Predicting our own demise

Prediction markets are having a moment.

Polymarket daily volumes are around $30 million. Kalshi is around the same, and recently raised a funding round at a valuation of $2 billion. On the regulatory front, the SEC and CFTC seem to have given up trying to limit Kalshi's transformation into a sport-betting powerhouse, making a mockery of state-level regulations on sports gambling. Indeed, prediction markets are ascendant. So is gambling in general. The world is long degeneracy.

And what's not to love, at least on paper. Economists such as Robin Hanson have been arguing for prediction markets for decades. There are strong theoretical arguments that they efficiently and accurately aggregate information from participants, much in the same way financial markets do. And who doesn't want better predictions about the future?

I want them. But prediction markets aren't going to provide them. And if these markets get big enough, they might just kill our society.

Prediction markets are bad markets

What makes a good market?

If we want to look at what makes a good market, let's look at the best ones we already have: financial markets. The best, most efficient financial markets (such as S&P500 futures, or Treasury bond futures) have a few common characteristics:

• Standardized products: Every share of AAPL is the same as every other share (in the same share class). This fungibility allows markets to pool liquidity and increase efficiency.
• Many participants: Markets work best when they maximize the number of participants. That way there's always someone willing to transact with you when you want.
• Low transaction costs: Compared to almost any other financial transaction, trading on liquid financial markets is incredibly cheap. You can buy $1 million in the SPY ETF and pay less than $10 in fees.
• Heterogenous participants: Markets find an efficient price when lots of different organizations are incentivized to contribute their own knowledge and understanding. The HFT, the earnings trader, the long-term investor: all contribute their own specialized knowledge and all are necessary for a well-functioning market.

It's that last key property, that of heterogenous participants, which prediction markets so resoundingly lack.

The missing hedgers

Fundamentally, financial markets work (and work well) because different participants have different risk preferences. This statement may come as a surprise, since we've all been taught that markets function well when people disagree with each other on the value of the traded product. While it's true that this drives a lot of trading, on its own this "belief disagreement" hypothesis would lead to almost no trading. Look up the No-trade theorem if you don't believe me. The problem is adverse selection.

Even if I have private information about some security (say that it's underpriced), I have to cross a bid/ask spread in order to put that position on. And whoever is on the other side willing to trade with me knows that the only reason I would trade is if I had private information that it's cheap. So they're going fade their offer up to the point where it would be unprofitable for me to cross the spread to trade with them. The result is no trading (or very very little trading).

But we do see lots of trading! The S&P500 future (ES) trades almost half a trillion dollars a day! But the reason is hedging. ES trades so much because people use it to hedge. They're transferring risk they don't want, and they're perfectly willing to pay a (small) price for the service the market provides. It's this activity that gives rise to all the trading we see in financial markets. People doing -EV trades, willingly, because those trades are positive utility for them. What's great is that they're positive utility for the other side too because the two sides have different risk preferences.

And it's precisely those hedgers, those participants with heterogeneous risk preferences, who are missing in prediction markets. There are two structural reasons.

Prediction markets are binaries

The outcome of a prediction market is either yes or no. One bit of information, known as a "binary contract". Binaries have been tried before. Many times in fact. And binaries have failed basically every time they've been tried. As far as I can tell, this is for two reasons:

1. They're approximately impossible to hedge, so liquidity providers hate them. Liquidity providers try to minimize risk. With a contract that settles to a continuous number, you can almost always find an instrument (usually the underyling security the derivative settles to) to hedge your risk. This makes providing liquidity relatively safe, so more people do so. With binaries, imagine the market is trading around 50%. If you're very short, you lose a lot if it settles to yes, and vice versa. But there's nothing you can do to reduce risk. All you can do is hope. That means there's a limit to how much liquidity you can provide in a given contract. All you can do is provide a small amount in a lot of contracts and hope the law of large numbers works out for you.

2. But there's a second and more fundamental problem: it's really hard to think of what natural risks exists that get optimally hedged with binaries. Maybe I'm a political appointee and have risk to the presidential election outcome. But even that seems like (a) a tiny market, (b) it's not obvious that there's much economic risk given K Street exists, and (c) that's probably better hedged by having a good social network. I think that if you dig, most or all of the supposed value of binaries for hedging evaporates under closer inspection.

It's worth noting that we already have a term for continuous-valued prediction markets: cash-settled futures.

Looking out for number one

The absence of natural hedgers in prediction markets ends up being its Achilles heel. Everyone in the market is either a noise trader (the polite term for "degenerate gambler") or a sharp. The former eventually run out of money, so a mature prediction market is full of people with the same risk preferences and goals: to make money trading. And that brings us back to the the no-trade theorem.

