We talk sometimes around here about hedge funds getting into weather forecasting. Weather affects financial markets (notably commodities prices), and it is hard but not impossible to predict; it is a problem to which you can apply skill and technology. If you are very good at predicting the weather, one particularly lucrative place to put your skills to work is in commodities derivatives trading. 

I wrote about this last year▸ AI summary:

The nicest thing you can say about the financial industry is that it lavishly rewards correct understanding of the world. If you know a thing, and other people do not know it, financial markets provide an efficient and fairly general way to turn your knowledge into large amounts of money. …

It seems to me that this creates nice incentive structures? … It is probably net good for the world that young people who are interested in rigorously understanding some specific domain know that they might get paid really well for that understanding.

I still think that this is basically true, but one could have a slightly more cynical and Keynesian model. Something like:

1. Financial markets reward, not correct understanding of underlying reality, but correct predictions of other people’s beliefs: Stocks go up because people believe that their future cash flows will be higher, not because the future cash flows will actually be higher.
2. These things are related: Financial markets are smart and competitive, and the main reason that people believe stuff is that they have good reason to believe it. So getting really good at predicting underlying reality is broadly, if indirectly, useful for predicting beliefs and, thus, prices.
3. But the overlap is not total. If you can choose between correctly predicting underlying reality and correctly predicting market perceptions, predicting market perceptions might be easier and more lucrative, and also more immediately lucrative. 

Bloomberg’s Joe Wertz has a story about predicting weather predictions▸ AI summary:

In Europe’s weather-driven energy markets, traders are turning to AI and machine-learning tools designed not to predict temperatures and precipitation, but to forecast the forecast.

That means predicting whether the European Centre for Medium-Range Weather Forecasts’ two-week outlook — the definitive reference point for traders repricing risk around heating demand, renewable output and system tightness — is about to shift warmer or colder.

To predict the next turn in the so-called Euro ensemble run before it crosses the wires, weather analytics firm Atmospheric G2 launched ForecastEdge in October. The prize is anticipating the ECMWF’s next forecast shift and profiting from it before gas and power curves move.

“We predict colder or warmer moves, that’s the key signal,” said Andrew Pedrini, an AG2 meteorologist. “The market reacts to the moves more than the actual accuracy of the model.” ...

AG2 won’t disclose exactly how ForecastEdge works, but Pedrini said it uses a combination of machine learning and AI with a statistical model of historical ECMWF forecast changes. He said the tool is optimized for directional accuracy, rather than precise degree changes. It focuses on the middle-to-end of the two-week outlook, capturing the most volatility and potential profits.

In artificial intelligence there is a concept of “reward hacking”: You want to train an AI model to do some task, you set up some reinforcement learning system for the AI to learn the task, and the AI learns not to do the real task but rather some simpler task that scores maximum points in training. (One example from Victoria Krakovna’s list: “I hooked a neural network up to my Roomba. I wanted it to learn to navigate without bumping into things, so I set up a reward scheme to encourage speed and discourage hitting the bumper sensors. It learnt to drive backwards, because there are no bumpers on the back.”)

This is a little like that: The task, for these machine-learning tools and also for the people running them, is not to predict the weather, but rather to predict how ECMWF will predict the weather, because markets are moved not just by the weather but specifically by the ECMWF forecast. If you optimize for that, you might get a good weather prediction, but it’s not essential.

Look, multibillion-dollar public-company takeover fights are exciting. They just are. Barbarians at the Gate is a thriller, and as I have said▸ AI summary before, I read it at an impressionable age and wanted to grow up to do mergers and acquisitions. I was a weird kid, but surely not unique. A lot of people still want to grow up to be characters in Barbarians at the Gate, to put on crisp pinstriped suits and stride purposefully into conference rooms to announce large risky bids to take over iconic companies. I wrote like a zillion columns about the recent battle between Netflix Inc. and Paramount Skydance Corp. to buy Warner Bros. Discover Inc. It’s a classic.

Some people like to write about that stuff, and some people like to read about that stuff, and some people — bankers and lawyers and corporate executives — like to do that stuff, to be in the room where it happens, to be part of the strategizing and negotiating and bluffing and skullduggery. Probably a lot of people in those rooms wish they weren’t — I did grow up to be an M&A lawyer, and I often wished I was home in bed! — but also a lot of people who are not in those rooms wish they were. Some for good sensible professional reasons — if you’re a media banker or M&A lawyer, and you weren’t on the Paramount deal, you got some dirty looks from your boss — but some just for fun. To a certain type of person, a type that includes me, it just seems like it would be cool to be a part of that high-stakes battle.

