The Ever Auric Age
Vol #156: November 30th, 2023
It’s been a little over a year since I started taking AI art generation seriously.
Thanks to the archives at the Till and the reminder of Microsoft's (finicky and irritating) OneDrive memory, I can pinpoint when this conversion happened. As my readers might expect, it was somewhere between Monet and merlot.
I first wrote about AI art as being fairly “mean”. Leaning into the nuts and bolts of how most machine learning algorithms work, I suggested that all AI art was good at, was averaging a lot of data points. “I’ve yet to see an example of it creating something truly artistically praiseworthy. A romantic might say that’s because a machine can never understand the human condition.”
The application that hooked me initially was an artist deconstructing impressionist paintings into fairly average contemporarily styled pieces. While this was very cool, and a little bit gimmicky, I still believed that Giverny could only truly come alive under Claude’s brush.
In an evolution that seems to shatter the principles of Moore’s law, within a few short weeks I was already creating unique, almost ‘good’ art. They were landscapes of wine regions of course. Taking the descriptions from Parr’s Atlas of Taste, or Johnson and Robinson’s own Atlas of Wine, I tried to create vineyard pictures that were representative of Bordeaux, Burgundy, and the Rhone. These microchips didn’t quite understand the terroir, but they were making a fairly good impression of it.
The images that I use for these blogs have gotten progressively better. I’m finding that OpenAI’s DALL-E tends to be slightly more creative, and is approaching a developed sense of context, perhaps even humor. The most contemporaneous LLM-meme is taking a stereotype and making it progressively more extra. Here Rohan Krisnan takes an e-mail template and re-writes it increasingly more “corporate”. (His AI book is excellent.)
This task requires an advanced level of awareness, and a pulse on the zeitgeist for nuanced interpretation. For us to find these amusing, they have clearly struck a chord of shared understanding.
When Deep Blue to beat Kasparov at chess, it was a demonstration of brute force power. But for DeepMind to win a similar battle in Go, it was through previously unimaginable techniques - AI thought about the game differently and won. DALL-E isn’t quite Degas, but increasingly the outputs have their own artist’s wink.
While these server farms are experiencing their silicon renaissance, the term “Golden Age” has been thrown around recently for something far less interesting - private credit. Previously a generic combination of two words, this hot ‘new’ sector is bombarding every investor's inbox. Real estate funds are pivoting to credit experts, and managers are raising billions of dollars to capture this auric opportunity.
Combine the developing wit of DALL-E with a new financial bauble, and the details are revealing. The “Golden Age of Private Credit” is nine white males with perfect jawlines and Patagonia vests, smugly basking in opulence. Half of them are double fisting wine glasses.
Private credit is a corollary to private equity. Without the restrictions of Sarbanes Oxley, 10-Q filings, and the generally pesky fourth estate; companies are free to grow and flourish. Private equity is sophisticated capital that can take risks others won’t, and brings a long horizon with business expertise.
Many managers and a few investors have gotten very wealthy from private equity. The results for the underlying businesses are more of a mixed bag. For those who came late to the game or are getting second tier deal flow - you probably would have been better off in an index fund.
If private equity is buying shares in a business with a 200 page private placement instead of a brokerage account, private credit is taking the other side of the capital stack and lending dollars with an attractive and streamlined deal process. Why go to a bank with all the red tape and invasive diligence process when a charming and puffered WASP is ready to write a check?
The Randian notion that deregulated industry is more efficient has an infectious quality. No one has ever taken out a mortgage and appreciated how it couldn’t get any easier. Getting a business loan is a grueling and humbling process. The KYC rules alone are enough to squash a deal before the merits of cash flow and credit worthiness have even been broached.
It’s both intuitive and appealing that a local or “small” business looking to extend a loan would know their customer better. This results in better service and more informed investment analysis. Underwriters who are familiar with the industry and environment will take smarter risks.
The above logic isn’t wrong, but it’s too broadly applied and generically accepted. There are distinct times when smaller and local is better, but there are also good reasons for large and highly regulated institutions. For investors looking to allocate capital, it’s important to parse the difference between these.
If you ever get too optimistic about the potential of humans unshackled from financial regulation, spend a few weeks in crypto. That’s all it takes for a full cycle of securities fraud, from rug pulls to market manipulation to bridge hacks, to leave all but the most grizzled libertarians skeptical and looking for more than a safe harbor.
Even if not as sinister as shadowy super coders, when predictions of 11+% returns on lending are coming in, who is paying that, and how sustainable is it? Businesses with significant costs of capital need to be highly performant. As DeFi OGs know, if you don’t know where the yield is coming from, you are the yield.
There is an important place in the capital stack for the entire bucket of exempt securities. Private equity, credit, real estate, or whatever else you can fit under rule 506 exist to capture unique opportunities. The right mix can be a good diversification tool.
While funds capture unique market opportunities, they can also be a massive obfuscation of fees and illiquidity. The simplest and best allocation for most investors in the long run is a basket of low cost diversified index funds. The specific mix should meet your timeline and risk tolerance, but you can get almost any reasonable exposure you want for less than a quarter of a percentage point.
Options strategies sit at the border of practical and fanciful. Most of us are drawn in because of the latter; hockey stick convexity or the ability to protect AND participate. Even after seeing the raw open guts of options market structure and how the participants make money, I still wake up in the middle of the night with an idea for how to tweak an earnings strategy.
There’s constantly a new way to make money and use options. Effective opportunity capture is separating the fools gold from the real opportunity. We know there are inefficiencies everywhere. Selling puts during a bull market can lull us into easy income generation fantasies. Being long a moving option is exhilarating. The best edge isn’t sexy, and requires a grind of analysis, consistency, and emotional wherewithal.
Most of what we should be doing in options fits into the boring and vanilla. Equity buffering, stock replacement, or collaring strategies are practical ways to use the big blade of the options swiss army knife. The simplicity and strength comes from systematic execution.
The more complex tools have their uses too. A straddle swap before earnings, or broken wing butterfly into an event can present positive risk/reward scenarios. Collecting risk managed premium within a rule set is a great way to diversify your portfolio.
This is where things quickly get murky, and whether you’re evaluating your own PnL or a prospectus, know what the risks are. I love what QQQY is doing every day selling daily ATM puts, but wouldn’t buy it if I didn’t grasp the FOMO of what a bull market underperformance meant.
There’s an interesting AI litmus test. If your strategy gets sarcastically painted by AI, then it’s probably too good to be true. If it looks boring and uninteresting, you’re probably on to something.
“A fund manager executing a passive automated options strategy, rebalancing timely.”
“An options day trader using technical analysis, and various prediction methods to trade, while getting increasingly fanatical.”