Money Stuff: Index Traders Had a Good Month

On Wednesday, 109.3 million shares of SpaceX traded on the stock market; of those, about 5.4 million were traded in the Nasdaq closing auction at 4 p.m. that set the closing price for the day ($157.54). [1] On Thursday, before the holiday, 61.3 million shares traded, 2.6 million of them in the closing auction; the stock closed at $162. Yesterday, 188.8 million shares traded, 74.9 million of them in the closing auction. The closing price was $160.42, down about 1% from Thursday; the broad market was up. SpaceX was down some more this morning.

Why was yesterday’s closing auction an order of magnitude bigger than it usually is? This is not a great mystery. The Wall Street Journal reported yesterday:

SpaceX will officially join the Nasdaq-100 on Tuesday, and investors holding some funds tied to the index will end up exposed whether they like it or not.

Mutual and exchange-traded funds with a collective $800 billion in assets under management that track Nasdaq’s flagship tech index, including the popular Invesco QQQ ETF, are set to buy SpaceX shares at Monday’s closing price in order to mirror the index’s performance.

That comes after Elon Musk’s artificial-intelligence and-rocket-making company was fast-tracked into the Nasdaq-100 under new rules that aim to include newly public megacap companies sooner. 

Index funds that want to mirror the performance of the Nasdaq 100 essentially had to buy SpaceX at yesterday’s closing price, since as of today the index includes SpaceX’s performance. “Bloomberg Intelligence analyst Rob Du Boff estimated that SpaceX’s inclusion in the Nasdaq 100 and FTSE Russell gauges would drive at least $5.4 billion in buying from index-tracking funds,” most of that from the Nasdaq 100. If you figure there was about $5 billion of index fund demand yesterday, that explains about 31.2 million of the 74.9 million shares that were purchased in the closing auction.

What about the rest? Well, it was a big day for SpaceX generally; the banks who ran its initial public offering initiated (bullish) research coverage yesterday. (That happened after the close, but was anticipated; maybe people were buying in anticipation of it.) Also, though, here is a 2024 paper by Alex Chinco and Marco Sammon titled “The Passive Ownership Share Is Double What You Think It Is,” which pretty much says that [2] :

While index funds held 16% of the US stock market in 2021, we put the overall passive ownership share at 33.5%. Our headline number is twice as large because it reflects index funds as well as other kinds of passive investors, such as institutional investors with internally managed index portfolios and active managers who are closet indexing.

If you are an investor who benchmarks yourself against the Nasdaq 100, then up until yesterday SpaceX wasn’t your problem. But as of the close yesterday, SpaceX is in the Nasdaq 100, and the default expectation for a Nasdaq-focused investor is to own some of it. 

So there was a ton of one-time demand to buy SpaceX yesterday. Was there a ton of one-time supply? Were there big SpaceX holders looking to sell? The big pre-IPO shareholders are all subject to lockup agreements, which start expiring later this month; none of them can sell now. There wasn’t bad fundamental news, or a SpaceX primary stock offering, to add to supply.

So a ton of unusual demand, all concentrated at the closing auction yesterday, with no unusual supply. Doesn’t that sound like the stock should have gone way up? In general, if you try to buy 74.9 million shares, all at once, of a stock that doesn’t usually trade that much, you have to pay much more than the price of the last trade. Note that this would be pretty bad news for the index funds: They’d buy the stock at a very high price caused by a temporary demand imbalance, and then immediately lose money as the stock returned to normal.

In fact SpaceX did trade up a bit into the closing auction yesterday: It was at $155.50 at 3:30 p.m., before closing at $160.42 at 4 p.m. On the other hand, that closing price was lower than where it started the day, or where it was trading at 1 p.m. The stock did not spike that much above its “normal” price to accommodate all that demand, though it did trade down today.

