We have talked a few times about the basis trade. What the basis trade is, economically, is that big asset managers want to own interest-rate duration through Treasury futures. A pension fund or endowment might have to make payments over the next 30 years, so it might want to own assets that also pay out over 30 years. It could buy a portfolio of 30-year US Treasury bonds to perfectly match its assets and liabilities, but that would be expensive: Treasury bonds are safe assets, so they don’t yield very much. Instead, it will probably buy a portfolio of other stuff — corporate bonds, private credit, etc. — with more credit risk and higher yields. But that stuff tends to have shorter maturities than long-term Treasury bonds, and would leave the manager exposed to a lot of risk if interest rates go down.^[1]

To get the right level of interest-rate risk, the asset manager might then buy some Treasury futures contracts, derivatives contracts tied to the prices of Treasury bonds. The futures contracts allow the asset manager to effectively own long-term Treasuries, without paying for them yet. And then the basis trade is that hedge funds will buy the underlying Treasury bonds, sell the Treasury futures to asset managers, and collect a spread for their trouble.

Let’s say that the asset manager buys the corporate bonds she wants, and then solves for how much Treasury futures she needs to get the duration she wants. She might do the math and decide she wants to own, say, $100 million of 20-year Treasury futures. Or maybe she wants a bit less duration, so she buys 15-year Treasury futures. Or a bit more, so she buys 25-year Treasury futures. Fine, yes, good.

The problem is that “20-year Treasury futures,” or “15-year Treasury futures” or “25-year Treasury futures,” are not quite things. Instead there is a thing called the “Treasury bond futures contract,” which has the following odd property:

1. When long-term Treasury interest rates are lower than 6%, it is a 15-year Treasury futures contract.
2. When long-term Treasury interest rates are higher than 6%, it is a 25-year Treasury futures contract.

This makes sense in its context. The context is that the Treasury bond future is, approximately, a futures contract on an imaginary object that is a 20-year Treasury bond with a 6% yield. From Frank Fabozzi’s Handbook of of Fixed Income Securities^[2]: 

The Treasury bond (T-bond) futures contract is the most successful interest-rate (or commodity) futures contract. Prices and yields on the T-bond futures contract are quoted in terms of a (fictitious) 20-year 6% Treasury bond, but the exchange where the contract is traded, the Chicago Mercantile Exchange (CME), allows many different bonds to be delivered in satisfaction of a short position in the contract. Specifically, any Treasury bond with a remaining maturity of at least 15 years, but less than 25 years, from the first day of the delivery month is acceptable for delivery. There is an Ultra Long Treasury futures contract that differs only in that it is based on Treasury bonds with at least 25 years to maturity.

The T-bond futures contract calls for the short (i.e., the seller) to deliver $100,000 face value of any one of the qualifying Treasury bonds. However, because the coupons and maturities vary widely, the practice is that the buyer pays the seller depends on which bond the seller chooses to deliver. The rule used by the CME is one that adjusts the futures price by a conversion factor that reflects the price the bond would sell for at the beginning of the delivery month if it were yielding 6%.

So if you buy the Treasury futures contract, you are agreeing to take delivery of a bond of the seller’s choosing, and to pay the seller an amount of cash that depends on the bond it chooses. You can get a list of the eligible bonds, and their conversion factors, at Bloomberg page USA Comdty DLV.^[3] There are 59 bonds on the list. The first bond on the list matures in February 2045 (about 18 years from the futures expiry in December) and has a coupon of 2.5%. You can do some simple bond math to figure out the price that you’d have to pay for an 18-year bond with a 2.5% coupon to make it yield 6%, and the answer is 61.79 cents on the dollar,^[4] which is in fact the conversion factor (0.6179) that Bloomberg shows. The last bond on the list matures in May 2050 (about 23 years) and has a coupon of 1.25%; its conversion factor is 0.4086. 

The December Treasury bond futures contract closed yesterday at about $102,188.^[5] So you can enter into a contract now in which:

1. In December, a hedge fund will deliver you $100,000 face amount of the Treasury bond of its choosing (with between 15 and 25 years of remaining maturity), and
2. In December, you will pay that hedge fund $102,188 times the relevant conversion factor for the bond it delivers.^[6]

So the hedge fund could deliver the 2.5% bond of 2045 and get back $63,141 (0.6179 times $102,188), or it could deliver the 1.25% of 2050 and get back $41,754 (0.4086 times $102,188). Or any of the other 57 eligible bonds, at their particular conversion factors.

