Thoughts on Prediction Markets

A few of my friends have been getting into the predictions market space. Some are trying to build algorithmic trading funds, social media platforms, cross-platform arbitrage bots, and even market creation.

paying to manage risk

In my view, there are two primary value creations for assets that trade on prediction markets. The first is hedging - for some assets the most direct way to hedge against regulatory changes, public health crises, geopolitical events, etc might be through a prediction market. For example, if you’re worried that the President’s decisions on a certain bill might impact your construction business’s ability to do deals, then you might use a prediction market to hedge against those actions. We can create scenarios like this ad nauseam, hedging against measle’s outbreaks, company politics, elections, etc.

paying for efficient markets

The second primary value creation for prediction markets is the ability to enable the transfer of money from dumb people to smart people. Quantifying this is hard, but there are people who are primarily interested in gambling - the dopamine they get from watching the markets shift up and down is worth the cost of losing money on average. The other group - the ones who pay for this “transfer service” are the ones using algorithmic trading / insider information to place bets that will on average lead them to win money. They create value for the hedgers by creating efficient, real-time, and liquid markets (note that an efficient market means the EV of any bet is 0). They get paid by the stupid people who aren’t focused on making the market efficient, but are focused on dopamine hits.

what next?

So what is interesting to build in prediction markets in that case?

First, you have to choose who’s worth serving.

If serving the hedgers:

• Creating useful markets or tools that enable more market makers

If serving the degenerate gamblers:

• Things that will increase dopamine hits (prettier UIs, stroking consumer egos, enabling them to trade more based on their “insights”)

If serving the algorithmic traders:

• Infrastructure to enable faster execution, making it easier for them to implement strategies, cross-platform support
• Tools for gathering information from other data sources to inform strategies

The midwit in me says that there’s actually a distinction between smart people and dumb people, but the truth might be that they’re the same people. The tools that you build for one are easily transferrable to those in group two. So really you can serve both simultaneously.

I’m probably not going to spend much time building in this space, but I do think it’s worth keeping an eye on as a mechanism for pricing risk.