By Benjamin (@__calabash__)

Prediction markets allow for people to express their opinion on certain real-world outcomes. The aggregate of all beliefs in a market allows everyone to gain accurate insights into possibilities of future outcomes.

But what about the belief of a group smaller than the entirety of the market? Gaining access to a group's belief within a wider market context brings a whole new dimension of statistical insights. Learning what Australians think of oil price, what gigabrain quants think of $BTC, or what insurance funds think of bushfire risk, could be handy to any number of decision makers and traders.

Using functionSPACE, I vibe-coded a team-based probability market web app that allows people to join teams/guilds/syndicates, express their individual belief, see it merged & aggregated with their team's overall belief, and (in the future) get paid out based on their whole team's performance, proportional to the amount of collateral each individual deposited into a specific market.

(This was as part of the community competition)

The What

I built a PoC of this missing prediction market primitive. A simple UI displaying the guilds with a leaderboard. The leaderboard is currently just the total guild stake, but this could be made to be more sophisticated (overall profit or profit margin). If Duolingo and Runescape have taught us anything, it's that people get addicted to chasing leaderboard highscores.

Inside a guild, we can see the aggregated guild belief versus the market's. This is the interesting signal. Additionally, we can see which members made which belief contribution. When the market is resolved, the members of a guild would be paid out proportional to the amount of collateral they put in.

What is the benefit to the traders? Or, why would a trader want to leak this alpha?

The first reason is for fun and competition. This naturally leans into the already-existing telegram groups for like-minded people to discuss, debate, and collaborate on predicting. Competition attracts competitive and bright (and maybe egotistical) people to prove they are the smartest.

The second reason is that market dynamics tend to favour early movers. For example, a position in functionSPACE increases in value if the market begins to agree with it (see the Cost Function). In other words, once a group has expressed their belief, there is an incentive alignment because there is no market-based reason to hide your trade. (This does not exhaust other reasons for privacy).

The third reason is the team dynamic and camaraderie. If your bet is completely wrong, but everyone else on your team got it very right, you still make it out in the green. Vice versa, if your whole team performs poorly apart from you, you still lose out. Being part of a team means winning together, and losing together!

The fourth reason is improved performance. It has been shown that teaming reduces bias and noise and increases information. The question becomes whether elite teams will choose to do their group belief aggregation in public or in private?

Why functionSPACE?

Aggregating a group's beliefs in a binary Arrow-Debreu market only gives you one data point: "which of these two options does the group think is most likely?" functionSPACE provides expressiveness in beliefs, and allows each member of a group to add their own nuanced take on the continuous outcome. Maybe one teammate thinks $BTC will be <$80K, while another thinks it will be somewhere between $78.6K and $78.9K. If they both agree to play in the same team, these two beliefs will be merged into one, thus representing their team's state.

Multi-outcome markets (e.g. who wins the Republican primary) in Polymarket and co. provide some use-case for teams, and bracketed markets (e.g. will ETH be between 10K-20K, or 20K-30K, etc.) provide another, but these still fail to provide a team with a combined expressiveness that would actually be as detailed and precise as the predictors belief. Combined with the other reasons for continuous distribution markets over binary prediction markets, and functionSPACE provides the perfect protocol for team-based prediction market trading.

The How

This is currently a PoC. Implementing this will be much more feasible once the engine is on-chain, rather than behind a centralised API/backend. Being on-chain means the entire team logic can be handled in a smart contract. You wouldn't actually need to do any in-depth mathematics to merge many probability distribution functions into one, you would simply need to submit them each separately, each as one account, and keep track of each user's collateral deposit.

One open question

functionSPACE provides an effectively infinite number of ways one can express their belief in a market. You need only to play around with the demo for a minute to get a grasp of how extensible it is. The same can be applied for displaying a market state. One question I am pondering is which combination of these works best for trading as a team? I see a potential in data visualisations that emphasise the team position over time as becoming very valuable. I am not yet sure what sort of trading modalities would be most appropriate to display to users.

Benjamin is a core contributor and developer in the functionSPACE team. functionSPACE is an open-source project exploring market-led resolution and novel economic instruments for prediction markets.