A daily market update from FS Insight — what you need to know ahead of opening bell

“The best way to predict the future is to create it.” — Peter Drucker

Overnight

The odds of a government shutdown seemed to drop after Tuesday’s meeting between the Big Four congressional leaders and President Biden at the White House

The Atlanta Fed’s real-time GDP tracker climbs back up to 3.1% for the current quarter

Bitcoin surpasses $57,000 as overall market value of digital currencies is now ~$2.2 trillion, recovering from a 2022 low of $820 billion

U.S. launches antitrust investigation of healthcare giant UnitedHealth

BOE may sell entire QE portfolio to better prepare for a future crisis

Barclays to offload $1.1 billion of card debt to Blackstone

Northeast municipal bond volume took hardest fall of any region in 2023

Chevron, Hess shares slip as Exxon considers bid to block Guyana deal

Office tower deal for $1 reveals anxiety among longtime buyers

130,000-plus Ugandans may sue over unpaid wages for U.S. security work in Iraq, Afghanistan

Three security pacts recently signed by Somalia highlight simmering tensions in the Horn of Africa threatening to set off a new conflict in an increasingly volatile region

Russia would use nuclear bombs on the battlefield despite not being attacked with them, leaked documents suggest; distrust of China

Shunted into the background by other conflicts, Sudan’s civil war spirals out of control

OpenAI files motion to dismiss New York Times lawsuit

Apple ends Project Titan, a quest to build its own electric vehicle; shifts some car-project employees to AI work

First news

• Microsoft wisely broadens its portfolio of AI partnerships, welcoming diversity in the approach to AI models by linking up with a young European startup
• Catastrophe bonds, the best hedge fund strategy of 2023, may be a disaster in waiting going forward.

Chart of the Day

MARKET LEVELS

MORNING INSIGHT

Good morning!

Forty-one companies are reporting this week. Of the 474 companies that have reported so far (95% of the S&P 500):

• Overall, 77% are beating estimates, and those that “beat” are beating by a median of 7%.
• Of the 23% missing, those are missing by a median of –6%.
• On the top line, overall results are beating estimates by a median of 5% and missing by a median of –5%, and 64% of those reporting are beating estimates.

Click HERE for more.

TECHNICAL

Consumer Discretionary vs. S&P is slowly but surely making strides in turning higher. Investors should watch for evidence of a breakout.

Discretionary is making good strides lately, with the Equal-weighted Invesco Consumer Discretionary / Equal-weighted S&P ETF ratio (RSPD 0.17% /RSP 0.09% ) moving to multi-week highs as stocks like AZO -0.56%  CCL 0.41%  CZR -0.58%  HAS 2.09%  NCLS ORLY -1.25%  VFC TSCO are all showing above-average gains in yesterday’s trading. 

However, on a larger scale, there hasn’t yet been a material breakout to suggest intermediate-term outperformance in Consumer Discretionary. We do suspect this will happen in 2024, and once this larger downtrend is broken in the relative ratio of RSPD 0.17%  to RSP 0.09% , we’ll see some outsized relative strength in Discretionary names as this sector will begin to outperform in the near-term, technically. 

It’s early to overweight, but we’ll be watching; this Discretionary sector certainly has some names of interest. As many know, we favor homebuilders like LEN 0.02% , PHM 0.26%  and DHI 0.49%  along with AMZN -0.08%  and DPZ -0.09%  (on my UPTICKS list), but also feel that casinos and cruise-liners are attractive stocks for mean reversion outperformance in 2024. For example, we like CCL 0.41% ’s move yesterday, technically, which is exceeding a two-month downtrend, and this might jumpstart this stock in the short run. 

Below is the ratio of Discretionary to S&P 500, both shown in Equal-weighted terms.

Click HERE for more.

CRYPTO

In this video, we discuss the Coinbase premium, what the return of animal spirits could mean for DOGE, what funding rates suggest about current market froth, and a potentially overlooked detail about Coinbase’s balance sheet.

Click HERE for the video.

FIRST NEWS

The C(h)at In the Windows. Seeking to expand its AI footprint beyond its relationship with Sam Altman’s maverick maker of ChatGPT, Microsoft announced a new partnership with young French startup Mistral AI – Europe’s answer to OpenAI. Mistral develops algorithmic models similar to those from OpenAI, used for chatbots and other AI services – but, crucially, Mistral models are shared openly. Under the terms of the deal, announced on Monday, Mistral’s large language models will be available on Microsoft’s Azure cloud-computing platform, while its ChatGPT-style multilingual conversational assistant “Le Chat” (which means the cat, but also the chat, in French) is rolled out.

As part of what Mistral announced as a ‘strategic partnership’, Microsoft will be investing 15 million euros ($16.3 million) at a 2 billion euro ($2.1 billion) valuation* into the not-quite-one-year-old business, to help it “unlock new commercial opportunities” and expand to global markets. Mistral will become only the second company, after OpenAI, to host its large language models (LLMs) on Microsoft’s Azure cloud-computing platform, while training and deploying their next generation models for AI on Microsoft’s AI data centers. The name of the offering is Mistral Large, and Microsoft is calling it its most advanced LLM. Customers will be able to use Mistral Large on Azure AI Studio and Azure Machine Learning for any language-based application requiring reasoning, understanding, coding, and text-generation capabilities.

The deal comes as Microsoft faces pressure from E.U. and U.K. antitrust regulators over its reported $13 billion investment in San Francisco-based OpenAI. Perhaps as a way of appeasing E.U. technocrats, Microsoft is casting this newest investment as a rising tide that lifts all ships – in this case, specifically those on the other side of the pond. Asked whether the partnership with Mistral was an effort to appease competition concerns, a Microsoft executive said it was going to be “an engine for technology, innovation and growth in Europe”, and added that the investment in Mistral AI would also see funds dedicated to R&D; among other things, to AI models for public-sector services in Europe.

