The Diff - Longreads + Open Thread

Longreads

• Abraham Thomas has a delightful retrospective on an effort in the 90s to do what then counted as high-frequency trading in treasury bonds. A fun read for anyone who wonders how designing data intensive applications worked back when RAM was measured in megabytes. Like many other quanty stories, it's right at the intersection of trading and tech; sometimes the constraints that mattered were things like getting funding for trades, sometimes they were limitations on processing power. And in one case, the two combined nicely: at the time, prime brokers couldn't continuously track positions, so they'd set collateral requirements based on what those positions looked like at the end of the day, so intraday strategies were very capital-efficient. One thing this particular kind of project showcases is the tradeoff between elegant solutions and acceptable hacks. In this case, "perfect" took hours, "good" took tens of seconds to minutes, and hacky updates that involved treating nonlinear features as linear-enough-to-get-by were basically instantaneous. As it turned out, being the fastest to get good-enough data was better than insisting on the best.
• Scott Sumner explores why the monetary base matters. It's a sort of asynchronous debate with Tyler Cowen and Alex Tabarrok. And it's a fun one, because both sides make some very intuitive points: if what we generally use is some form of credit (most of the "cash" you have is a database entry in a bank, and it's tempting to think about how a financial system would work if it had an absolutely tiny monetary base—like Feynman's electron, a single penny zipping around the system to settle whatever debts can't be settled through corresponding debit and credit entries at various intermediaries). But another way to look at those bookkeeping entries is that they're all a form of leverage, making the value of the underlying asset supremely important.
• Ivana Greco on the economics of homemakers. (The term "homemakers" sounds a little fussy and old-fashioned compared to the modern "stay-at-home mom," but as many people can attest, being the primary caregiver for children, primary shopper, primary runner-of-errands-that-occur-during-business-hours, etc. involves plenty of moving around and not that much staying.) One of the interesting points the piece makes is that this work is not counted in GDP, and in fact it's GDP-accretive to replace in-kind services provided by a full-time parent with lower-quality ones provided through their after-tax income—but also that 34% of government spending goes to Social Security and Medicare, two more measurable substitutes for this previously-unmeasured effort.
• A fun paper from David Leite combines spending and firm-formation data in Portugal to argue that business owners route personal spending through their companies, avoiding taxes that amount to roughly 1% of GDP. The paper shows that when people form companies, there isn't a big effect on their personal expenditures for things like rent, utilities, and schools. But they show a big drop in spending on travel, dining, car repair, and other plausibly business-related activities. The aggregate data is important, but the real fun is in the details: "I find that business expenditures on hotels and restaurants significantly increase by 9.8% in the birthday month of the owner-manager and by 6.1% in the birthday month of the owner-manager’s spouse (only owner-managers whose spouse does not work in the same firm are considered). Only expenditures on hotels and restaurants react to the birthdays, and I find no evidence of changes in the pattern of business expenditures around the birthdays of employees."
• A team at DeepMind looks at how AI can be applied to science. Most of the paper is high-level, but they do provide a great summary of which problems are amenable to these approaches: " huge combinatorial search spaces, large amounts of data, and a clear objective function to benchmark performance against." It's also a good meditation on how scientific progress happens: the tools one generation uses to make new discoveries are the product of a previous generation of discoveries. There's a lot to be optimized in this space: talking to STEM grad students, I get the impression that a PhD is basically a poorly-paid four-year stint in MLOps, with a paper at the end.
• In this week's issue of Capital Gains, we look at passive investing, particularly the fact that the more ubiquitous passive investing gets as an investment approach, the fuzzier its definition becomes. A 100%-passive world would be weird, but the more you dig into the concept, the harder it is to imagine any realistic path to its existenceOver time, there will probably be a cycle between active and passive, and it only looks like an inexorable trend because we're in the first wave of that cycle.

Open Thread

• Drop in any links or comments of interest to Diff readers.
• What are some good pieces like the treasury bond trading one above that go into some detail on older software projects. Everyone who learns to program gets a sense that people who learned after them don't know how much they take for granted, and it's always helpful to extend that sentiment in the other direction.

Diff Jobs

Companies in the Diff network are actively looking for talent. See a sampling of current open roles below:

• A growing pod at a multi-manager platform is looking for new quantitative researchers, no prior finance experience necessary, 250k+. (NYC)
• A company building the new pension of the 21st century and enabling universal basic capital is looking for a head of capital markets to help scale capital markets (debt and equity) efforts. Background in banking, hedge funds, or fintech, including embedded lending, preferred. (NYC)
• An AI startup building regulatory agents to help automate compliance for companies in highly regulated industries is looking for account executives or consultants with direct experience selling into large financial institutions (banks, asset managers, insurance companies). (NYC)
• YC-backed AI company that’s turning body cam footage into complete police reports is looking for a tech lead/CTO who can build scalable backend systems and maintain best practices for the engineering org. (SF)
• A hyper-growth startup that’s turning customers’ sales and marketing data into revenue is looking for a product engineer with a track record of building, shipping, and owning customer delivery at high velocity. If you like to build, this role is for you. (NYC)

Even if you don't see an exact match for your skills and interests right now, we're happy to talk early so we can let you know if a good opportunity comes up.

If you’re at a company that's looking for talent, we should talk! Diff Jobs works with companies across fintech, hard tech, consumer software, enterprise software, and other areas—any company where finding unusually effective people is a top priority.