Money Stuff: Untruth Machines

The order of magnitude is that artists get a few dollars for every 1,000 times their songs are streamed on Spotify, but bots can generate fake Spotify streams at a cost of something like $1 per 1,000 streams. [1] If you can generate 1,000 streams for $1, and those 1,000 streams pay you $2, then you’re in business, sort of. We have discussed the economics of that business. It’s not a business that Spotify likes, or US federal prosecutors. We discussed it because a guy who allegedly did it was arrested for fraud.

But the key point here is that bots can, and sometimes do, generate fake Spotify streams for something like $1 per 1,000 streams. And then the question is: If you can generate fake Spotify streams at a cost of around $1 per 1,000, and you don’t mind doing a little crime, what is the best way to monetize that? Getting paid $2 per 1,000 streams as artist royalties is not terrible. (It’s like a 100% return on investment?) I’m sure that there are more complicated big-picture sorts of answers. (“Generate lots of fake streams and get a record deal,” etc.)

But now there are better, simpler answers. Bloomberg’s Denitsa Tsekova reports:

Spotify spotted and removed over 500,000 artificial streams that had unexpectedly turned Malcolm Todd’s song “Earrings” into one of the most popular songs on its charts, according to a person familiar with the matter. The artificial numbers had already been used to settle a Kalshi market tied to the most frequently streamed Spotify song in the US in June, which had attracted $3 million in trading.

Todd had been declared one of the winners based on figures published before Spotify completed its investigation. Spotify reached out to the companies after finding the problem, the person said.

“We’re in touch with Spotify and are actively investigating this matter,” said Elisabeth Diana, a spokesperson for Kalshi.

Don’t take my numbers too literally — there is not, like, a Bloomberg chart of the price of illegal Spotify bot streams — but, as far as I can tell, generating 500,000 artificial streams would cost on the order of hundreds of dollars. Kalshi’s contracts on Todd’s song traded well below $0.10 for most of June, and then paid off $1 when it unexpectedly finished first. (After Spotify’s investigation and removal of fake streams it fell back to fourth, but the Kalshi contract had already settled.) Open interest, by the end of the month, was about 76,000 contracts, meaning that you could have made tens of thousands of dollars on Kalshi by spending hundreds of dollars to manipulate Spotify, on the order of a 10,000% return on investment.

Did someone? Is there a connection between the Kalshi market and the fake Spotify streams? I dunno, but everyone seems to think so. Here are stories about the controversy from Wired and the Financial Times, and here is a Substack post from prediction markets trader Caleb Davies explaining the likely manipulation. 

I wrote last week about egg market manipulation:

There’s a big market (egg producers selling eggs to supermarkets etc.), and there’s a small market (egg producers selling extra eggs to each other on an electronic exchange). The price in the small market determines the price in the big market. Participants in the small market are also participants in the big market. You can spend a little money in the small market to move the price, which can make you a lot of money in the big market.

As I also wrote last week, “‘Big market’ and ‘small market’ there are imprecise terms.” Naively, at least a few years ago, I would have thought that the music industry is pretty big and prediction markets are pretty small. But it turns out streaming markets are pretty cheap to manipulate, and now the payoff in prediction markets is big enough to make it worthwhile. And I wrote in April:

One, uh, accomplishment of modern prediction markets is that they can, in theory, allow for an arbitrarily large amount of betting on arbitrarily small quantities of reality. ...

And so one meta-strategy for trading on prediction markets is to try to find the largest gaps between (1) the amount of money to be made and (2) the amount of underlying reality. And then, you know. Make a big bet, and cause a small change to reality so your bet pays off.

Regular financial markets sometimes create incentives to manipulate other financial markets. Any time the stock market goes up, you could have some theory about quant funds conspiring to inflate it or whatever, and of course people sometimes really do make trades in one financial market to influence prices in another. And remember what Keynes said. You can always question whether some financial-market price “reflects reality” in a meaningful way. 

What prediction markets have accomplished is opening up all of reality, not just financial markets, to that sort of doubt. You can make money on Kalshi, now, not just by manipulating stock or Bitcoin or whatever markets, but by manipulating Spotify streams or temperature gauges or the words people say at press conferences or the location of a gorilla. The upshot is that now you can’t be sure whose song was the most popular last month, or what the temperature is in Paris, or why a gorilla is wandering around your neighborhood.

Prediction markets are “truth machines,” in the sense that they give people incentives to figure out what is true and bet on it. But they are also, obviously and increasingly, falsehood machines, in the sense that they give people incentives to make bets and then distort reality to make those bets pay off. Because of prediction markets, it is now lucrative to figure out what song will be the most popular, but it is even more lucrative to make a different song look most popular. Prediction markets can make it harder to know what is true.

