Money Stuff

Dark Pools and Money Camps

Also robot psychologists, parody ICOs, and the soothing return of worry.

Market structure. 

For a long time, a big worry about dark pools, in the U.S., was that they were marketed as places for institutional investors to quietly trade large chunks of stock with each other away from the prying eyes of high-frequency traders, but they were actually full of high-frequency traders. That just turned out to be a feature of how stock trading works: Big institutions tend not to want to sell big chunks of stock at exactly the same millisecond that other big institutions want to buy similar-sized chunks of the same stocks, so high-frequency traders -- proprietary trading firms, or the trading desks of the banks that ran the dark pools -- serve a useful role in buying from the sellers and selling to the buyers. The purity of big institutions just buying and selling to each other, off in a quiet corner away from the public exchanges, had an obvious appeal -- no one likes a middleman, and everyone worries about "front-running" -- but it rarely seemed to work out that way.

In Europe these days, though, dark pools have a different problem:

The rules known as MiFID II, just six months away, loom large over European operators of dark pools -- they introduce a cap on transactions that would by one estimate bar three quarters of big European stocks from the trading venues. Investors love dark pools because they allow them to trade big orders without tipping off the market to their intentions. That demand isn’t going anywhere, so firms have been rushing to roll out services that allow investors to sidestep the rules.

American regulators worry that dark pools are false advertising, claiming to be places for institutions to quietly trade orders without affecting the price while really being full of high-frequency traders. European regulators worry that dark pools really are places for institutions to quietly trade orders without affecting the price, and "that the market’s shift to the dark could make prices on public exchanges less accurate, weakening the efficiency of public markets." (This is perhaps symbolic of a more general difference -- European regulators regulate economics, while American ones regulate disclosure -- though I wouldn't press too hard on that.) So the European solution is to cap dark-pool trading.

But MiFID II will allow some alternatives. One way is to run periodic auctions, allowing investors to trade with pre-trade price transparency but without leaving standing orders on exchanges for very long -- an approach to curing the evils of high-frequency trading that academics (and the Bloomberg View editorial board) have long advocated. Another, though, is what are called "systematic internalizers":

If SIs become the venues of choice for dark trading post-MiFID rules, that would mark a victory for the large banks whose internal dark pools (dubbed broker crossing networks) account for about half of dark trading. Banks will probably be the biggest SIs, but high-frequency trading firms can also set up systematic internalizers from January when MiFID kicks in.

SIs have their limits: the European Commission has banned them from matching orders between a bank’s customers, meaning either the buyer or seller in the trade must be the operator of the SI.

A systematic internalizer, that is, is sort of like a dark pool, except that institutions using the SI will only trade with the bank or proprietary trading firm running the dark pool. And that's what regulators are encouraging. Again, it's sort of the opposite of the approach in the U.S.

Elsewhere in market structure: New open-outcry trading floor!

Box Options Exchange LLC hopes to open a new floor for about 40 human traders at the Chicago Board of Trade Building. The exchange, whose electronic platform has one of the smallest market shares in U.S. options, is trying to build up its business by vying for orders executed via open outcry.

Not everyone is a fan:

Box’s initial plan, though, sparked critical letters this year from Chicago Board Options Exchange owner, CBOE Holdings Inc., and the NYSE, as well as trading firms that said another trading venue will make it tougher to do business as activity potentially becomes less transparent and more fragmented.

Nasdaq Inc., which also operates options exchanges, has filed three comment letters about its concerns. A main one is that a Box trading floor, if approved, could sit empty for a few months, potentially leading to worse prices for customers if not enough market makers are competing.

A key issue in market fragmentation is that if you are serious about being a broker or market maker, and a new exchange opens, you feel compelled to connect to that exchange. If you're a broker, you have to seek out the best execution for your clients wherever you can find it; if you're a market maker, competitive pressures push you to make markets everywhere. Usually, in the equity markets, this means that if a new exchange opens, you have to pay some money to put a computer next to its servers to run your trading algorithms. But if someone opens a new open-outcry exchange, do you have to hire traders and dress them in funny jackets and send them off to the exchange? It suggests a business model: Build a new open-outcry exchange in a resort location, have it open for an hour before lunch every Monday through Thursday, and force a bunch of market makers to staff it up with very well-rested traders.

