_{Brought to you by:}

“Securities markets are a vehicle for amalgamating unorganized knowledge.”

— Michael Maloney and Harold Mulherin, authors of a study on the stock market reaction to the Challenger space shuttle crash

Prediction markets work — even when they don’t

The wisdom of crowds was discovered in 1907 when the British polymath Sir Francis Galton observed that while none of the 800 people guessing the weight of a cow at a county fair guessed correctly, the average of their guesses was spot on.

Contrary to popular opinion, this is not how prediction markets work.

The promise of prediction markets is not that a large enough group of misinformed people will inadvertently stumble on the correct answer — it’s that a small number of insiders and superforecasters will be incentivized to share their knowledge with us.

The purpose of the misinformed crowd is simply to provide that incentive. 

Francis Galton’s oft-invoked cow is therefore not the applicable case study to judge newly popular prediction markets by.

Instead, it’s the Challenger space shuttle.

When the Challenger space shuttle shockingly exploded shortly after liftoff in 1986, it took a presidential panel of experts many months to determine what had gone wrong.

The stock market figured it out much faster: “Within an hour,” an academic study on the efficiency of markets found, “the market seems to have placed the blame for the crash on Morton Thiokol, the party ultimately judged by authorities to have been at fault.”

Morton Thiokol was not the only candidate. The stock prices of all four major suppliers to the space shuttle program initially traded lower on the tragic news, but three of them recovered quickly while shares of Morton Thiokol sank further, closing 12% lower on the day — a remarkably accurate estimate of the eventual damage, according to the study.

How, exactly, the market got to that estimate so quickly remains a bit of a mystery to the study’s co-authors. “We are unable to detect the actual manner in which particular informed traders induced price discovery,” they wrote.

Safe to say, though, it was not by averaging out the guesses of a crowd of uninformed speculators.

Instead, the market somehow managed to access and incorporate the knowledge of the “low-level managers and engineers” at Morton Thiokol who immediately knew that the “O-ring” sealants their company supplied to the space shuttle (to stop ultra-hot gasses from leaking into reserves of liquid-hydrogen propellant) were most likely to blame.

“On the morning of the launch,” the authors note, “Morton Thiokol engineers in Salt Lake recommended that the launch be postponed because of concern over the O-rings given the weather at the launch site.” 

It was 18 degrees in Florida on that January day in 1986 and cold weather was a known problem for O-rings, but no one outside of a small group of Morton Thiokol engineers had ever worried about it — because who ever worries about cold weather in Florida???

But the information was there to be had: “The existence of prior knowledge of the O-ring problem suggests that investors who were aware of this private information facilitated the price discovery process on the day of the explosion.”

You can imagine how this might have happened — investment analysts and hedge fund managers putting calls into the four big suppliers to the space shuttle would have heard three of them saying, “We have no idea” and a fourth saying, “Well, we’re not sure, but there is this one thing…”

That’s all the information they would have needed to act, and unlike a government report, the process of gathering that information was accelerated by the incentive of making money (or not losing it).

In short, the knowledge of select insiders was quickly incorporated into prices by a small group of well-informed traders who were incentivized to act on that knowledge by a larger group of uninformed traders.

Prediction markets work the same way.

Talk less, bet more

People seem to enjoy dunking on prediction markets when they assign a low probability to something that ultimately comes to pass.

But they may be dunking mostly on themselves because they appear to be missing the point of both probabilities and prediction markets.

The predictive power of Polymarket, for example, has been called into question because it originally assigned only a 3% chance of Kamala Harris selecting Tim Walz as her vice presidential pick.

But there are only two valid takeaways from that, and neither of them is bad for prediction markets.

Either Polymarket was correct and Tim Walz really did have only a 3% chance of becoming the nominee for vice president…

This is possible because 3% percent chances happen approximately (let me check the math here…), 3% of the time.

Or, Polymarket was incorrect and Tim Walz had a much higher than 3% chance of becoming the nominee right from the start — in which case the 3% prediction was an opportunity to make money.

This is possible because markets are supposed to give you a chance to make money!

In the case of Tim Walz, it was an unusually high-profile opportunity that will surely attract the insiders and superforecasters who will make future Polymarket odds more accurate. 

So, right or wrong, Polymarket should be celebrated: It’s either telling us the future or giving us a chance to make money.

Dunking on them for offering Walz at 3% is like dunking on the stock market for offering undervalued stocks.

Stop dunking and start buying!

Take my bet, please

Counterintuitively, prediction markets may need some degree of imprecision to function. If the markets are exactly right, there’s no incentive for the insiders and superforecasters that make them that way to participate. 

Some say the same about equity markets, usually when they’re complaining about ETFs — equity markets, they warn, are only efficient if the smart investors that make them that way think they are inefficient enough to actively participate in.

But even a perfectly efficient equities market would be worth participating in because equities are a positive sum game (if your time horizon is long enough, even the most expensive stocks might turn out to be good investments).

Prediction markets, by contrast, are a zero-sum game (or negative-sum, after fees), so there has to be enough dumb money involved to attract the smart money that makes them accurate.

Fortunately, the persistent popularity of casinos and sports betting demonstrates that there is plenty of dumb money around — people (myself included) enjoy playing negative-sum games. 

I do a little sports betting, for example, despite the outrageous fees charged by betting sites (Tony Soprano would be more than happy with the vig they typically take). 

I know that stacks the odds against me, but I do it anyway, for one of two non-monetary reasons: to have a rooting interest in games I don't know anything about or to have the satisfaction of being right in a game I think I do know something about.

The beauty of prediction markets is that you get to take the other side.

And the reason they work is because if you don’t, someone who knows better will.

— Byron Gilliam

Tired of panels? We got you covered at Permissionless. Hear from the leading voices across different ecosystems on the design debates that will shape the future of the crypto space.

Finding Consumer Gems in Crypto

Qiao and Imran from AllianceDAO discuss infrastructure exhaustion in crypto, the Fat App Thesis and the cultural differences between blockchains. Join us to unpack whether there’s a long-term bull case for L1s and how crypto can potentially solve the failing business model of traditional media.

Watch or listen to Bell Curve on YouTube, Spotify or Apple.

Blockworks Research is conducting a survey to gain insight into the institutional staking landscape. This data will help industry leaders adopt their strategies as the industry matures.

If you're an institutional staker, we want to hear from you (and if you’re new to Blockworks Research, get 20% off of our service while you’re at it!)