You could always just … lob in a bid? Send Warner’s board a letter like “hello I have been studying your little company on behalf of my gigantic but very secretive family office, I think you are doing good work, and I would like to pay $32.50 per share in cash for your whole company.” That would be, nominally, the highest bid for the company — Paramount ended up paying $31 — and maybe they’d go for it. Send you back a merger agreement to mark up, schedule a call to go over your financing, invite you to fly out and sit in a boardroom and say cool M&A stuff. Worth a shot! Obviously if you don’t have a gigantic family office, or the $110 billion it would take to buy Warner, this would not go very far. But you might get a few fun hours of pretend M&A negotiations out of it. Plus, if this sounds like a good idea to you, you might be a little delusional; maybe you think it will go far!

Anyway none of this is legal advice, and in some quarters fake merger proposals are frowned upon and considered “securities fraud.” Usually, though, that’s because the fake merger proposals are announced publicly and have the effect, and the intent, of moving the target’s stock price: You buy some stock, you announce the fake merger, the stock goes up, you sell, profit, fraud. Sometimes, though, we talk around here about fake merger proposals that are not like that. A guy really wants to get in the boardroom to pretend to negotiate a pretend merger, and maybe also get on TV. He’s not even trading the stock. He’s not doing it for money. He’s doing it for sheer love of the game.

At Puck, William Cohan reports:

Back in February, about a week before David Ellison [of Paramount] and David Zaslav [of Warner Bros.] announced their merger, a mysterious Singaporean entity called Nobelis Capital, Pte., Ltd., appeared out of nowhere with a $32.50-per-share, all-cash bid for all of WBD. From the jump, there were red flags: The Nobelis bid “did not include any evidence of equity or debt financing, nor a definitive transaction agreement,” according to a revised WBD proxy filed on March 16. What’s more, nobody on Wall Street seemed to have heard of them. And when WBD took the time to try to figure out if they were legit, it came up empty too. “We assumed it wasn’t real based on several factors,” a WBD spokesman told me, before pointing to the recently filed proxy statement.

From the proxy:

On February 18, 2026, WBD received an electronic communication from Nobelis Capital, Pte. Ltd., an organization based in Singapore (“Nobelis”), purporting to submit a “binding offer” to acquire 100% of the WBD Common Stock for $32.50 per share in cash (the “Nobelis Proposal”). The Nobelis Proposal did not include any evidence of equity or debt financing, nor a definitive transaction agreement. The presentation accompanying the Nobelis Proposal indicated that the proposal might be for $32.50 per share in cash or “$28.00 + Equity Participation”. The Nobelis Proposal indicated that Nobelis had deposited $7.5 billion into a “Tri Party Escrow account” with J.P. Morgan to cover the regulatory termination fee, and in the attached presentation described a $10 billion “immediate deposit” into HSBC Bank to cover the $2.8 billion Netflix Termination Fee that would be payable by WBD to Netflix, $7.2 billion in an “Execution Bond” and the $7.5 billion regulatory termination fee, among other terms. WBD’s legal and financial advisors conducted preliminary due diligence regarding Nobelis, including through professional contacts in Singapore, as well as an investment banker identified in the proposal, but were unable to verify that Nobelis owned or controlled any material assets, and could not find the purported deposit at J.P. Morgan. The investment banker advised counsel to WBD that he had no knowledge of Nobelis and had not been retained by them.

“Tri Party Escrow account,” I love it. How fun would it have been, for the real bankers and lawyers on the Warner Bros. deal, to spend half a day on “preliminary due diligence” on the obviously fake bidder? “Not fun at all” is the practical answer; those people were super busy and did not have time for this nonsense. Still I hope they got some joy out of it. They were living the dream, and Nobelis, whoever that was, just wanted to be a part of their world.

No, fine, there was probably a financial angle. From the proxy again:

On March 9, 2026, Nobelis Capital sent further communication to WBD which, among other things, threatened various legal actions against WBD unless it entered into a “settlement framework” within forty-eight hours, which would include public disclosure of the Nobelis Proposal, WBD giving seven-day access to Nobelis to verify certain matters, and payment by WBD to Nobelis of a termination fee equal to “3.5% of the transaction value,” plus “full Expense Reimbursement for the costs of establishing our $10 Billion escrow architecture.” 

I mean, Netflix got a termination fee, shouldn’t a fake bidder get one too?

My high-level theory of private credit is that:

1. Traditionally banks would take short-term deposits and make long-term loans to people and businesses, creating risky maturity mismatches.
2. Now private credit funds take long-term locked-up equity investments and make long-term loans, a far more sensible funding model that avoids maturity mismatches.
3. Banks are still around, still taking short-term deposits, still making long-term loans. But instead of the traditional business of lending to people and businesses, now the banks lend to the private credit funds. 