Some technocrats aren’t buying the explanation. The Mistral investment is set to be analyzed by the European Union’s competition watchdog while Microsoft’s deep ties to OpenAI also come under regulatory scrutiny.

Named for the strong, cold wind that blows from southern France into the northern Mediterranean, Mistral is founded in part on the idea that a lot of the money that American Big Tech is throwing at AI is being wasted. In that sense, it’s as if Mistral were a wind of change. Co-founder Artur Mensch says the startup exists “to be the most capital-efficient company in the world of AI”. Proof: Mistral’s new model cost less than €20 million ($22 million) to train. OpenAI CEO Sam Altman ballparked the cost of training his company’s biggest models at “much more than” $50 million to $100 million. (Altman also famously estimated that he would need $7 trillion to build out a chip-making infrastructure servicing AI’s enormous needs.)

The refreshing approach of not throwing limitless amounts of cash at problems, and instead solving big, thorny challenges with relatively small amounts of cash – more or less the classic modus operandi for startups up until a couple of decades ago – is getting noticed. Mistral has already partnered with and sold small stakes to other tech behemoths, including enterprise-software company Salesforce as well as Nvidia.

Mistral’s through-line has been efficiency and the use of AI scaling laws (understanding the relationship between the size of an AI model, how much data is used to build it, and how well it performs) to build qualitatively as-good or better models at a fraction of the cost of competitors’ offerings.

Mistral’s release of its initial AI systems as open source (that anyone could use or adapt free of charge) was important in principle, while also a good way to get noticed by developers and potential clients who value greater control over the AI they use. Following a kind of freemium model, Mistral reserves its most advanced models, including the one unveiled Monday, for paying customers.

OpenAI also started out as a champion of open-source, but quickly turned into the very opposite. Mistral may have learned the lessons from OpenAI’s quick path from the equivalent of Google’s Don’t be evil (‘Don’t be venal’) to a gleeful embrace of closed-source and world domination. Says Mensch, “It’s obviously a thin balance between building a business model and sticking to our open-source values. We want to invent new things, new architectures, and we still want to have something to sell extra to our customers.”

Mensch added that while Mistral was not yet making any money, he expected that to change “before the end of the year”, as the company readies a new platform for customers to access its AI models.

For now, the most valuable thing Mistral has offered is a lesson in classic startup pluck. Is an elaborately designed mousetrap better for having cost enough at the R&D stage to be worth its weight in gold? Or is a beautifully engineered contraption quickly, elegantly wireframed with copper wire and plywood worth more, in principle, for having been developed on a mouse-sized budget – with the same elephant-size impact? CNBC, WSJ, Microsoft, Reuters, TIME, FT

^{* Microsoft’s investments will convert into equity as part of Mistral’s next funding round. Interestingly, the $2 billion mark is roughly the same as what Mistral was valued at two and a half months ago, when it was valued at 2 billion euros post-money during a 400-million-euro equity-and-convertible-debt funding round led by Andreessen Horowitz. Other investors in Mistral (apart from those mentioned above) include former Google chief Eric Schmidt, French telecoms billionaire Xavier Niel, Bpifrance (the French state-backed investment bank), BNP Paribas, and General Catalyst.}

The Cat Who Sat in the Eye of the Storm. As the best hedge fund strategy of 2023 becomes a magnet for mainstream investors, the risk models it relies on are getting a lot tougher to crack. The strategy in question is tied to securities that are themselves linked to a kind of insurance dominated by catastrophe bonds (known as cat bonds). In 2023, no other asset class produced a better-performing bet for hedge funds, with firms including Fermat Capital Management and Tenax Capital booking their biggest-ever returns.

Cat bonds have been around for over a quarter-century and are used by the insurance industry to shield itself from losses that would be – well – catastrophic for the finances of any one firm. That risk is instead transferred to investors who lose money if a pre-defined catastrophe hits – such as a hurricane causing $500 million in insured losses or an earthquake reaching a magnitude of 7.0 – and rake in force-majeure-level returns if it does not.*

Yet calculating catastrophic risk is much more complex than it used to be on account of a growing concentration of property in areas prone to increasingly frequent storms, fires, and floods. Individually, each event tends to be less intense than a major earthquake or hurricane. Taken together, however, the losses from them can be a lot more serious, and that has major implications for the growing numbers of investors now adding exposure to cat bonds.

Traditionally, cat bonds have been used to protect insurers from the kinds of losses associated with once-in-a-generation natural disasters. While those primary perils, as they’re known, accounted for only 14% of global losses in 2023 – per broker Aon Plc, a category known as secondary perils has “outpaced their cumulative costs in the 21st century by a large margin.” 

Fund managers monitoring the development of the niche space say these secondary perils, which most often materialize in the form of destructive thunderstorms, aren’t being consistently captured by models designed to measure cat bond risk. “We see that some models are actually not pricing these perils adequately,” says Etienne Schwartz, head of investment management at Twelve Capital, which holds $3.7 billion of cat bonds. In fact, he warns that “the expected loss on paper is way below what we actually think it is.” Bloomberg, Chicago Fed

^{* Unlike traditional reinsurance, where it is possible for the reinsurer to fail to pay out following a loss event, CAT bonds are 100% collateralized and structured to eliminate counterparty risk. In a typical cat-bond contract, the primary insurer enters a reinsurance agreement with a special-purpose vehicle (SPV) specially created for the transaction. A legal entity created to hold the capital raised from investors, the SPV issues cat-bonds to capital market investors up to a specified limit.}