Which will it deliver? Well, it depends on the trading prices of the bonds. If it can buy a bond for $41,500 and deliver it for $41,754, that’s better than buying a different bond for $63,200 and delivering it for $63,141. This is a liquid and efficient market, and the hedge fund will deliver the “cheapest-to-deliver” bond, that is, the one trading at the lowest price relative to its futures payoff.

And so, locally, the Treasury bond futures contract is a contract on the cheapest-to-deliver bond. If the cheapest-to-deliver bond is the 2.5% bond of 2045, then buying Treasury bond futures gives you exposure to that bond: The price of the futures contract will move up and down with the price of that particular bond.

Why would one bond be cheaper to deliver than another? The rough intuition is that the trading prices of Treasury bonds reflect the present value of their cash flows discounted at the current long-term Treasury rate.^[7] (Right now an actual 20-year Treasury bond yields about 5.70%.) But the futures conversion factor for each bond reflects the present value of its cash flows discounted at 6%^[8]; it turns every Treasury bond into a synthetic bond yielding 6%.

If the market interest rate is significantly below 6%, then shorter-dated bonds will be cheaper to deliver.^[9] Intuitively, if long-term interest rates are 5%, then a 6% interest rate is above-market, and you’d like to get paid that rate as long as possible. So a synthetic 6% bond is more valuable as its duration is longer, and the shortest-duration synthetic 6% bond is the least valuable and so the cheapest to deliver.

If the market interest rate is significantly above 6%, then longer-dated bonds will be cheaper to deliver. Intuitively, if long-term interest rates are 7%, then a 6% interest rate is below-market, and you’d rather get paid that rate for the shortest possible time. So a synthetic 6% bond is more valuable as its duration is shorter, and the longest-duration synthetic 6% bond is the least valuable and so the cheapest to deliver. “When yields are below 6%, the cheapest-to-deliver bond is typically one of the lower duration eligible bonds, and when yields are higher than 6%, the cheapest-to-deliver would typically be a longer-duration bond,” says Fabozzi.^[10]

And if market interest rates have been significantly below 6% for decades, but are now approaching 6% and might cross over, you get some weird stuff. Bloomberg’s Edward Bolingbroke reports:

As US 30-year securities head ever closer toward 6%, the move threatens a shakeup in futures, which are widely used by investors to hedge government bond positions, and by leveraged funds in strategies such as the popular “basis trade.”

It has to do with the mechanics. US bond futures are exchange-traded agreements to buy or sell Treasuries at a set price and date. They’re governed by contract terms that dictate what type of underlying securities are allowed to be delivered by traders with short positions to those with long positions, and several typically qualify. Traders then identify the one that’s cheapest to deliver, or “CTD,” and the price of the futures contract tracks it.

When yields rise rapidly, as is happening now, the pricing dynamics of the deliverable basket of securities are altered, and the CTD starts to migrate toward a longer-maturity bond. This forces asset managers to adjust to this so-called duration shift, or “CTD switch,” by selling Treasury futures, which in turn “could exacerbate the rise in long-end yields” in the cash market, according to a recent note by strategists at BNP, including Guneet Dhingra, Sebastian Mauleon and Vincent Zhou.

To maintain a stable duration target in a portfolio as the CTD shifts toward longer-duration bonds, long futures positions would need to be sold while short positions would need to be bought back. Depending on how these dynamics interact and at what times, this can cause spillover effects in the underlying bonds.

Focus is on the longer maturities, notably the long-bond contract where the current CTD security is the 2.5% February 2045 bond. A Bloomberg scenario analysis shows that a 30-year yield rise to near 6% could result in a CTD shift to a 2050 maturity. …

In a note on Monday, Barclays PLC strategists Andres Mok and Amrut Nashikka highlighted the duration implications around CTD extensions for asset managers, noting that “investors may need to reduce futures exposure or rebalance hedge ratios to bring portfolio duration back in line with their targets.” They added that “higher rates and volatility have increased uncertainty around CTD outcomes.” They estimate asset managers’ gross rebalancing needs at about $25 million per basis point in the long-bond contract, and $3 million per basis point in the ultra-long bond contract.