I mean. A memecoin is:

1. Some person or thing (the “meme”) is in the news.
2. Someone (the “promoter”) who may or may not be associated with the meme produces a crypto token (the “memecoin”) named after the meme.
3. The promoter sells the memecoin to people who buy it while the meme is in the news.
4. At the end of this, the buyers have the memecoin and the promoter has their money.
5. Nothing else happens. There is no ongoing economic activity. The memecoin has a perpetual existence and a market price; people can keep trading it for as long as they want at whatever price they want. But the promoters don’t do anything and the memecoin has no cash flows.

People  pretend not to understand this, because it is embarrassing, but this is really all that is going on. If a memecoin is an investment of real money in nothing, why would you buy one? The main answers seem to be:

1. You think the meme is funny and it amuses you to buy memecoins.
2. The meme is a focal point for some sort of online community, and you buy the memecoin to feel like you are part of that community.
3. You hope there’ll be some greater fools and you can sell the memecoin for more than you paid for it.
4. You are confused.

Anyway Donald Trump is, among other things, a meme, and people who are associated with him launched a memecoin last year, and that memecoin did exactly the only thing that a memecoin could ever possibly do, which is transfer money from its buyers to its promoters. The New York Times has an update:

Nearly 1 million people who bought President Trump’s memecoin have lost money through the end of June, according to a report by the cryptocurrency analytics firm Nansen. Their losses total $3.81 billion.

The analytics firm’s assessment was calculated this week after Mr. Trump signed an annual financial disclosure showing that he walked away with a $636 million payout on the same crypto bet, part of a haul of at least $2.2 billion from all of his business ventures in 2025. ...

Three days before his inauguration, Mr. Trump unveiled … the $TRUMP memecoin, a type of novelty currency with little practical value.

“It’s time to celebrate everything we stand for: WINNING!” Mr. Trump wrote on social media. “Join my very special Trump community. GET YOUR $TRUMP NOW!” But that turned out to be bad advice.

I feel like a crazy person. “GET YOUR $TRUMP NOW” is simply a way to say “give me money to have.” If I say to you “give me $100 now,” and you do it, then (1) I will have the $100 and (2) you won’t. Did I give you “bad advice”? Yes? But also: What? And:

Nicholas Pinto is among the losers. A frequent crypto trader who voted for Mr. Trump in 2024, Mr. Pinto said he invested a total of roughly $500,000 in the $TRUMP coin, and has now lost about half that investment.

“He is leveraging the power of being president to launch currencies, when he seems trustworthy in the public’s eye,” Mr. Pinto said in an interview. “It is kind of incredible. It is almost a legal scam.”

Sure? [2]  But also: What? He “invested” $500,000 in a memecoin? What on earth did he think was going to happen? Again, the main reasons to buy memecoins are (1) comedy, (2) community, (3) greater fools and (4) confusion; it is no wonder that the Trump memecoin tapped a rich vein.

Last week, Susquehanna Investment Group sued a bunch of anonymous traders who had bought put options on Chinese brokerage shares just before China announced a crackdown on those brokerages that caused their shares to drop. Susquehanna argued that the only possible explanation for all the put buying was insider trading, and that, since it had sold a lot of the puts, it should get to take the put buyers’ winnings back. Susquehanna got a court to freeze the alleged insider traders’ accounts, as a first step to giving Susquehanna back the money.

One problem with this is that there is no direct proof that any of the traders who bought puts were actually insider trading, though the lawsuit is just starting. Another problem is that Susquehanna was not the only trader selling puts: If the put buyers were insider trading, and if some money can be clawed back from them, it would be unfair to give all that money to Susquehanna when Citadel Securities is sitting right there. So:

Citadel Securities LLC said it was the victim of the same alleged insider-trading scheme that rival Susquehanna International Group claims made more than $100 million on options bets placed ahead of a Chinese regulatory crackdown.

Ken Griffin’s market-making firm late last week asked for court permission to join Susquehanna’s June 29 lawsuit, saying it lost about $28 million as the “victim of a brazen insider trader scheme.”

Susquehanna sued 100 John Doe defendants in Manhattan federal court, acknowledging it did not know who made the trades. But it said only the possession of inside information could plausibly explain the “high risk, high reward” bets on US exchange-traded options in securities firms like Futu Holdings Ltd. and Up Fintech Holding Ltd. that were affected by the Chinese government in a May 22 announcement.

Here is Citadel Securities’ motion to intervene, and here is its draft complaint, which is largely similar to Susquehanna’s. One interesting point in the complaint is that it contains a chart that “shows a significant increase in put options transactions in FUTU and TIGR that were flagged as potentially suspicious by Citadel Securities in the lead up to the CSRC Announcement.”

