Money Stuff

Distressed Divas and Family AIs

Also tax reform, floor traders, AI religion and angels.

Lynn Tilton.

Lynn Tilton -- the "Diva of Distressed," as her TV show called her -- ran collateralized debt obligations named "Zohar," essentially investment funds that would buy the debt of distressed companies and profit if Tilton could turn those companies around. Tilton's firm, Patriarch Partners LLC, was paid based on how many assets it managed, that is, on the value of the loans in each Zohar CDO. The value of those loans was determined by Tilton, an obvious but often unavoidable conflict of interest.

The Securities and Exchange Commission's enforcement division decided that Tilton was doing a bad job of valuing the loans, in ways that increased her own fees. The Zohar funds had two categories of loans: Category 1 were bad, defaulted, written-off loans; Category 4 (yes, 4) were good, promising loans. Category 4 loans were held on Zohar's books at whatever price Zohar had paid for them; Category 1 loans were written down. The SEC noticed that lots of Category 4 loans were in fact pretty troubled: The companies weren't making interest payments, or the interest payments had been reduced, but Zohar was still holding them at the same price it had paid for them. (And Tilton was still getting paid on those high values.) So the SEC sued Tilton in its own courts, and Tilton went all the way to the Supreme Court to try to avoid being tried before an SEC administrative law judge, claiming that it was unconstitutional. She lost, so she had her trial, and yesterday she won completely: The SEC judge ruled that Tilton's alleged violations "are unproven," and dismissed the case. 

Tilton's defense -- which we discussed a couple of years ago, when the SEC brought its case -- was basically that (1) all of her investors knew what was going on, and (2) it was, after all, distressed investing. The whole point of the Zohar funds was to invest in companies that were struggling with their debt, and to try to fix them by, in part, dealing with that debt. It would make no sense for Zohar to buy the debt of a company, lower its interest payments, and then immediately write down the debt; the point of lowering the interest payments was to increase the value of the debt by making the company more likely to survive. Tilton explained in the SEC court:

[I]f we were working up the company, we would hold it as a Category 4, amend and restructure as we saw fit to create value. If it were in a free-fall bankruptcy without a prepackaged plan, we would hold it as a Category 1. Or if it came to a point that we didn’t believe by putting more money in that we could create more value than where we were today, or it was going to take too much time, too much money and too much effort, then we would hold it as a Category 1.

So Tilton would waive various defaults on the debt if she was still trying to create value. Category 4 -- the good category of debt -- included debt with no "'event of default' or any 'default' ... which has not been waived," and the Zohar fund documents gave Tilton the right to waive anything in the underlying loans. So in her mind, debt that wasn't current was still Category 4 -- still valued at cost -- as long as she wasn't worried about it.

Not just her mind, though: That is what the documents said, and also what investors expected. "Everyone knew what we were buying, and everyone knew what the strategy was," Tilton told the SEC court. And the SEC judge agreed: Tilton had no intent to defraud investors, and investors had all the information they needed to understand what was going on, so there was no fraud:

The investors knew that the Funds’ business model was to lend to a number of distressed companies with the idea that, while some would succeed, enabling the Funds’ investors to be paid, others would fail. Thus, it would be unreasonable to expect that all the borrowers would make 100% of their interest payments.

It is not perhaps the most rigorous accounting you could imagine, but it's good enough not to be fraud.

Some AI.

There are two overly simplistic ways to think about artificial intelligence in investing. One way to think about it is, well, it's like a big regression analysis. You get a bunch of data, your computer crunches it and figures out which signals tend to correlate with stocks going up, and then it buys stocks that exhibit those signals. The computer, which can tirelessly look at a lot of numbers, can spot statistically useful patterns in those numbers that elude human pattern-spotters.

The other way to think about it is, like, "Blade Runner": You build a computer, and endow it with thoughts and feelings and a soul, and then it gets up and walks around and talks to you, and grows and changes and evolves as a person/computer, and your interactions with it are surprising and risky and heartfelt and moving. The artificial intelligence is a substitute for human intelligence; the computer is like you, but different: smarter and more logical and more data-driven, but with less capacity to feel love.

Here is a story about Man Group Plc.'s use of artificial intelligence that is delightfully sci-fi. They built the computer, and it got up and walked around, and they cowered in fear of it:

Sarcastically, he says the creation was kept on a separate server, as if it could somehow infect Man Group’s main computer system. “It used to sit in a nuclear bunker in the corner,” Ellis jokes, slumped casually in a chair off the main trading floor in London, drinking a soda. Ellis, who became CEO last year, was on the executive committee then. “Were we scared by it? Yes. You wanted to wash your hands every time you looked at it.”

