What Prediction Markets Got Right (and Wrong) in 2024

There was a specific moment in October 2024 when prediction markets stopped being a niche curiosity and became part of the mainstream political conversation. Cable news chyrons started quoting Polymarket odds. Pundits debated whether the numbers reflected genuine intelligence or just money chasing momentum. And for the first time, ordinary readers were Googling "what is a prediction market" — not to understand a concept, but to follow a news story.

That moment mattered. And now, with some distance, it's worth asking the honest question: what did prediction markets actually get right in 2024, and where did the story get more complicated?

The most striking thing prediction markets did in 2024 was hold a consistent directional view on the presidential election well before polling averages caught up. While national polls seesawed through the fall, market odds were pointing toward Trump with a confidence that polling aggregators simply weren't showing.

This is the core argument for prediction markets as an information tool: prices aggregate dispersed private knowledge. Traders who believe strongly in an outcome are willing to put money behind it, which means the odds reflect more than just opinion — they reflect conviction with skin in the game.

On that front, the 2024 presidential race looks like a data point in favor of the skeptics-of-polls, friends-of-markets camp. The directional call held. Weeks before Election Day, anyone watching market odds was seeing a picture that diverged meaningfully from the polling consensus — and the markets turned out to be closer to the final result.

That is a real signal, and it deserves to be taken seriously.

But here is where intellectual honesty requires slowing down.

Not all prediction market platforms performed equally. Polymarket — by far the most liquid and most-cited platform in 2024, processing billions of dollars in volume — recorded an accuracy rate of approximately 67% on election contracts. PredictIt, a smaller and older platform, came in at 93%. That is not a rounding error. That is a meaningful gap between two platforms supposedly measuring the same thing.

Why the difference? Liquidity, it turns out, is a double-edged instrument. Polymarket's scale attracted not just informed traders but also speculative volume — noise traders chasing momentum, social media herd behavior, and large individual positions that could move prices without necessarily encoding real information. The market that looked most like a truth-telling mechanism was also the one most susceptible to distortion precisely because of how much attention and money it attracted.

A 2026 Vanderbilt study made this structural problem even sharper: only about 3% of traders on these platforms were actually responsible for meaningful price discovery. The rest were following the crowd. That is a long way from the democratic "wisdom of crowds" story that gets told about prediction markets in most coverage. It is a small number of informed participants doing the epistemic work, surrounded by a much larger pool of noise.

Markets also performed unevenly depending on the type of question. Events closer in time were predicted with more reliability than distant ones. Specific, binary outcomes outperformed complex or conditional questions. The further out you go, and the more moving parts involved, the less signal the odds tend to carry.

None of this means prediction markets are broken. What it means is that they are a signal, not an oracle — and reading them well requires understanding how they work and which platforms to trust.

The 2024 experience is best understood as a proof of concept with important caveats attached. Yes, markets can outperform polling averages on directional calls. Yes, there are real conditions under which aggregated market prices encode information that other forecasting tools miss. But the mechanism is more fragile and more platform-dependent than the triumphalist coverage suggested.

The right frame is not "markets versus polls" as if one always wins. It is: under what conditions does market price discovery work, on which platforms, and for what kinds of questions? That is a more useful and more honest question than most of what was published in 2024.

There is a genuine gap in how financial and political media covers prediction markets. Most pieces either oversell the technology — citing odds as if they were certainties — or dismiss it after a bad data point. Very little coverage actually does the work of explaining platform differences, trader composition, or the structural conditions under which market prices are and are not reliable.

That gap is exactly why Consensus exists. Not to be a scoreboard for prediction market wins and losses, but to be the translation layer — helping readers understand what the odds actually mean, when to weight them heavily, and when to treat them as noise.

2024 showed that these markets are worth taking seriously. It also showed that taking them seriously requires more than just reading the number.

The signal is real. Reading it well is the work.

— Tony
Founder at Consensus
www.PredictionMarkets.media
@ReadConsensus