We talked last week about a reader’s suggestion that, when confronted with a suspicious rush of put option orders from Chinese brokerages, Susquehanna should have used that information to inform its own trading: It should have concluded “welp, if everyone is buying puts all at once, something bad is coming,” and gotten short Chinese brokerages itself. Of course you can’t always copy whatever retail traders do. “Probably,” I wrote, Susquehanna has a “nuanced model of when retail flows are informative and when they’re stupid.”

Citadel Securities apparently does! I mean, it has a model of when retail flows are “potentially suspicious,” which does not exactly mean “informative” but doesn’t exactly not mean that either. The complaint also says “Citadel Securities sold the put options to Defendants at a market price that was informed exclusively by publicly available information and its role as a liquidity provider to the market,” so perhaps the suspicious-trade-flagging system doesn’t inform the pricing system.

When we talked about private credit in 2024 and even 2025, there was a sort of standard narrative. Private credit, as a category, had grown from almost nothing into a huge business. Private credit managers were having a really easy time raising money, but a harder time deploying it: A lot of investors wanted to get into private credit, but there were only so many good borrowers looking for loans. I wrote in 2024:

A huge boom in private credit is a difficult thing for a private credit manager to manage: There’s so much money coming in, so many competitors getting into the space, that it’s hard to stick to tough underwriting criteria and do only the good deals. You have to do so many deals to keep up, and they can’t all be good.

With all that money flowing in, you’d expect rushed deals, high leverage levels, low interest rates, borrower-friendly covenants, and generally bad decisions that would plant the seeds of losses in a few years.

Did that happen? It’s kind of too early to tell, and the negative narrative about private credit these days is less “the covenants were bad” and more “AI happened and ate all of software.” But the point is that now there is a negative narrative about private credit, and the rush of retail money into private credit has reversed.

Perhaps that means the whole story has reversed? With all that money flowing out, there might be less competition to make new loans, so the new loans might have lower leverage, higher rates and more lender-friendly covenants. The Financial Times reports:

Large investors are committing billions of dollars to private credit funds as big institutions seek to profit from an exodus of smaller retail clients. ...

Brad Marshall, who co-runs Blackstone’s flagship $45bn private credit fund, said that many investors expected returns to increase, particularly if outflows from retail-focused vehicles limited how much money those funds were willing to lend.

“Periods of volatility [are] usually the best time to invest capital because people are nervous, capital structures are a little more conservative and pricing is a little wider,” he added, referring to the extra interest, or “spread”, that lenders are able to charge above benchmark rates. ...

“The institutional side seems to be pretty clinical about how they’re approaching” direct lending, one private credit executive said. “Leverage is a little lower [on new deals], documents are a little tighter and price is wider. Those are the dynamics institutional investors see: that this market is getting better, not worse.”

Obviously there are some flaws in a story like “private credit has gotten better because there’s less money coming into it, so put more money into it,” but it’s probably basically right.

One casual way to think about it is that, if you invest in stocks, the main decision you have to make is how much exposure you want to the AI factor. There are some companies that are bets on the continuing boom in artificial intelligence, and there are some other companies that are bets against it, and there are some other companies that are relatively unaffected. If you are enthusiastic about the AI boom, you will want a lot of AI exposure; if you are pessimistic, you will want less (or even negative) AI exposure. You might have more nuanced views — you might think some AI companies will benefit more than others from the particular AI future you envision, etc. — but your rough first cut might be “more AI” or “less AI.”

If you are picking stocks, you will pick the stocks that give you the AI exposure you want. If you are mostly a passive investor, you won’t pick stocks. But the AI factor is so big and so polarizing that, passive though you mostly are, you might care about your AI exposure. If a big index provider asks market participants “hey should we rush the big AI companies into the index,” you might answer “yes, I am a passive investor and I need that AI exposure,” or “no, I am a passive investor and I don’t want that AI exposure jammed into my portfolio.” In fact the big US index providers did ask that question, and got different answers, and emotions ran high on both sides. Even mostly passive investors care a lot about their AI factor exposure, and how much AI is in the index is an active choice made by index providers.

Naively you might think that passively investing in “emerging markets” would get you relatively little AI exposure. But in fact that too is an active choice, and depending on how you count, the emerging markets index might be roughly 0% AI or roughly 50%. [3] Bloomberg’s Ye Xie and Carolina Wilson report:

While BlackRock’s more than $150 billion iShares Core MSCI Emerging Markets ETF (ticker: IEMG) returned nearly 40% in the 12 months ended June 30, Vanguard’s rival offering notched roughly half that during the same period. For funds that moved in near lockstep for years, the differential is notable.

