Prediction markets have spent the last few years trying to prove themselves as serious financial instruments rather than glorified betting pools. Then this week, someone apparently decided to rig a Spotify chart to win one.
Spotify confirmed it removed over 500,000 streams from Malcolm Todd’s track “Earrings” after detecting plays that, in the company’s words, didn’t appear to come from genuine listeners. The surge in streams coincided with heavy trading on Kalshi, a US prediction market platform that had opened a market tied to the most frequently streamed song on Spotify in the US – around $3 million was traded on that market. Spotify has since asked Kalshi and Polymarket to stop using its branding and clarified it has no partnership with either platform.
Nobody has been formally charged – the mechanics of who did what are still being investigated. But the shape of what appears to have happened is unusually clear: someone bet on a Spotify chart position, then manipulated the streams that determined the chart position, then collected on the bet. It’s fraud as a product feature.
How This Kind Of Manipulation Works
The core concept is simple enough: for a prediction market to function, it needs a verifiable, real-world event to reach a resolution. Markets tracking objective, hard-to-influence events like election results or GDP figures are inherently difficult to rig. Conversely, markets tied to platform metrics – such as stream counts or chart positions that are easily manipulated by bot activity – carry a far greater risk of subversion.
Streaming fraud is hardly a new phenomenon – bot farms and coordinated listener accounts have been a feature of the industry for as long as royalties have been tied to play counts. What’s new is the financial incentive sitting on top. Previously, manipulating Spotify streams was about gaming royalty payments or inflating an artist’s profile. Both are reasons to do it, but neither creates the kind of concentrated, time-sensitive financial incentive that a $3 million prediction market bet does. That changes the economics of running a manipulation operation considerably.
While Spotify’s policy of using detection systems to deny royalties on manipulated streams is the right approach, the situation highlights the unavoidable reality that detection remains a reactive process. There’s a window between manipulation happening and being identified, and if a prediction market settles within that window, the damage is done.
The Bigger Problem This Points To
Ultimately, this is neither a Spotify story nor a prediction market story. It’s a case study on the risks inherent in using platform-generated data as the foundation for financial instruments.
Prediction markets have grown significantly as financial products. Kalshi and Polymarket between them have handled billions in trading volume. Their appeal is that they can create liquid markets around almost any real-world outcome, including cultural ones: chart positions, box office numbers, social media follower counts, view counts on individual videos. These are all platform-derived metrics – and they’re all, to varying degrees, gameable.
The same logic extends beyond prediction markets. Advertisers use platform engagement metrics to make spending decisions. Record labels use streaming data to sign or drop artists. Investors in media companies treat monthly active user figures as signals of underlying value. Any of these contexts creates an incentive to inflate the metric if doing so produces a financial gain. Prediction markets made the incentive unusually direct and measurable.
Kalshi is currently investigating and remains in contact with Spotify, yet the underlying truth is that neither party has a clear-cut solution to this problem. Prediction markets can’t easily stop people from trading on events tied to manipulable metrics unless they restrict the events they accept, which limits the product. Platforms like Spotify can detect and remove manipulated activity, but they can’t always do it fast enough to prevent a financial market from settling.
The Part Every Founder Should Read Twice
The Spotify incident is an early and unusually visible example of a problem that will get more common. As prediction markets grow and as more financial products tie themselves to platform metrics, the incentive to manipulate those metrics grows with them.
This is worth consideration for those building businesses that treat platform data as a basis for financial decision-making. Metrics such as stream counts, follower totals, review scores, app store rankings and engagement rates remain susceptible to artificial inflation whenever the incentive is sufficient. In most contexts, the cost of manipulating them outweighs the benefit. Attach a large enough financial instrument to any of them and the calculation changes.
The regulatory picture around prediction markets is still developing, particularly in the US where Kalshi operates under CFTC oversight and Polymarket navigates a patchwork of state-level rules. Whether this incident accelerates regulatory scrutiny of which real-world events are permissible as market settlement conditions is the interesting thing to watch. Someone found a very creative and apparently profitable gap in the system – that gap is now visible to everyone.