The only way for prediction markets to sustain themselves is on the back of new gamblers willing to lose money to the sharps. Is it any wonder that Kalshi, Polymarket and all the others are so pro-deregulation? After all, all the real money is locked up in institutions: pension funds, mutual funds, etc. If prediction markets provided value to institutional participants, why the obsession with getting retail onto their platforms? The answer is clear. Prediction markets only work when there is a steady supply of dumb money.

Markets are reflexive

I can hear you saying "Ok but that's just your opinion. What if you're wrong?"

Maybe I am. Maybe a natural class of hedgers will emerge for prediction markets that I can't think of. But even if that happens, prediction markets are still bad. Possibly fatal for the kind of society we all currently enjoy. The reason is reflexivity.

A clean prediction market

Consider a prediction market on, say, solar radiation. The sun's activity varies over time, and periods with more sunspots have higher radiation. Sometimes the sun can be pretty inactive for long periods, with corresponding effects on our climate.

So it might make sense to trade a prediction market on whether the number of sunspots exceeds some threshold. Maybe this helps farmers hedge crop prices in the event the weather becomes uncooperative as a result of reduced solar activity. So far, so good. And the reason I picked such an exotic-seeming example is that, as far as I know, we have no ability to influence solar activity. The sun is going to do its thing no matter what we do, and in particular no matter what our prediction market predicts it will do. The prediction market on sunspots doesn't causally affect the number of sunspots.

The tail wags the dog

Now consider this tweet.

The main selling point of prediction markets (other than ease-of-gambling arguments) is that they're a way to aggregate information and, through simple self-interest, create a truth-seeking mechanism. But this view completely ignores reflexivity. If the existence of a prediction market directly affects the outcome, then it's not truth-seeking anymore.

Once a prediction market becomes large enough, it transitions from being truth-seeking to being tautology-seeking.

Again, the very existence of the prediction market influences and hence distorts the underlying event that's being predicted!

We've seen this before

This isn't a new phenomenon. In financial markets, we see this anytime a derivatives market becomes larger (in liquidity, trading volume, etc) than the underyling security that defines its value. And this is especially true in cash-settled derivatives, which prediction markets are by definition. We need look no farther than the recent action by the Indian regulator SEBI against Jane Street, with the latter being accused of manipulating the underlying equities market to increase profits in the options positions they held.

The same thing isn't only possible in predictions markets, it's actively incentivized.

The vibes-based economy

The Twitter user @goodalexander has written extensively about what he calls either the distraction economy or (the term I prefer) the "vibes-based economy".

1. Define "bullshit" as the irrelevance of whether something is true or not.
2. Define the vibes-based economy as one where the ability to capture the public with bullshit is centrally economically valuable.

As an example, whether what Joe Rogan says is true doesn't matter. The more he says it, the more it "becomes true" in the sense of people believing it's true and acting accordingly. Obviously the current resident of 1600 Pennsylvania Ave knows this dynamic as well as anyone.

Back to prediction markets. They're a way of disintermediating the vibes/bullshit generation process by giving participants a direct economic stake in those vibes. And now remember reflexivity. By creating an economic incentive to manipulate probabilities, the vibes can actually become true if they're even remotely manipulable.

A concrete example

Let's consider a prediction market on "Will Trump run for reelection in 2028?" If the market is small, it's a true prediction market. It doesn't affect outcomes.

Now consider a prediction market with billions of dollars a day trading on the question. People are talking about it, thinking about it, figuring out ways it might be possible. It's now a vibe. Large amounts of societal resources are being deployed to argue for or against it. This feedback cycle directly increases the probability of the rare, changeable events through this deployment of resources.

Similarly, very probably events are made less probable.

Now ask yourself if this process is good for society? I claim it's clearly not good. Because society is a tenuous, fragile thing.

• It's easier to screw something up than to make it better.
• Salience, or vibes, of bad things have a much larger natural audience.
• The profitability of betting that a rare event will happen has a much higher upside.

So prediction markets will naturally converge much more on "will there be a civil war by 2030" types of questions than on "will we discover anti-gravity by 2030" questions.

Thus, the market for vibes (and hence predictions) will be dominated by irrelevant stuff that sucks economic resources (sports), or societally negative outcomes.

But aren't financial markets already like this?

No.

The reflexivity in financial markets (usually) leads to positive externalities, not negative ones. People plowing money into TSLA or NVDA creates risk capital to fund their competitors. New ideas get funded, technological progress happens, we all get richer, and the market becomes more efficient as a result.

The meme stock phenomenon is arguably an exception to that. To the extent that financial markets have been colonized by the vibes-based economy, that's bad thing for those markets and for the world. But that's a discussion for another day.

Conclusion

• Prediction markets are bad. They're bad market design and they induce bad incentives.
• But there's still time to put the genie back in the box.
• So let's do that.