There is a Robert Harris novel, "The Fear Index," on exactly this subject. (In fact it features in the kicker of this article.) In that book, a hedge fund's AI rises up to (spoiler alert) try to kill its founder. It is possible that everything you need to know about hedge-fund AI is in that book, in the sense that, if you are a hedge-fund founder dabbling in AI, and your AI does rise up to kill you, that will have been the most important thing you could have known about hedge-fund AI, much more so than any technical questions about the structure of neural networks. 

But this is not a story of murder. It is a story of growth and change, of two different cultures (human, AI) learning to love and respect each other. And so Man Group Chief Executive Officer Luke Ellis later says:

“It went from a total isolation to ‘OK, you are allowed to sit at dinner with the rest of us, but don’t talk’ to the point where it’s become a part of the family,”

I will be disappointed if this is just the CEO anthropomorphizing the AI to journalists, and not a real part of the company culture. Does the AI have a name? Are they polite when they make requests of it? Do they invite it out to after-work drinks? Also how do they discipline and stimulate it?

AI engineers use punishments and rewards to guide the machine; it’s like teaching a mouse to push a button for food. In what’s broadly known as deep learning, algorithms are trained to hunt for predictive patterns within libraries of historical information. They are “stimulated” when they find similarities in, say, the pricing data of stocks or commodities. In another approach, called reinforcement learning, the machine recalibrates itself on the go according to the success or failure of certain actions. Researchers also program penalties into algorithms to discourage certain behavior by the AI, such as creating strategies that are too similar to those humans already employ.

Given the general commitment to anthropomorphism, I imagine a leather-clad dominatrix standing over the computer, ready to administer punishment as necessary.

There are other delightful sci-fi elements in the story. For instance, here's how the program does stuff:

He hits enter and a stream of numbers rains down the screen, like those Hollywood graphics that run behind the opening credits of a techno-thriller:

0.3426383
0.237250642
0.53534377

The cascading numbers show the computer “thinking,” crunching data at speeds a human could never achieve.

Presumably you don't have to display all the numbers the computer is thinking about, right? Like, the numbers are inside the computer; it doesn't have to put them on the screen to read them. Perhaps the first priority in designing an AI system is to make it look like "The Matrix"?

"Finance is perhaps AI’s most daunting challenge," but I don't know, man. You can go a long way in finance with some regressions. There is a lot of data, and a lot of it is fairly transparent and structured, and people have been profitably finding patterns in the data for many decades. Driving a car -- responding to the chaotic stimuli of the physical world -- is hard. Building artificial-intelligence replicants with the appropriate balance of conscientiousness and empathy to hunt other replicants is hard. Spotting head and shoulders patterns or whatever in stock prices is relatively easy.

Elsewhere: "Wells Fargo Analysts Build the Robot That Could Take Their Jobs."

Taxes.

Yesterday the Trump administration unveiled a pretty large middle-class tax increase to pay for tax cuts on partnership income and millionaire heirs for some reason, probably populism. This plan is slightly less napkin-like than the previous tax plan from the administration, though it still leaves a lot of details to be filled in. Many of the fun details are in business taxation: The plan would reduce the top corporate tax rate to 20 percent, which is easy enough -- but it will also move to a territorial system of taxation, a more complicated transition that is not especially spelled out. The plan would allow businesses to immediately deduct a lot of capital expenditures, though only temporarily, but it also provides that "the deduction for net interest expense incurred by C corporations will be partially limited." How? Obviously if you are a bank, and you make billions of dollars of interest revenue and pay most of it out in interest expense, you will be keenly interested in whether your interest expense remains deductible.

Also income from pass-through entities will get special tax treatment:

The framework limits the maximum tax rate applied to the business income of small and family-owned businesses conducted as sole proprietorships, partnerships and S corporations to 25%. The framework contemplates that the committees will adopt measures to prevent the recharacterization of personal income into business income to prevent wealthy individuals from avoiding the top personal tax rate.

I don't know what they're talking about. As the proprietor of a family-owned small business, Money Stuff Business Thing LLC, that produces a high-quality artisanal newsletter out of locally-manufactured parts and sells that newsletter to a big American-owned company, Bloomberg LP, I am horrified by the idea that wealthy individuals might try to avoid the top personal tax rate by recharacterizing their personal income into business income. 

NYSE humans.

Here is "Why robot traders haven’t replaced all the humans at the New York Stock Exchange—yet," a question I have often wondered about; I assumed that the answer was "shouting humans make for a better backdrop for television shows." I still do, really, but there is another piece of the answer:

The humans on NYSE’s floor have a special advantage: Brokers can use the d-Quote, which gives them almost 15 minutes of extra time to tweak or add stock orders at the end of trading, which can be the most important price of the day. In the world of computerized trading, as one trader put it, that quarter of an hour is like a few months in human time: news can break, and thousands of other trades can take place during that 15 minutes.

The only way to access the d-Quote is through a floor broker with a handheld device. This order type is incredibly popular, and it means that a significant amount of vital trading still involves human reaction. Market-structure experts say it could probably be done without a trading floor, however.

