The Three-to-One Rule That Prediction Markets Don’t Want You to Understand

The Washington Post did something unusual recently. They actually checked the receipts on prediction markets — and what they found should reframe how everyone from casual bettors to institutional traders thinks about this industry.

The Math Nobody Bothers to Do

Here’s the finding that matters: for every high-profile miss in the 2022 midterm primaries, prediction markets got roughly three other races right. Candidates sitting at 75 percent odds actually won about three-quarters of the time.

That sounds almost tautological when you say it out loud. A 75 percent favorite wins 75 percent of the time. The probability matched reality.

And yet. This is the piece of information that gets buried every single time a prediction market blows a call. When the underdog wins, the takes flood in. “Markets got it wrong.” “So much for the wisdom of crowds.” “Prediction markets are just gambling dressed up in math.”

What those takes miss — what they almost always miss — is that prediction markets were never claiming certainty. A 75 percent probability explicitly acknowledges a 25 percent chance of the opposite outcome. That’s not a bug. That’s the entire point.

The Post analysis represents something rare in coverage of this industry: an attempt to evaluate prediction markets on their own terms rather than cherry-picking the misses. And the verdict is more favorable than the usual discourse suggests.

Why the Misses Make Better Headlines

The prediction market industry has a narrative problem it can’t seem to solve. Wall Street’s sharpest traders have increasingly turned to these platforms for signal, but the broader public still encounters them primarily through stories about spectacular failures.

Consider how memory works. Nobody remembers the Senate race where the 80 percent favorite cruised to victory. That’s not a story. That’s a confirmation of expectations. But the primary where the 30 percent longshot pulled off an upset? That gets months of coverage. Think pieces. Retrospective analysis. Hand-wringing about whether these markets can be trusted.

The asymmetry creates a persistent perception gap. Prediction markets look worse than they actually perform because their successes are invisible and their failures are loud.

This matters beyond mere reputation. As Congress has started paying closer attention to the prediction market space, the prevailing narrative shapes regulatory appetite. If lawmakers believe these platforms are essentially coin flips dressed up as sophisticated instruments, they’ll regulate accordingly. If they understand that a 3-to-1 success rate at the 75 percent threshold represents genuine predictive value, the conversation changes.

Calibration Is the Real Metric

The technical term for what the Post measured is calibration. A well-calibrated prediction system is one where stated probabilities match observed frequencies over time. When a platform says something has a 70 percent chance of happening, that thing should occur roughly 70 percent of the time across many trials.

Perfect calibration is the gold standard. It’s also essentially impossible to achieve in practice, especially for low-frequency events. You can’t run the 2024 presidential election a thousand times to check whether the 55 percent favorite wins 550 of them.

But elections have something most prediction targets lack: volume. There are enough races, primaries, and down-ballot contests to build a meaningful sample. The Post’s analysis suggests that at least for events with sufficient historical precedent, prediction markets are entering a new era of demonstrated reliability.

This doesn’t mean the markets are infallible. It doesn’t mean they won’t occasionally give confident odds on outcomes that never materialize. It means that when you aggregate enough predictions, the stated probabilities tend to track reality with reasonable fidelity.

For anyone making decisions based on these numbers — whether hedging portfolio risk or planning campaign strategy — that’s the relevant question. Not “will this specific prediction be right?” but “over many predictions, does the confidence level mean what it claims to mean?”

The Structural Advantages No One Discusses

Why would prediction markets achieve better calibration than, say, pundit forecasts or polling averages?

The answer involves incentives. When real money moves on outcomes, participants face consequences for overconfidence. The know-it-all who insists a race is 95 percent likely will occasionally be right. But over time, if they’re miscalibrated, they’ll bleed capital. The market eventually weeds out consistently wrong players and rewards those whose probability assessments match reality.

This is different from political commentary, where being confidently wrong carries no financial penalty. The pundit who declared the 2016 election impossible to lose never had to refund anyone. They just pivoted to explaining what they missed.

Markets create accountability that opinions don’t. That’s not a moral judgment — it’s a structural observation about how information gets priced when stakes are involved.

The platforms themselves are increasingly sophisticated about this dynamic. Kalshi’s valuation surge reflects investor confidence that properly structured prediction markets can maintain calibration as they scale. And as regulatory clarity improves, we’re seeing increasing mainstream coverage of political prediction markets that treats them as legitimate forecasting tools rather than curiosities.

What This Means for the Industry’s Future

The 3-to-1 finding matters because it lands at a pivotal moment. Prediction markets are simultaneously experiencing explosive growth and intensifying regulatory scrutiny. Whether they emerge as established financial infrastructure or get pushed to the margins depends partly on whether they can demonstrate genuine utility.

Critics will continue pointing to individual misses. That’s inevitable and not entirely unfair — high-profile failures do reveal information about market limitations. But the aggregate picture painted by systematic analysis like the Post’s offers a counterweight.

The industry’s challenge now is ensuring that this kind of evaluation becomes the norm rather than the exception. Polymarket’s latest markets and their competitors need advocates who understand probability well enough to explain why a 25 percent event happening isn’t evidence of market failure.

They also need to maintain the calibration that makes the case compelling. As more money floods into these platforms and the psychological toll of constant betting becomes better understood, there’s a risk that distorted incentives or manipulation could degrade the very quality that makes prediction markets useful.

For now, though, the data suggests something that should be obvious but apparently isn’t: prediction markets largely do what they claim to do. The favorites usually win. The probabilities mean something. And the exceptions, dramatic as they can be, fall within the bounds that the markets themselves predicted.

That’s not a guarantee of future performance. But it’s more than most forecasting methods can honestly claim.