Music Streaming is a Prediction Market
Kalshi and Polymarket as a model for Spotify
I believe we have a new financial model to explain how music streaming functions: prediction markets.
The Wall Street Journal reports,
“Kalshi and its competitor Polymarket advertise themselves as life-changing tools for regular people — implying everyone has a fair chance to score… Instead, casual traders are bleeding cash while a small number of sophisticated pros — including trading firms with access to vast streams of data — eat their lunch, according to a Journal analysis of platform data and interviews with traders. On Polymarket, the Journal found, 67% of profits go to just 0.1% of accounts. That means less than 2,000 accounts netted a total of nearly half a billion dollars.”
Polymarket’s 2,000 big winners come from a pool of 2.3 million accounts. Prediction markets are, the Journal explains, machines for “wealth concentration” – a tiny percentage of accounts are soaking up everyone else’s money. And it’s not because those 2,000 hit the lottery by chance.
“Traders have been shelling out for access to big-data streams from third-party providers to have a leg up. Computers use the data and algorithms to predict price movements and manage risk, faster than any human. The pros also make use of their scale to make frequent, strategic trades — sometimes tens of thousands a day — and book profits on incremental moves, with a measure of attention and discipline rarely seen in recreational users. Casual traders ‘have no chance. Systematically,’ said Michael Boss, a former professional poker player and a statistician by training. On Kalshi, Boss places 60 trades a minute and modifies his bids and asks 30 times a second.”
In prediction markets, it would seem like all participants have an equal chance since they are betting against one another, not the house. And yet it’s not an equal split between winners and losers. Instead, nearly all bettors lose, while algorithm-wielding pros who move faster than the eye can see win. It’s three-card monte at scale.
To musicians, the net result of all this might look eerily familiar…
Music streaming, as we know, has resulted in the same absurdly steep split in earnings as the Journal reveals for prediction markets, with a teeny-tiny club of big winners at the top. What’s less obvious in streaming is how cash could be draining up from below, since those who do not profit from streaming do not pay to participate.
Or do we.
We now know that streaming numbers are skewed if not fixed by platforms’ manipulation of algorithms, manipulation of consumers, and by plain-old payola from labels and marketing companies. Those at the top of the pyramid – the major labels, the well connected and the well funded – benefit from these advantages. The recent controversy over marketing company Chaotic Good and their public brag about manipulations of platforms on behalf of the band Geese has made this widespread practice better known among consumers of streaming music. But content providers have pitched these pay-to-play distortions as marketing programs for years.
As Liz Pelly recently summarized for the Financial Times,
“Artists are now sold the idea that paying for algorithmic boosting is just part of marketing — or even empowering, through the old myth that digital platforms have democratised access to the music market. These narratives help platforms avoid regulation and having to take responsibility for the types of works that circulate, or the effects on artists’ livelihoods. There is also a ripple effect on the music: algorithmic promotion most benefits artists whose work is optimised towards platform logic — smoothed-out and streaming-friendly — and visibility is awarded to those with the biggest budgets.”
Meanwhile, down below, 88% of tracks in 2025 had 1,000 streams or fewer, according to annual stats collected for the music industry by Luminate. Spotify has built a financial wall at this 1,000-stream point by refusing to account for any tracks below that level – all such tracks now stream royalty-free, by unilateral decision of the platform. And yet they are streamed. What’s more, they make up the vast majority of content on the platform. Nevertheless, whatever contribution they make to value is being redirected upwards to those at the top. They have become, to use language the Wall Street Journal applied to prediction markets, unprofitable accounts.
Yet others do profit from these 88% of unprofitable streaming tracks. What would Spotify, Apple Music or Amazon Music be like without 88% of tracks on the platform? It would take a very different pitch for consumers to subscribe for access to 22% of music. (Not every song on my own albums for example, just some of them.)
All of us in recorded music are paying in to the system of streaming, even if only with our content, as surely as every bettor is paying in to prediction markets. Neither of these types of platforms can operate without a vast quantity of participants receiving no financial gain in return. That is their financial model.
Listening to: Alabaster DePlume, Dear Children of Our Children, I Knew: Epilogue
Cooking: beans
The type of music factors in here. Many noise / avant-garde / free jazz artists don't hit that level often.
Nice article read.
Your point about the 88% is well made.
Some observations:
While Spotify might have built the wall, it was at UMG's behest (threat).
The less-than-1,000 streaming threshold is a bit like Brewster's Millions. Once you aggregate them, you have a large pot of money.
It's more like enclosure rather than a prediction market, even though market dominance plays out in the same way. First and foremost, it's always about licenses and not music. Licenses can be traded, music less so. Licenses are fungible.
As a UK Government Digital Creator Report stated:
There is no evidence that there was ever a time when recorded music was the basis of substantial income for large numbers of musicians, even when total revenues were higher, in the 1990s. (Hesmondhalgh et al., 2021, p.18).
Cheers