Because, of course, this problem is quite well-known and the market has more or less solved it. Here is one conceptual solution that you might think of:

1. Index funds and quasi-indexers know by late June that they will collectively have to buy, say, 65 million SpaceX shares [3] at exactly the closing price on July 6.
2. Everyone else knows that too.
3. Each fund could go to an investment bank in late June and say “hey, we will need 10 million SpaceX shares at 4 p.m. on July 6, can you sell them to us?”
4. The bank is in this business — block trading, liquidity provision, risk management, etc. — and will quote the fund a price. “Well, the stock is at $153,” the bank might say, “but you’re looking to buy a lot, and we know that a lot of other funds will also be looking to buy a lot, and this is a lot of risk. We can’t lock in a $153 price for you. But, tell you what, we could do $160. That leaves us some cushion.”
5. “Done,” says the fund. 
6. The bank now owes the fund 10 million shares of SpaceX at $160 per share. The bank is effectively short 10 million shares of SpaceX at $160. [4]
7. The bank then spends the next week or so buying SpaceX stock for its own account. It is short 10 million shares, and wants to hedge that by getting long 10 million shares. So it goes and buys SpaceX stock. Ideally it does this at prices below $160. If it buys 10 million shares before July 6 at an average price of $158 per share, it will make a $2 per share profit. If it buys at $154, it will make $6 per share. If it buys at $162, it will lose $2 per share. If it buys 6 million shares before July 6, delivers the agreed 10 million shares on July 6, and buys the remaining 4 million shares after July 6, that’s its business. [5]  The index fund has locked in a price; how the bank delivers that price — and how it manages its risk to get the shares — is its problem.

That’s a good first cut at a solution, and a lot of markets work more or less like that. It does have one flaw, though, which is that in Step 4 the index fund has not quite locked in the right price. The index fund doesn’t want to buy SpaceX at some fixed price ($160) on July 6. It wants to buy SpaceX at the closing price on July 6, whatever it turns out to be: That’s how it tracks exactly the index performance. It just doesn’t want that closing price to be abnormally high.

So we can modify the trade. In Step 4, instead of saying “We’ll lock in $160 for you,” the bank says: “Sure, we’ll sell to you at whatever the closing price is on July 6.” But then every other part of the trade is the same. The bank’s job is now not to buy 10 million shares below $160, but rather to buy 10 million shares below the closing price on July 6, which it doesn’t yet know. 

This is a messier problem but, you know, banks are smart, they can figure it out. Roughly speaking the job is to guess (1) what the closing price will be on July 6 and (2) when the stock will trade below that price. You might think something like: “Well, I’m going to be buying up until July 6 to hedge my obligation to deliver stock to this index fund. A lot of other banks are going to be doing the same trade for other index funds. Therefore, the stock will creep up until July 6, peak then, and fall afterwards. So I’ll buy like 6 million shares fairly quickly, stay out of the market on July 2 and July 6 when the stock is peaking, deliver my shares at the closing price on July 6, be short a bit at that point, and then buy the remaining 4 million shares on July 7 when the pressure is off and the stock drops.”

Of course this needs some refinement. For one thing, the stock might move for reasons other than the index fund demand; you might have fundamental reasons to buy early or late or hedge or do something else. For another thing, lots of people are doing similar trades and making similar decisions, so your reasoning about them has to go a few levels deeper: If everyone thinks “I’ll buy early and late to avoid the rush to buy right on July 6,” then the rush to buy will be on June 30 and July 7, not July 6, and you should actually buy on July 6, etc. 

The point is that this is the bank’s problem, or rather the banks’ problem, if a bunch of them are selling this service to a bunch of index funds. The index funds just pay the closing price on July 6; how the banks deliver that price — and how they manage their risk to get the shares — is up to them.

One possible approach, for a bank, would be to do nothing. You could just wait until the closing auction on July 6 and then put in a bid to buy 10 million shares. You’d get filled at whatever the closing auction price was, and then deliver the shares to the index fund at that price. No risk, no effort, easy. If every bank did this, then the stock price really would spike in that closing auction: There’d be a ton of demand (from the banks, for the index funds) and no additional supply.

But most banks probably wouldn’t do this, for two reasons. First of all, it is bad customer service: If your customer, the index fund, has agreed to buy from you at the closing price, and then the closing price spikes from $150 to $200 in the last half hour of trading, the customer is going to be annoyed. The customer will blame you for pushing up the price going into the close, and might stop using your services.