The banks are “re-tranching”: Instead of doing some risky activities (like lending to businesses) directly, they lend to people who do the risky activities. They take a more senior claim on the risky activities; someone else puts in the junior capital and takes more of the risk.

Obviously this theory is exaggerated in many respects (banks still mostly lend to people and businesses, not private credit firms), and of course in the last few weeks the main story about private credit has been about people getting mad at private credit funds for locking up their investments for the long term. (“Apollo, Blue Owl Lament Private Credit’s Liquidity Confusion,” Bloomberg News reports today.) This is not by any means a perfect theory. But I do think that it is the directionally correct way to understand the rise of private credit and its interaction with banking. Private credit keeps doing▸ AI summary more of the business▸ AI summary of banking▸ AI summary, and banks do more of the safer business of making senior loans to private credit.

This theory explains a lot. For instance, at the Wall Street Journal, Telis Demos writes that “New Bank Regulations Could Favor Loans to Private Credit▸ AI summary”:

One of the goals of some newly proposed U.S. capital rules is to help foster more bank lending. The thinking goes that tougher capital requirements for banks since the aftermath of the 2008 crisis have helped give rise to more nonbank lending, including the now trillion-dollar-plus private-credit market. …

But some of the changes could also incentivize banks to lend still more money to nonbank lenders. … That is because some bank lending to other lenders can also potentially be treated by the capital rules like a securitization. ...

The bank will lend against a special entity holding loans as collateral, and it will lend out only a portion of the underlying collateral’s value, for example. With this cushioning, along with other features, such lending can be viewed under the capital rules as the bank having a more senior “tranche” of exposure, making the treatment for this lending less risky than the underlying loans themselves would be. ...

So the upshot for a bank is that lending to a financial intermediary that makes certain loans, even at a lower yield, can be much more profitable than making those loans itself. This can be due to the capital treatment, and things like potentially lower loss rates and lower costs of dealing with one client rather than dozens.

The important point here is that this is, in some obvious sense, “correct.” If a bank makes a loan to a business and the business defaults, the bank loses money. If a bank makes a $50 loan against a $100 pool of private loans to 20 businesses, and two of them default, the bank does not lose money. That loan does have a more senior tranche of exposure, which means that it is less risky, which means that the capital treatment for it should reflect less risk. Obviously one can argue about details — if the private credit loans are riskier than the loans the bank would make, or if they are highly correlated, or if they are fraud, then this is bad — but the directional point here is that, when banks re-tranche, that makes their lending safer, and capital regulation reflects that.

Similarly, Bloomberg’s Alex Harris reports that “Apollo’s Insurance Arm Rises to Second-Biggest FHLB Borrower”:

Apollo Global Management Inc.’s insurance arm was the second-biggest borrower last year in the Federal Home Loan Bank system, a Depression-era program designed to shore up mortgage lending that has morphed into a go-to — and controversial — source of cheap financing for banks and other financial institutions.

Athene Holding Ltd. owed the FHLB $23.3 billion in loans, known as principal advances, as of Dec. 31, second only to Truist Financial Corp. and ahead of every major US bank. That’s up from 2024, when Athene was the seventh-biggest borrower in the system with a balance of $15.6 billion.

I get that it is unintuitive for Apollo, a private-equity giant, to get money from “a Depression-era program designed to shore up mortgage lending.” But in a world in which banks are increasingly re-tranching, and insurance companies like Athene are increasingly getting into mortgage lending because their liabilities match up well with mortgage assets, sure, why not?

At Bloomberg Businessweek, Brent Crane has a fun story about Theia Group Inc., a satellite startup that collapsed in 2021; its founder, Erlend Olson, is now in jail on federal fraud charges. Theia seems to have positioned itself in the spy-satellite space, and Olson’s claims were appropriately shadowy and grandiose. (“He says the company, particularly its aviation subsidiary, was packed with employees who were also active CIA agents.”) And he had a way with risk factors:

In an email to prospective investors [in 2020], Olson had listed the three biggest threats to his company’s existence. One, “Covid kills nearly everyone on earth.” Two, “The entire Theia executive team are killed somehow.” And three, “The US defaults on treasuries and loses the $6B we have in escrow.”

The ultimate threat proved more prosaic: Theia just never managed to do anything.

I feel like if you get an investment deck and the risk factors are “US default,” “we get assassinated” and “everyone on earth dies,” you send that around to all of your friends but you definitely do not invest. Those are all huge red flags.

But here is my favorite red flag in the story, identified by a city councillor in Albuquerque, where Theia supposedly planned to build a giant satellite monitoring center:

Pat Davis, a city councillor who represented the district selected for this Orion Center megaproject, was more skeptical. For starters, Theia hadn’t pushed for the usual tax credits or other financial incentives.