Money camp.

Now is roughly the time when new analysts are starting at investment banks, and the Wall Street Journal has a report from inside one bank's training program. Here's one new analyst with an appropriate nickname and attitude:

“I like money,” said Kashious, who goes by Kash. “And you could get really rich.”

Not everyone has the same focus, though, and the trainees will occasionally devolve into frat-boy antics:

During an hourlong trade lesson, a group in the back corner took the opportunity to practice their best fake-flatulence noises. 

And it can be hard to get them to pay attention:

During lectures, some children squirmed in their seats or attempted to wander around, promptly attended to by counselors who promised to hand out colorful stickers if they were quiet. 

Ha wait no this is not an analyst training program at all! It's the Junior Money Matters camp for 7- to 11-year-olds. "When you train puppies, it’s better to train them when they’re younger," explains a 9-year-old. "It’s the same with people." It does read a little like "Liar's Poker," though. Do you think I could start one of those "adult summer camp" things, only for money camp? Money Stuff Money Camp? 

My favorite part might be this, though:

“How do countries make money?” a camp counselor asked during one day’s international trade lesson.

“By suing people!” one boy shouted.

That is not really how international trade works, but it might be the most American thing I have ever heard. The way to create wealth is by suing people! The kids are going to be all right.


"Career of the Future: Robot Psychologist," it says here, and that sounds like a good job except I feel like it would be automated away pretty quickly? Wouldn't the robot feel more comfortable with a robot psychologist who was also a robot? Wouldn't the robot robot psychologist understand the robot better than a human robot psychologist would?

That is of course the problem:

Unlike with humans, we can’t just ask a robot why it does what it does. Artificial intelligences can excel at narrow tasks, but even those that talk have introspective powers about on par with a cockroach.

It is a difficult enough problem to crack that the Defense Advanced Research Projects Agency, better known as Darpa, is funding researchers working on “explainable artificial intelligence.”

So people are resorting to the tools of cognitive psychology to try to figure out why neural networks do what they do. But you can sort of schematically see the issue in investing. If you have a neural network that tells you what stocks to buy, and its insights into those stocks are easily explainable and understandable to human intelligence, then your confidence in those stock picks might be high, but in some sense you didn't really need the neural network. A spreadsheet would have been fine. The neural network was really just doing some kind of calculation or administrative task to confirm or support essentially human insights.

The really interesting use of artificial intelligence is if it can come up with insights that humans would not have -- insights that are inscrutable and inexplicable to the humans. If your really advanced investing robot tells you what stocks to buy, you (and it) won't be able to explain why those stocks are good stocks. This will of course leave you uneasy: How confident can you be that the robot is right, if you can't understand what it's thinking? Mostly what you can do is buy the stocks and see if they go up. (If they don't, you fire the robot.) But I suppose you can also hire a robot psychologist to try to figure out if your robot is thinking about stocks in a good way even if it can't explain its thoughts to you, to try to meta-cognitively understand the robot whose actual insights you can't understand.

Elsewhere: "Robo Advisers Are Gaining Traction as Insurance Salesmen," and I look forward to noir novels in which robot insurance salesmen are seduced into committing murder for the insurance proceeds. And elsewhere in automation: "Are goats taking jobs from union workers?"

Blockchain blockchain blockchain.

You know, on Friday I said that I knew of at least two parody ICOs -- real initial coin offerings on the Ethereum blockchain that are also parodies of initial coin offerings on the Ethereum blockchain -- but of course there are more. A decentralized permissionless financial architecture -- a decentralized permissionless anything -- lends itself to amateur comedy routines. Here is "Useless Ethereum Token":

“I saw that people were investing in ICO after ICO, with each having its own slew of problems (both technical and in principle) and still making absurd amounts of money,” UET CEO wrote the Observer in an email. “My first two thoughts were ‘This is ridiculous’ and ‘How can I get in on this?’ I didn’t have a product… but I realized that people didn’t really care about the product. They cared about spending a little bit of money, watching a chart and then withdrawing a little bit more money. So why not have an ICO without a product, and do so completely transparently just to see what happened?”