Also, though, you can make more money doing the other thing! If you’ve locked in a sale at the closing price on July 6, your job is to beat that price for your own account. If you just buy in the closing auction, you will make $0 per share: no risk, but no reward. If you cleverly buy before and after the closing auction, you might make $2 or $5 or $10 per share, which is tens of millions of dollars. You might lose money, too, but you’re in this seat because you’re trying to make money.

The banks’ incentives here are largely benign: Their job is to smooth out demand. Instead of sudden demand for 65 million shares all at once at 4 p.m. on July 6, the banks buy some shares before July 6, and some more shares after, so that they can deliver the whole 65 million shares to the index funds at the relevant time. The banks take the risk that they might pay more than the July 6 closing price to source all those shares, but if they pay less then they keep all the savings. Their incentive is to buy tactically at low prices, so they don’t push the price up too much at the close.

This structure — customer puts in buy-on-close order with a bank, bank “pre-hedges” the order and thus smoothes out demand — is again fairly common in other markets. We have discussed the “trade at settlement” mechanism in the oil market, and pre-hedging the fix in foreign exchange markets, which are variations on this approach.

The actual structure in the stock index addition trade is almost like that, with a couple of tweaks [6] :

1. Banks used to do this business, but now it is largely done by big multimanager multistrategy “pod shop” hedge funds, which do the risky markets businesses that banks used to do.
2. The hedge funds don’t have customers or take orders. It’s not like each hedge fund has an order book from index funds for 5 or 10 or 20 million shares, and they each plot how to get their shares to deliver to their customers. Instead, they guess. Each hedge fund estimates how many shares indexers will demand on July 6; each hedge fund also estimates how many shares other hedge funds will supply on July 6. Then each fund decides how many shares it will want to supply — roughly, the answer is “demand from index funds minus supply from other hedge funds,” though of course that is circular — and then figures out when and how to buy all those shares at the lowest possible prices, as well as how to manage the risk of market moves. (For that matter, they also guess what stocks will be added and deleted from indexes: This trade is better, in general, if you buy the stock of a company that will be added to the index before everyone else knows that it will be added, though that was not especially relevant in the SpaceX case.) And then each hedge fund offers shares into the closing auction on July 6, hopefully roughly matching the number of shares that the index funds (and quasi-indexers) bid for. 

If it all works, supply matches demand in the closing auction, and the price doesn’t move too much: The hedge funds have effectively spread out the index funds’ buying over the days before (and after!) the index-add date, averaging them into the index over time. The index funds get to buy SpaceX at more or less a “normal” price, rather than a price that has spiked up momentarily on huge one-time demand. Here is a 2024 paper by Robin Greenwood and Marco Sammon on “The Disappearing Index Effect,” finding that the stock-price spike caused by index additions has declined over time, because “the stock market has simply become more efficient in the context of providing liquidity to S&P 500 index additions and deletions.” Despite not having customers, the hedge funds provide good customer service.

There are various ways for it to work out badly, with the main ones being:

1. The demand is not fully smoothed out: The price spikes a lot in the closing auction and falls afterwards, leaving the index funds with an immediate loss. 
2. Some hedge funds get their estimates wrong and lose a ton of money.
3. Some hedge funds make a ton of money. Good for them! In some sense, how much money the hedge funds make is their business: The index funds just want to buy shares from the hedge funds at the closing price, and don’t care when or how the hedge funds buy their shares. In another sense, though, the hedge funds’ profits come from the gap between the price at which they buy the stock and the price at which they sell it to the index funds, and the bigger that gap is the worse off the index funds are.

Bloomberg’s Nishant Kumar reported yesterday:

Two teams focused on trading index changes at Millennium Management made billions of dollars between them in June, lifting the multistrategy hedge fund’s monthly gains.

Pods run by Glen Scheinberg and Pratik Madhvani made about $3.7 billion in total last month, according to people with knowledge of the matter. The gains accounted for more than half of the about $6.6 billion profit generated by Millennium before fees in June, the people said, asking not to be identified because the details are private. …

Scheinberg and Madhvani specialize in index rebalancing trades, which involves highly leveraged bets on which securities will enter or exit various indexes. This strategy, along with bond basis trading, are core to many multistrategy hedge funds looking to deploy large chunks of capital and earn steady overall returns.