The challenge, in building a fraudulent business, is convincingly imitating real businesses. A real business, in deciding where to launch a multibillion-dollar construction project, would apparently haggle over tax credits. A fraudulent business might forget to do that: If the construction project is fake, you probably don’t need the tax credits; the margins on fraud are generally higher than those in heavy industry. But if you forget to haggle for the tax credits, that’s a subtle tip-off that you’re doing fraud.

Elsewhere in subtle tip-offs of fraud, Jen Wieczner has a fun story about two alleged scammers who allegedly went around throwing fake fundraisers to get cash and party photos out of wealthy people and their family offices. Apparently they pretended to be scions of famous wealthy families themselves — a Rockefeller, a Cartier — in order to ingratiate themselves with their target social set. The advantage of this, as an approach, is that family offices have a lot of money and not necessarily a lot of financial sophistication that might help them spot frauds:

Family offices control an estimated $5.5 trillion of capital, ranking them with Wall Street’s biggest banks. But in the financial world, they’re often the butt of jokes. There are notable exceptions, yet overall, family offices have a reputation for being somewhat unsophisticated — a sort of dumber money. Many of them make investments based on the whims of one very rich dude or, worse, the whims of his children who have no business experience or financial acumen of their own. … 

There are few specific qualifications or regulatory requirements for setting one up, making them both a potentially easy target for cons as well as ripe for fakers. 

The disadvantage is that they have other sorts of sophistication that might help them spot frauds:

“Honey, no one from the Cartier family would tilt her head that way in a photo; she would have gone to finishing school by the age of 14,” Soozin’s mother said. “Look at the spacing of her highlights. That hair treatment is under $200. No one from the Cartier family would have a hair treatment that cost less than $600.” 

And:

She was wearing what appeared to be a white Birkin bag, but upon closer inspection, a well-heeled attendee observed that it was fake. “What Birkin has a grommet?” said the guest.

If you’re looking to scam a certain sort of target, you don’t have to get all the financial details right, but your highlights and your fake Birkin have to be impeccable.

Last week, Elon Musk lost a securities fraud trial over claims that he misled investors about his plans to get out of buying Twitter Inc. for $54.20 per share in 2022. His lawyers have said he will appeal, but in the meantime, one of them sent this real letter to the court:

I am writing on behalf of my client Elon Musk to alert the Court to a serious issue with the verdict indicating that the jury’s solemn process to find the truth based only on the faithful application of the law to the evidence—without favor, bias, or outside influence—was corrupted in this case. The jury used its verdict to mock Mr. Musk and the process, making a numerical joke—coloring and emphasizing in bright blue the number $4.20 in its damages verdict—to send a message and signal to my client. The jury’s bizarre and highly questionable method of completing the form, this “joke” (which was no doubt intentional), was just the final example in a parade of issues and events that illustrated and confirmed that Mr. Musk was deprived his right to a fair trial adjudicated by an impartial jury dedicated to finding the truth. …

The jury revealed when completing its verdict that its decision to find liability in the first place was driven by a desire to send a message to Mr. Musk, rather than to faithfully apply the law. When writing its damages verdict, the jury wrote each deflation number in black ink, except for August 9, 2022. On that date, the jury colored in blue ink and larger font the number $4.20 to draw attention to it. … The bright blue number in a sea of black figures immediately jumps off the page, as was the apparent intent. The jury’s emphasis on the $4.20 number, which had no significance to its damages determination, but appears to be a mocking reference to a number previously associated with Mr. Musk, shows that the verdict was a mockery of justice: a commentary not on whether Mr. Musk committed securities fraud (he did not) but on the jury’s views about Mr. Musk himself.

As I have often pointed out in connection with Musk’s various M&A shenanigans, 420 is a weed joke. “Previously associated with Mr. Musk”? Has he moved on to 6 7 now?

The Well-Timed Trades▸ AI summary Made Moments Before Trump’s Policy Surprises. Wall Street Bonus Pool▸ AI summary Jumps to a Record $49.2 Billion for 2025. Ares Private Credit Fund Posts Steepest Monthly Loss▸ AI summary on Record. Monte dei Paschi revokes chief executive’s powers▸ AI summary. KKR Strikes Deal for Nothing Bundt Cakes▸ AI summary. Fibonacci▸ AI summary retracement level. The MIT Professor Tangled Up in a Tech CEO’s ‘Ponzi-Like▸ AI summary’ Scheme. ‘Five Nights at Epstein’s▸ AI summary’ Game Goes Viral at US School Campuses. Polymarket bar scene report. OpenAI puts erotic chatbot plans on hold▸ AI summary ‘indefinitely.’

If you'd like to get Money Stuff in handy email form, right in your inbox, please subscribe at this link. Or you can subscribe to Money Stuff and other great Bloomberg newsletters here. Thanks!