That is: NYSE's rules favor floor brokers by giving them an advantage in late-day (entirely electronic!) trading. If it just gave that advantage to everyone -- or to any group of favored electronic traders selected however NYSE liked -- then the human advantage would vanish, and you'd be left with vague puffery like "you lose value when you no longer allow for human interaction," as well as the TV-set thing.

What dumb thing are tech bros worshipping now?

Ah sure right Anthony Levandowski is starting a religion:

 In September 2015, the multi-millionaire engineer at the heart of the patent and trade secrets lawsuit between Uber and Waymo, Google’s self-driving car company, founded a religious organization called Way of the Future. Its purpose, according to previously unreported state filings, is nothing less than to “develop and promote the realization of a Godhead based on Artificial Intelligence.” 

That is a bit of a bait-and-switch; the article is actually a deeply reported profile of Levandowski and the Uber/Google dispute over self-driving cars. There's not much more about his religion. But good lord -- good bot? -- there should be. I assume they need someone to write their scriptures. (Or will a neural network do it?) "Thou shalt move fast and break things." "Thou shalt not covet thy neighbor's lidar." "A million dollars is without form, and void, but dost thou know what is cool? A billion dollars." 

Blockchain blockchain blockchain.

Here is an information sheet from the Australian Securities & Investments Commission finding that initial coin offerings are likely to be regulated "managed investment schemes" -- even if they are characterized as pre-sales of products:

In some cases, ICO issuers may frame the entitlements received by contributors as a receipt of a purchased service. However, if the value of the digital coins acquired is affected by the pooling of funds from contributors or use of those funds under the arrangement, then the ICO is likely to fall within the requirements relating to MISs. This is often the case if what is offered through the ICO has the attributes of an investment.

This generally seems right to me: Most people buying tokens in today's ICOs are doing so to speculate on the value of an enterprise, not to pre-buy a product, and so if you are a securities commission that regulates speculative investments in enterprises, it kind of makes sense that you'd want to regulate ICOs.

Elsewhere: "This 31-Year-Old Is Trying to Revolutionize Cryptocurrency Trading." "How does Ethereum work, anyway?" And: "Showtime's Websites May Have Used Your CPU to Mine Cryptocoin While You Binged on Twin Peaks."

People are worried about unicorns.

Here is a New York Times article about Jason Calacanis, who invested $25,000 in his friend's tech company and ended up making $100 million because the friend was Travis Kalanick and the company was Uber Technologies Inc. Now he has a book out called "Angel: How to Invest in Technology Startups — Timeless Advice From An Angel Investor Who Turned $100,000 into $100,000,000," and I wonder if any actual lottery winners do that. "Powerball: How to Invest in the Lottery -- Timeless Advice From A Lottery Winner Who Turned $1 into $400,000,000." "I used to be just like you, sitting on my couch, wishing I was rich, playing the lottery. Then I won the lottery and now life is great! Let me tell you my secrets." I mean why not. Luck, if you have it, is easy to mistake for genius.

Also Calacanis wants everyone to become an angel investor to stave off a "full-on revolution in the streets": The tech industry, he expects, will create massive unemployment by automating away jobs, and the only way to avert social unrest and revolution is if all the people automated out of jobs get rich as tech investors themselves. Let them eat seed rounds, is I guess the message.

Elsewhere, Elaine Ou writes: "The most important tech skill, then, isn’t computers or engineering — It’s the art of getting paid to control vast amounts of money." And here's a unicorn map.

Things happen.

Puerto Rico creditors jockey for position in Hurricane Maria aftermath. What happens in Vegas ... the messy bankruptcy of Caesars Entertainment. Equifax Will Offer Free Credit Locks for Life, New CEO Says. JPMorgan Ordered to Pay More Than $4 Billion to Widow and Family. Carlyle Group in Talks to Sell TCW Group Stake. Value of private equity dealmaking at highest level since 2007. Hain Celestial in Pact With Activist Investor Overhauling Board. The siren call of T+0, or real-time settlement. Agencies propose simplifying regulatory capital rules. Saudi Insurers Soar After Decision to Allow Women to Drive. Merger Negotiations in the Shadow of Judicial Appraisal. SEC Probes Departure of PepsiCo’s Former Top Lawyer. Uber Shutting Down U.S. Car-Leasing Business. "Value has shifted away from companies that control the distribution of scarce resources to those that control demand for abundant ones." The Coming Software Apocalypse. The Male Model Who Lost His Hair. Legal Weed May Be a Windfall for McDonald’s and Taco Bell. The Repressive, Authoritarian Soul of "Thomas the Tank Engine & Friends."

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 mlevine51@bloomberg.net

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

    Before it's here, it's on the Bloomberg Terminal.
    LEARN MORE
    Comments