Five events in June created a particularly rich environment for index rebalancing strategies: the S&P 500 quarterly rebalancing, Nasdaq 100’s quarterly changes, Russell’s annual reconstitution, the fast-track inclusion of SpaceX and quarter-end multi-asset rebalancing.

Good for them! [7]  At a very high level, in aggregate, in expectation, this trade is “the hedge funds buy stock for the index funds over time and get paid a reasonable fee for liquidity provision.” But, given all the guessing, there’s a ton of variance: Sometimes the fee is negative hundreds of millions of dollars; other times it is positive billions of dollars.

The deal is that, if you borrow as much money as you can to buy a thing as it goes up, you will push up its price. The thing you own is now more valuable, which means that you can borrow more money against it. If you borrow more money to buy more of the thing, you will push up the price some more, in a continuing cycle. Eventually you will have borrowed a lot of money and own a lot of the thing and feel pretty good about yourself. You will have billions and billions of dollars of gains.

And then what? I suppose one answer is: “You stop, the thing continues to go up on its own for fundamental reasons, you occasionally sell some over time to pay down borrowings, and after a few years you own billions of dollars’ worth of the thing free and clear.” 

The answer that I write about more often is: “You reach the limits of your borrowing capacity, a slight breeze knocks down the price of the thing, you get margin calls, and it all collapses in 20 minutes.” That was famously the outcome when Bill Hwang of Archegos Capital Management did this trade with some stocks. 

My model of Strategy Inc., the original digital asset treasury company, is that it is “soft fuzzy banking,” or I guess “soft fuzzy use of leverage to buy Bitcoin as it goes up.” Strategy borrowed a lot of money to buy Bitcoin on the way up, but its borrowing mostly came in the fairly benign form of perpetual preferred stock. As long as Bitcoin rose more than the (quite high!) cost of the preferred, everything was fine. Now Strategy seems to have reached the limits of how much it can borrow, and Bitcoin has gone down, so the cycle has reversed. Bloomberg’s Vildana Hajric and David Pan report:

Michael Saylor’s Strategy Inc. sold $216 million of Bitcoin last week, marking the first major step in the financing overhaul the company unveiled in recent days after a prolonged slump in the cryptocurrency and its own shares.

The transaction is the company’s largest Bitcoin sale since it began building its holdings in 2020 and only its third overall. Strategy shares fell about 2% in premarket trading Monday, while Bitcoin traded 1.3% lower at around $61,800, well below the firm’s average purchase price of about $75,000 per token.

For years, Strategy’s business model rested on a simple premise: raise capital, buy Bitcoin and don’t sell it. The latest transaction marks the clearest sign yet that the company is moving toward a more flexible approach in which the token becomes another source of liquidity alongside equity and debt markets. …

Strategy’s latest Bitcoin sale also carries implications beyond one company. The company had become one of Bitcoin’s largest and most consistent corporate buyers, helping underpin institutional demand during the bull market. 

If Strategy had taken out margin debt to buy a huge chunk of the world’s Bitcoin, and then the price of Bitcoin fell to well below its average purchase price, the result would have been obvious and quick: margin calls, forced liquidation, further price drops, bad stuff. But Strategy sold weird preferred stocks to buy a huge chunk of the world’s Bitcoin, so the reversal of the cycle is a bit calmer and slower, “a more flexible approach” rather than dumping all the Bitcoin immediately. Still, Strategy is selling Bitcoin at $61,800 after buying it at $75,000, because the cycle did reverse.

Meanwhile, Bloomberg’s Reshmi Basu reports:

Distressed-debt funds that snapped up Strategy Inc.’s beaten-down preferred shares are in talks with one of the company’s bankers about swapping them for other securities, pitching it as a win-win as the relentless rout in crypto puts the Bitcoin accumulator in an increasingly tough spot.

The investors have had conversations with Moelis & Co. about exchanging their holdings for other preferred shares at a discounted price, or potentially for common shares, which have plunged by roughly 75% in the past year, according to people familiar with the matter.

Again, not quite a margin call, but sort of a more flexible version of a margin call.

Elsewhere: “Strategy’s Bitcoin Treasury Model: Corporate Omphaloskepsis, Polypharmacy of Risk, and Shareholder and Societal Welfare,” ah.

