposing role in the deal, which had spilled into public view, according to people with knowledge of the matter. Blankfein stood his ground, expressing disagreement with the way it was structured.
Disclosure, I used to work at Goldman, and I certainly want Blankfein to enjoy his lunch, but I think I am team Blackstone here. You can't "disagree" with how this trade was structured. The first-order question is, did it work? Did it follow all the rules in a way that actually created the payout that GSO desired? If it worked -- and my sense is it did, though it is not final yet -- then the next question is, are you the sort of financial institution that is down with doing creepy things that technically follow the rules even though they kind of surprise everyone, or are you not? I do not think that, at this late date, you can show up to lunch and be all "Goldman Sachs would never take advantage of a clever reading of the documents!" Sure sure they have the friendly consumer-loan business now, and Clarity Money and Honest Dollar, but the Goldman Sachs I know was also pretty good at reading documents.
A.G. Schneiderman Launches Inquiry Into Cryptocurrency “Exchanges”
Virtual Markets Integrity Initiative Seeks to Improve Transparency and Accountability of Major Cryptocurrency Trading Platforms to Protect Virtual Currency Investors
AG’s Office Sends Letters to 13 Virtual Currency Trading Platforms or “Exchanges” Requesting Disclosures on Their Operations, Use of Bots, Conflicts of Interest, Outages, and Other Key Issues
I have to say the scare quotes on "exchanges" seem unnecessarily harsh? Like I am willing to concede that a Bitcoin exchange is a Bitcoin exchange. It is not a nationally registered securities exchange, sure. Perhaps -- if for instance it trades initial-coin-offering tokens that count as securities under U.S. law -- it ought to be, and the scare quotes are a low-key intimation of illegality. And I realize that electronic trading platforms for many financial products -- currencies, Treasury bonds -- tend not to use the name "exchange," in order to avoid confusion with the particular set of characteristics (national registration, public quotes, membership, etc.) that official securities exchanges have. But colloquially I have no problem with saying that a place where you can go to trade things is an "exchange," and the fact that cryptocurrency exchanges are new, that they trade something that was not previously traded on exchanges (or at all), is not necessarily a knock on them.
On the other hand there are lots of other concerns! The thing about established securities exchanges is that there are rules around transparency and market manipulation and asset custody and so forth. The thing about cryptocurrency exchanges is ... well, they might have rules like that, but there certainly aren't uniform rules that are set and enforced by government agencies. So here is Schneiderman's questionnaire for the crypto exchanges, which has questions like:
Describe any practices, procedures, safeguards, and monitoring your platform has in place to detect, prevent, block, or penalize suspicious trading activity or market manipulation, the types of customer behavior or trading activity that qualifies for such response, and when such measures came into effect.
Describe the precautions you have in place to safeguard the Fiat Currency, Virtual Currency, or financial instruments in your custody.
They're fair questions! Also I mean they are basic stuff. You'd hope that customers of the exchanges would have asked similar questions, and that the exchanges would have robust answers ready and perhaps already published on their websites. But I wonder sometimes if Bitcoin's emphasis on trustlessness and decentralization makes its users too willing to trust centralized intermediaries. "It's trustless," I suspect the thought process goes, "so why should I bother asking how they keep my money safe?"
Elsewhere in crypto.
The embattled political data firm Cambridge Analytica quietly sought to develop its own virtual currency in recent months through a so-called initial coin offering, a novel fund-raising method that has come under growing scrutiny by financial regulators around the world. ...
The goal of Cambridge Analytica’s own coin offering? Raise money that would pay for the creation of a system to help people store and sell their online personal data to advertisers, Brittany Kaiser, a former Cambridge Analytica employee, said in an interview. The idea was to protect information from more or less what the firm did when it obtained the personal data of up to 87 million Facebook users.
Since the scandal of Cambridge Analytica's use of Facebook data broke, there has been some debate over whether the company is a shadowy organization that has developed powerful political-manipulation techniques that delivered the U.S. presidency to Donald Trump, or a dumb useless scam that wasted gullible donors' money. This is a data point!
Elsewhere: "Icelandic bitcoin heist suspect escapes jail and flees the country — on the same plane as the Prime Minister."
Morgan Stanley just left Morgan Stanley. No, really.
The fixed-income trader, who just happens to share a name with the 82-year-old Wall Street firm, has left the bank, a person with knowledge of the matter said. Stanley joined in 2012 after graduating from Boston College, according to records from the Financial Industry Regulatory Authority.
Stanley, whose middle name is Adam, didn’t return LinkedIn messages seeking comment on his next career move. A spokesman for the New York-based bank declined to comment.
My first reaction was: I can't believe he wasn't at Goldman Sachs. Honestly I hope there was a bidding war for his talents/name among the big banks. But answering the phone "Goldman Sachs, this is Morgan Stanley" is just too good. Obviously "Morgan Stanley, this is Morgan Stanley" is pretty good too. "Yes I'd like to talk to the trader who just sold me these bonds." "Yes that's me." "Who are you?" "Morgan Stanley." "No I know but what is your name?" "Morgan Stanley." You could put off a lot of customer complaints that way. He's not an investment banker; he's an investment bank. I'll be disappointed if he left for a tech startup like everyone else. What if he left to found his own boutique advisory firm? What if he's calling it Morgan Stanley & Co.?
Elsewhere in names, here is a breakfast-themed story about Greg Coffey and Louis Bacon that may not be to all tastes but that I grudgingly admired.
At this point it's kind of a boring cliché to make fun of Google for its erstwhile motto "Don't Be Evil," but still it is jarring to see former Google CEO and current Alphabet Inc. board member Eric Schmidt argue that we need to develop intelligent robots that can kill humans?
At a Tuesday hearing of the House Armed Services Committee, Schmidt gave a strong endorsement of the Pentagon’s proposal for a new AI center and its collaboration with technology companies.
"The nature of AI is a long-term technology that will be useful for defensive and perhaps offensive purposes as well," said Schmidt, who also chairs the Defense Department’s Innovation Board.
Yep definitely important to teach the computers to wage offensive war, what could possibly etc.
Morgan Stanley First-Quarter Profit Rises to Record High. (Earnings release.) The Biggest U.S. Banks Made $2.5 Billion From Tax Law—in One Quarter. Bill Ackman Calls Newell’s Icahn Agreement ‘Deal With the Devil.’ Russia Sanctions Throw Global Aluminum Industry Into Chaos. United Trims Some of Its Capacity Growth Plans for This Year. I.R.S. Website Crashes on Tax Day as Millions Tried to File Returns. Jeremy Grantham saw his shadow so it's seven years of bear markets. "Swipe right for inequality.' ‘Nut Rage’ Redux: Another Korean Air Executive Is in Hot Water. Flood of tumbleweeds submerge California town.
If you'd like to get Money Stuff in handy email form, right in your inbox, please subscribe at this link. Thanks!
This column does not necessarily reflect the opinion of the editorial board or Bloomberg LP and its owners.
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.
For more columns from Bloomberg View, visit http://www.bloomberg.com/view.
©2018 Bloomberg L.P.