It comes down to Korea, where an AI-driven rally in Samsung Electronics Co. and SK Hynix Inc. fueled a more than 170% surge in the country’s benchmark Kospi index during the period — benefiting the BlackRock fund, which is exposed to the market even as Vanguard’s isn’t. And it’s shining a light on the potential perils of set-it-and-forget-it investing in an era when market concentration can skew returns.

The BlackRock ETF tracks the performance of MSCI Inc.’s emerging-market gauge, and because the index provider continues to classify South Korea as an emerging market — a decision it reaffirmed just last month in its latest annual review — BlackRock’s fund is capturing the country’s stellar, if choppy, equity gains. By contrast, the roughly $120 billion Vanguard FTSE Emerging Markets ETF (ticker: VWO) tracks a FTSE Russell emerging-markets index that doesn’t include Korea, because that provider considers the country a developed market.

The less interesting classification question here is: “Is Korea a developed market or an emerging one?” The more interesting practical question is: “Do I want AI-exposed memory chip stocks in my emerging markets allocation or not?” But you can’t answer that question passively.

Elsewhere: “The Investors Scrambling to Keep SpaceX — and Elon Musk — Out of Their Portfolios.”

I used to be a big Twitter user, so I have a grudging respect for companies that hate their power users. Here is a funny Wall Street Journal article about Lucky Strike Entertainment, “America’s largest bowling conglomerate,” which was more or less founded on the premise that bowling is boring and stupid and needed to be jazzed up:

Lucky Strike Chief Executive Thomas Shannon is incredulous. He says his strategy is crucial for bowling to survive.

“We saved bowling in America,” Shannon said. “That’s the reality.” …

Shannon is thinking of new ways to reinvent bowling, including through “gamification.” Players could earn rewards like a new car or cash, not for winning traditional games but other challenges like rolling a certain number of strikes in an allotted period. 

People who think bowling is good and fun hate this:

Competitive players say the company is ruining the sport with its focus on frivolous entertainment while allowing lanes and equipment to fall apart. 

But those people, while they bowl a lot, do not pay the bills:

Today, 10% of Lucky Strike’s revenue comes from league bowlers while 25% comes from party bowlers. The rest is “walk-in retail,” which includes bowling, arcades, food and beverage. …

“League bowling is in decline,” [Shannon] said. “We have yet to find that it has real commercial value.”

The league bowlers have sued; here is their complaint, which is sort of touching although not all that legally compelling:

Plaintiff Mitchel Feuer, who is a resident of Yonkers, New York, has been bowling regularly for nearly 50 years. Lately, he enjoys taking his grandson bowling or getting in some bowling practice on his own. Mr. Feuer has gone bowling at both Bowlero White Plains and Lucky Strike Nyack in the last four years. Mr. Feuer has been dismayed at how much bowling prices have increased during this period; as a longtime bowler, he finds Bowlero’s prices to be “extravagant.” Mr. Feuer estimates he was charged approximately $45 an hour to bowl and $12 to rent shoes at a recent visit to Bowlero White Plains—and, when he arrived at his designated lane, he realized that Bowlero had failed to even oil it.

Disclosure: I have actually been to Bowlero White Plains. It was fun. I have no idea how oiled the lanes were because I am not, at all, interested in bowling. In that sense, I agree with Shannon. But I can see why the bowlers are frustrated that the main bowling company seems to hate bowling.

Memory Chipmaker SK Hynix Kicks Off $28 Billion US Listing. Quant Hedge Funds Extend Worst Run Since 2023 as Momentum Slides. Small Hedge Funds Beat Multistrats in Rollercoaster First Half. China Quant Funds Draw Billions as AI Trounces Human Traders. Big Tech Has Suddenly Flipped on the AI Jobs Wipeout Scenario. Citi Joins Small Group of Banks Running London’s Gold Vaults. Saudis Slash Main Oil Price to Rare Discount as Market Dives. Wall Street banks recover in China amid trading boom. Judge Rules JPMorgan Still Has to Pay for Charlie Javice’s Legal Defense. How Bending Spoons built a $23bn tech empire from struggling brands. How BlueCrest’s Michael Platt fell foul of the UK taxman. How Rogue Nations Are Using Cryptocurrencies to Evade Sanctions. “I feel a little bit like I’m not good enough to live here anymore because I don’t work at an A.I. company.” Snow rooms. Belgium Fumes After FIFA Clears Balogun Following Trump Appeal. UK finds lost US Declaration of Independence copy after 250 years.

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!