One early version of “effective altruism” was: “You should go work at the highest-paying job you can find, which will be at Jane Street, live modestly, and donate your earnings to save lives by buying mosquito nets in developing countries.” This idea — often called “earn to give” — was actually quite controversial, though it always seemed pretty reasonable to me.

It was, however, pretty quickly superseded by another form of effective altruism, which was roughly: “You should go work at Jane Street and donate your earnings to think tanks that are trying to stop runaway artificial intelligence from destroying humanity, because that is where you will get the most bang for your charitable buck.” This version was much more controversial, basically because it is relatively easy to measure the lifesaving impact of mosquito nets and relatively hard to measure the lifesaving impact of, YOU KNOW, buying a castle to give philosophers a congenial place to think about how to stop runaway AI. If runaway AI really was going to destroy humanity, and the castle really allows the philosophers to come up with a way to stop it, then that really does save more lives than all the mosquito nets. But. You know.

A slight tweak to that form of effective altruism was “You should quit your job at Jane Street to start a high-risk crypto empire to have even more money to use to stop runaway AI.” If the basic theory is “you should make as much money as possible to stop runaway AI,” then working at Jane Street is fine, but for a while the real money was in crypto, and stopping runaway AI would take real money.

That was the state of the art a few years ago. Now, of course, the way to make real money is by working at a frontier AI lab. This suggests that the optimal current form of effective altruism is:

1. Work at a frontier AI lab that is sprinting to build artificial superintelligence as quickly as possible.
2. Get paid like $100 million a year.
3. Live modestly and donate all of your earnings to a charity that is trying to stop frontier AI labs from building artificial superintelligence that might wipe out humanity.

I can see no flaws in this approach. Andrew Fedorov reports on “The Effective-Altruism Comeback”:

Matt Lerner, the managing director of research for Founders Pledge, which advises entrepreneurs on their charitable giving, [said] it is clear the “influx of capital from the AI sector is going to significantly change the philanthropic ecosystem.”  …

The new AI money has given EA a chance to return and come back larger than ever before.

To be fair, some of the zillions of dollars of AI-to-EA money will probably be used for Effective Altruism 1.0 projects (“ending factory farming, stopping nuclear war, halting climate change”), not just stopping AI. But mostly stopping AI:

The optimism about the good this money could bring is also closely accompanied by ambient panic about the technology’s wider consequences. Ironically, the thing that many effective altruists want to do with their AI wealth to do the most good is fund organizations working to make sure AI is developed responsibly. …

Amid the swirl of goals and causes, the one widely believed to be set to receive the most money was AI safety. PauseAI, which advocates for a moratorium on development, is the only major AI safety organization that doesn’t accept funding from the industry as a matter of policy, but even they haven’t ruled out donations from staffers of the large AI labs. “That would have to be the board’s decision, not just mine,” said Holly Elmore, founder and executive director of PauseAI US. “They might rule that it’s against the interest of the org for me to say, ‘No, I don’t want to be part of this person’s blood money.’”

Organizations whose coffers are willing to welcome support, surprisingly, include the Machine Intelligence Research Institute, which was founded by [Eliezer] Yudkowsky. “MIRI would like to applaud and encourage people taking marginal right actions, even within larger systems that are moving in the wrong direction,” said Malo Bourgon, MIRI’s CEO. “A polluting industry that pays for carbon offsets is better than a polluting industry that does not, for instance, even if it would be better overall for the industry to stop polluting entirely.”

Right, I mean, if you are an effective altruism organization, it would be weird to refuse to take anyone’s money on principle.

HSBC pulls back from riskier private credit lending. Amazon Seeks to Raise at Least $25 Billion From Bond Sale. Michaels Goes From ‘ Grandmother’s Store’ to Unlikely Success Under Apollo. SK Hynix’s Planned $28 Billion Offering Brings Korean Market Fever to the U.S. AI Legal Startup Norm Valued at $1.2 Billion in Funding Round. Davos Founder Klaus Schwab Plots a Return to World Economic Forum. Canada tells UAE it is not ready for its C$70bn investment. Don’t heat a NeeDoh.

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!