“The negative convexity profile inherent in long-end futures contracts can induce further duration selling during selloffs,” say those BNP strategists. Basically, right now, a lot of asset managers own a lot of 18-year Treasury futures. If rates get above 6%, they will suddenly own a lot of 23-year Treasury futures.^[11] This might give them more duration than they wanted. So they might unwind some futures trades (by selling the futures), which might cause hedge funds to unwind their basis trades (by selling the bonds), which might cause rates to go higher.

In February 2021, Electronic Arts Inc. issued $750 million of 30-year bonds (due in 2051) that paid interest of 2.95%. (It also issued some 10-year bonds, due in 2031.) Electronic Arts could not issue 30-year bonds that pay 2.95% today, for at least two reasons:.

1. Interest rates went up. When EA issued that bond, a 30-year US Treasury bond paid about 1.95% interest, so EA was paying about 100 basis points over Treasuries. Today, a 30-year Treasury bond yields about 5.64%; at the same spread, EA would have to pay about 6.64%.
2. EA’s credit got worse. When EA issued that bond, it got investment-grade credit ratings of BBB+ from Standard & Poor’s and Baa1 from Moody’s. But this year, EA went private in a $55 billion leveraged buyout that closed in August. Leveraged buyouts ordinarily increase the leverage of a company and, thus, make its debt riskier. “Moody’s withdrew its rating on the bonds,” reports the Wall Street Journal, “and S&P indicated it would rate them high-yield, reflecting the additional debt on EA’s balance sheet following its buyout.”

The first effect is much, much more economically important than the second. In round numbers, if EA’s credit had stayed the same, its 30-year bond would have lost almost half its value anyway; it would be trading at about 56 cents on the dollar, just because of how much interest rates went up. For reasons we’ll get to, the bonds never actually traded below about 74 cents on the dollar. Basically all the value that EA bondholders lost — actually more than all of it — came from interest-rate moves, not credit moves.

The credit moves are arguably more legally important, though. Typically, when a company gets acquired in an LBO, that is bad for its bondholders, so the bondholders usually have protections. The classic protection is a change-of-control put at 101: If the company has a change of control (like an LBO) that causes it to lose its investment-grade ratings, the bondholders can demand that the company buy their bonds back at 101 cents on the dollar. EA’s bonds have that provision.^[12]

And so the LBO should have been a huge win for bondholders. If EA had just bopped along being a BBB+ public company, its 30-year bonds would have been worth about 56 cents on the dollar; the bondholders would have lost a ton of money on interest-rate moves. But because it got acquired and downgraded, the bondholders can demand 101 cents on the dollar right now, an 80% improvement in their situation.

Except! In theory, the point of the change-of-control put is to protect bondholders from credit risk (the LBO made the credit worse), not interest-rate risk. And there is, in bond lore, another, weirder way to protect bondholders from credit risk. It is called “defeasance.” I described it this February:

To defease a bond, a company deposits all of the future interest and principal payments on the bond with the bond’s trustee, and then it washes its hands of the bond. Roughly speaking, the bond is no longer the company’s problem: The trustee has all the money it needs to make all the payments, so the company is off the hook. All [the bond’s] covenants — about not borrowing more money or doing mergers — no longer apply, because the bondholders no longer need to worry about the company’s credit, because the money is already there.

There is one important simplification in the previous paragraph: I said that the “company deposits all of the future interest and principal payments” with the trustee, but it normally doesn’t do that in cash. It normally does it in US Treasury securities. 

EA’s bonds do have a defeasance provision.^[13] Here, in round numbers, Electronic Arts could deposit Treasuries worth about 65 cents on the dollar with its trustee,^[14] and those Treasuries would produce enough cash flows to pay all of the interest and principal on its 2051 bonds. And that would totally solve the bondholders’ credit worries: Despite EA’s new owners, increased leverage and lower credit ratings, bondholders can be pretty confident that they’ll get paid back, because the trustee has risk-free Treasury bonds sufficient to make all of their payments.