We have talked a few times in the past about the difficulty of shorting unicorns: Investors can buy shares in the big venture-backed private tech companies, but they can't sell those shares short, which arguably leads to those shares being overvalued as enthusiasts join in but skeptics are excluded. As I once said, though, "the way to profit from a bubble is by selling into it, and that people sometimes focus too narrowly on short-selling into it": If you think that unicorns as a category are overvalued, the way to profit from that is not so much by shorting Uber as it is by founding your own dumb startup, raising a lot of money from overenthusiastic venture capitalists, paying yourself a big salary, and walking away whistling when the bubble collapses.

Same here! If you are skeptical of the ICO trend, the right thing to do is not to short all the new tokens that are coming to market. It's to build your own token, do an initial coin offering, and walk off with the proceeds. For the sake of your own conscience, you can just go ahead and say that that's what you're doing, right in the ICO white paper. No one seems to mind.

Of course this is not legal advice, or investing advice, or advice about whatever ICOs are (not investments!). But as I type it I realize it's a million-dollar idea, bigger even than Money Stuff Money Camp. The first thing you could do is offer the Token Short Token. White paper abstract:

Cryptocurrency and Ethereum tokens and initial coin offerings are good, but it is too difficult to sell them short. The Token Short Token offers the first convenient opportunity to short the ICO market. It offers that opportunity to me. I will sell you the tokens, and then if the market collapses I will have profited by that collapse. If the market does not collapse, I will also have profited. I cannot exactly tell you what your tokens will be worth in either case, but don't you want to find out?

But the real opportunity is the Token Short Meta-Token. White paper abstract:

Cryptocurrency and Ethereum tokens and initial coin offerings are good, but it is too difficult to sell them short. The Token Short Meta-Token offers speculators an opportunity to short the ICO market: TSMTs can be used to create Token Short Tokens, which can then be sold to investors in subsequent initial coin offerings, allowing the TSTs' creators to profit from the collapse of the ICO market. Of course if you buy TSMTs from me I will also profit from the market's collapse, but maybe you can find a greater fool to buy your TSTs in the meantime.

Obviously there are some details to flesh out, but really someone should do this and give me a portion of the proceeds as a royalty. (Not the tokens; I don't want the tokens.) 

People are worried that people aren't worried enough.

Good news everybody! People are starting to worry again, which means that you don't need to worry so much about how they're not worried enough:

The evaporation of market volatility has baffled many analysts and investors, given the noisy political environment and vulnerable economic recovery, and frustrated traders, who depend on turbulence and healthy turnover to generate profits.

But a sell-off in technology stocks and a spate of hawkish central bank speeches have caused volatility to make a modest but noticeable return in recent weeks.

People are worried about unicorns.

Here is Elaine Ou on Silicon Valley's sexual harassment problem: "When you have an industry that revolves around glorified panhandling, it’s inevitable that participants end up getting kicked and spat on." 

Things happen.

AT&T’s Blockbuster Deal for Time Warner Hangs in Limbo. Sunac to Pay Record $9.3 Billion for Wanda Hotels, Theme Parks. China prepares fresh round of state-orchestrated megamergers. Yieldcos are out of fashion. Fiduciary-Rule Review Zeroes In on Industry Costs, Liabilities. This Giant Metals Exchange Is Taking on the Gold Elite. How the Qatar Crisis Shook Up the World's Supply of Helium. Goldman Sachs is the worst-selling fund manager in 2017. Smart beta funds stalked by chaotic ‘factor zoo.’ "Given that security analysts are supposed to be sophisticated investors, and all the anomalies analyzed were in the publicly available literature, it is surprising to find that while analysts’ return forecasts predict stock returns, they do so in the wrong direction." The Private Equity Firm That Quietly Profits on Top-Selling Drugs. Delinquent ducks arrested for loitering in Pennsylvania. MTA chair: Commuters should just go to work earlier.

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.

    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.