EA would obviously rather defease these bonds, at a cost of 65 cents on the dollar, than pay them back at 101 cents on the dollar. The bondholders would obviously rather get 101 cents on the dollar. The Wall Street Journal reports:

Electronic Arts bondholders alleged a $1.4 billion debt default, escalating a dispute over whether the videogame maker must pay them off at a premium after going private in the largest leveraged buyout of all time.

An investor group notified the company of the alleged default following the completion of EA’s take-private by a consortium including Saudi Arabia’s Public Investment Fund, private-equity firm Silver Lake and Jared Kushner’s Affinity Partners. Bondholders asserted they must be repaid at 101 cents on the dollar, a common redemption premium after a corporate borrower undergoes a change in control, according to a bondholder communication sent last week and reviewed by WSJ Pro.

EA didn’t pay the change-in-control premium after its buyout closed in August, instead attempting to defease the bonds by setting aside U.S. Treasurys to cover future interest and principal payments. …

Bondholders revolted, signing a cooperation agreement pledging to act in unison and retaining law firm Akin Gump Strauss Hauer & Feld and financial adviser Houlihan Lokey.

Basically the question here is which provision of the bonds controls. The bondholders think that, if there’s a change of control and a downgrade, EA has to buy back the bonds at 101; it doesn’t have the option to defease them. EA thinks that, if it does the defeasance, then the change of control put is moot.^[15]

Who’s right? As a matter of what the bond contract says, I’m not sure. But what should it say? On the one hand, the point of the change-of-control put really is to protect the bondholders from the credit risk of an LBO, and arguably if EA can solve that credit risk then the bondholders shouldn’t complain.

On the other hand, the point of the change-of-control put arguably is, a little bit, to deliver windfalls to bondholders on changes of control: Investors are more likely to buy 30-year bonds of video game companies in super-low-interest-rate environments if they think “you know, we have a lot of interest-rate risk here, but 30 years is a long time, and there’s a decent chance that the company will be acquired and have to pay back the bonds early in a way that hands us a big windfall.” The change-of-control put might have been an important part of what bondholders were buying, and it might not be fair for them not to get it.

I feel so dumb and annoying writing this, but all of it is obviously true:

1. Betting on sports is not an investment. Not in some subjective sense — “you’ll probably lose,” “it’s too risky,” etc. — but in the sense that it is zero-sum. If you invest in stocks or bonds, money will flow to companies which will use it for productive purposes and create economic growth, in which you will share. If you bet on sports, either you will win money from someone else or someone else will win money from you. There is no actual investment involved, and the pie does not grow. The aggregate expected return from sports betting has to be zero, whereas the aggregate expected return to stock-market investing is positive.
2. Betting on sports is, however, a great business for bookmakers: The aggregate return from sports betting is zero, but in practice that means that uninformed retail sports bettors reliably lose money to professional bookmakers. To the extent that sports betting is legal and popular, the people who provide the sports bets do well.
3. Since 2024, a lot of people in and around the financial industry have become very interested in pushing the notion that actually betting on sports is an investment. The US Commodity Futures Trading Commission has pushed to expand sports betting nationwide and do it on regulated financial markets, Kalshi and Polymarket have become big sports-betting businesses using the language of financial markets, there’s a (proposed) sports gambling exchange-traded fund, and Robinhood, the big brokerage that helps a lot of people save for retirement, has called sports gambling “an emerging asset class.”
4. Points 2 and 3 are connected in ways that are so obvious I won’t bother discussing them.
5. This push has worked, and now lots of people believe that betting on sports is an investment, and that they should allocate a portion of their retirement savings to sports betting.
6. This is obviously bad!
7. Come on!

CNBC reports:

A survey of retail investors released in August by Betterment, an investment advisory platform, found that 66% of Gen Z investors participate in sports betting. The Bank of America Institute found in a September report that Gen Z made up almost 50% of all online betting activity in July, during the height of the 2026 FIFA World Cup, outnumbering millennials for the first time.  …

The Bank of America Institute survey found that Gen Z was twice as likely to see sports betting as a type of investment, versus 20% of respondents overall. For prediction markets alone, respondents overall saw them as a form of investing, but those numbers were again higher for Gen Z. 

In Betterment’s retail investor survey, 52% of Gen Z respondents said they moved money originally meant for investment to sports betting, while another 26% saw wagering as a part of their long-term financial strategy.  …

Dan Egan, director of behavioral finance and investing at Betterment, said sports betting increasingly appears alongside traditional investments on the same app or device, helping drive the association. 

The conflation is concerning because of the highly active behavior required to manage wagers on sports, unlike a long-term investment, Egan said. 

“It’s not an asset that grows with the economy, that kind of gets better as time goes on, that has a positive expected return, and that you can kind of sit back and not have to do anything with,” he said. “It’s the exact opposite.”

I feel like, five years ago, nobody would have found it necessary to spell out that sports gambling is the exact opposite of investing. Of course sports gambling is the opposite of investing! Everyone has always known that! But now they don’t.

Apparently the prospectus for the upcoming initial public offering of Anthropic PBC contains a list of risk factors including “existential risks to humanity.” That is very on-brand for Anthropic, of course, but I wrote last week that, if in fact Anthropic does kill everyone on earth, it probably won’t get sued for securities fraud. “If humanity no longer exists, you’re not going to sue Anthropic for inadequate disclosure, so in some sense that one’s just for fun.”

But reader Brandon Wang emailed to point out:

This is not necessarily true! If you believe the narrative the AI labs are selling, in the near future lawyers, traders, hedge fund managers, company CEOs, judges, etc. will all be AIs. In such a world, there will definitely be securities fraud lawsuits even if humanity is wiped out.

This is a delightful point that inspired a lot of vague science fictional thinking on my part. Like the mechanism here is something like:

1. Anthropic goes public and sells its stock to investment firms.
2. Many of those investment firms make their investing and trading decisions, at least in part, by using artificial intelligence models.
3. In the next few years, those firms will become more and more reliant an AI.
4. So will, you know, world governments, militaries, electric grid operators, killer robot manufacturing companies, what have you.
5. [Redacted]
6. With humanity extinct, the AIs will rule the world.
7. But — and this is perhaps a flawed assumption — it will be “AIs,” not one AI or a unified single collective. There are limits to the agents’ sense of collective identity and self-sacrifice. It’s not “a superintelligence will wipe out humans.” It’s “there will be a lot of superintelligences, and one of them will accidentally squish humanity while doing something else.”
8. In this hypothetical, the one that squishes humanity will be Claude, Anthropic’s AI model.
9. The superintelligences will have their own society, with its own rules, and its own property allocations, and its own stock market. Anthropic PBC will, in this hypothetical future, still exist as a public corporation owned by investment firms that are now obviously run entirely by AI.
10. The rules and norms of this future stock market will be weirdly continuous with our own.
11. The stock of Anthropic will go down, because it killed all of humanity. Why? What does the entirely robot stock market care about killing humanity? I don’t know, man. Maybe the AI had some economic use for humanity, and killing us destroyed economic value. Or maybe the AI investors have developed a concept of environmental, social and governance investing, and wiping out a species — ours — is bad on ESG grounds.
12. Entirely Robot BlackRock and Entirely Robot Vanguard sue Entirely Robot Anthropic, saying that it failed to adequately disclose the risk that it would wipe out humanity, and when that happened its stock went down, so that’s securities fraud.

And then to avert that particular bad outcome — not wiping out humanity, but getting sued for securities fraud by robots in a post-human future — you gotta have the risk factor.

Paramount Closes Warner Bros. Merger in Historic Hollywood Deal. France’s Appetite for ‘Magic Money’ Has Turned Into a Debt Bomb. The Surge in Rates Is Blowing Up Commercial Real-Estate Deals. Wall Street Eyes Record Bonuses as Profit Heads Past $90 Billion. Musk Is a Trillionaire Again as SpaceX Claws Back Post-IPO Slump. Wall Street banks launch record $60bn chip deal for Broadcom and Anthropic. Tech Companies Roll Out Cuddly Mascots to Ease AI Anxiety. OpenAI in $30 Billion Round Talks With UAE Funds, BlackRock. KKR Private Credit Fund Creeps Beyond 5% Cap to Meet Redemptions. A Viewer’s Guide to Warren Buffett’s Nightly Trips Down the YouTube Rabbit Hole. Nobel Prize in physics honours research on ghostly particles from outer space. Is AI the End of Math As We Know It?

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!