So, What Exactly ARE Prediction Markets?

Here is a full run down of what Prediction Markets are all about...

So, What Exactly ARE Prediction Markets?
Here's a full rundown of what Prediction Markets actually are and what they're all about...

So you’ve seen us on Twitter, you’ve checked out our website, you’ve heard about us on the Relay Chain Podcast, perhaps you even experimented in our Kusama Derby or Beta application. But do you know who we really are, and what we do?

It’s likely that you’ve heard we’re all about prediction markets. But it is here where newcomers often trip up…

In this blog post, we’re going to run through what prediction markets actually are, how they work, and even scratch the surface of what the Zeitgeist software is all about. We’re going to do all of this in an easy-to-understand manner.

The Gist:
As you may well know, Zeitgeist is a project that is building a software application where…

  • anyone can create their own prediction market
  • users can participate in these markets
  • companies can white label the protocol to create their own prediction market applications
  • governance can be run via prediction markets instead of standard one-man-one-vote democracy (this is known as futarchy).

We’re doing all of this in a decentralized, permissionless, and open-sourced manner, secured as a Substrate-based blockchain by connecting as a “parachain” on the Kusama network.

It goes without saying that we’re essentially a software company, building a protocol. And that protocol is an entire prediction market ecosystem that anyone can participate in.

This means that instead of users coming onto our platform and looking at what markets are worth predicting on, they can also create their very own market. This is great for variety, and provides a myriad of possible markets.

Users could come across an event in their everyday life, like while reading an interesting newspaper article, and think to themselves “Hey! This would make for a great prediction market”, and head over to our platform to create the market.

Of course, there will be ample opportunity for the more passive user who doesn’t necessarily want to create markets, but instead wants to participate in already-active markets. Such a user can merely log on, and scroll through any markets they believe would interest them, and make predictions on said interesting markets.

But let’s pause for a second and get to the quintessence of this blog post:

What exactly is a Prediction Market?

A prediction market is merely an open market where specific outcomes can be predicted using financial incentives.

Some astute wikipedia user defines them as “exchange-traded markets created for the purpose of trading the outcome of events. The market prices can indicate what the crowd thinks the probability of the event is.” (That summary seems a little narrow, so hang on a sec while I jump over and edit the entry…)


…ok. I’m back. Only got distracted by forty two more wikipedia articles and eventually ended up on “toilet paper orientation”. Curse you internet and your effect on us ADHD types!

Back to the article… The key to a prediction market is its “openness”. It acts as an open, free-market economy, like an economy operating under capitalism. Unlike betting, where odds are predetermined by bookmakers using a strict set of formulae, prediction markets start at even odds, and as time progresses, the trading activity in a market naturally regulates the price to the most likely outcomes.

Let’s apply the example to sport… And as a football (soccer) fan, let’s use the Men’s FIFA World Cup Final happening in December 2022. Of course, that final is going to be contested by Argentina and England (right?).

Argentina of course deserve to be in the final due to the strength and skill of their team, while England had a few strokes of luck, and come into the final as underdogs.

Centralized bookmakers set the odds of the game at 3/2 in Argentina’s favor based on all of the data these bookmakers have at their disposal. Those odds mean that the bookmakers are predicting that Argentina have a 67% chance of winning the game, while England have a 33% chance.

Argentina Football Team. Source: Wikimedia Commons

Again, a prediction market works without the centralized bookmaker in the middle.

Anyone could create a prediction market about the FIFA World Cup Final using the Zeitgeist application. They would enter the question “Who will win the FIFA World Cup Final”, and if they wanted to keep it simple, they would simply enter “Argentina” or “England” as the possible outcomes (this is known as a binary prediction market). They would then set an “oracle” that provides and establishes the final outcome data when the match is finished.

(The user could also add numerous outcomes such as “Draw”, “Argentina on Penalties”, “England on Penalties” etc. - but for the sake of this example we’re keeping it simple.)

Considering there are two possible scenarios in our example market, there will be two “outcome tokens” available for sale in this market:


Each token starts at an exactly even price, so in this case they’re priced at 50/50. Our ecosystem is denominated in the Zeitgeist token: ZTG. (There is the option for certain markets to be denominated in the Polkadot-based stablecoin “aUSD”, tethered to the price of the US Dollar - but more on that in another blog post.)

So, with an evenly priced binary market of 50/50, the opening price of each token would be 0.5 ztg.

As users begin to purchase which token they believe reflects the most likely outcome, the price increases or decreases based on the simple economics of supply and demand. If more people are buying “ARGWIN” then its price will increase while “ENGWIN” will decrease. Eventually, as people with insights into football make their purchases, the market will “self-regulate”, and the price of the tokens will reflect the most likely result of the match - most likely somewhere within the region of 67/33, similar to the odds already predetermined by the bookmakers.

So you see, a Prediction Market allows for accurate results without the investment into forecasters and data analysts. Prediction markets “allow the market to do the work”. The majority of people won’t make predictions unless they have some helpful insight or information into what will be the most likely outcome.

Hopefully that sporting example has helped you gain some clarity about basic prediction markets. Of course everyone has their own interests, and new markets appear all the time. For instance; a ship getting stuck in the Suez Canal - you could create a market predicting when it will be freed. Etc.

But what happens if results look skewed?

If a market looks skewed in one direction, then you have the chance to make some money by buying the other token. But keep in mind, the people who have purchased tokens so far aren’t necessarily stupid, they might just know something you don’t. The point is though: Rational and informed predictors will always bring the market into alignment with the most likely result. That’s the point of a prediction “market”, it incentivizes users to participate, and to do so accurately.

Back to our sporting example, if Argentina were overwhelming favorites to win the World Cup, but the price of the ARGWIN token was considerably low for whatever reason (irrational England supporters?), informed and rational predictors would buy up the ARGWIN token and align the market back into its more accurate outcome.

The counterexample is if someone is maliciously trying to skew the market. If you know this is happening (eg. a politician pumping their own campaign) then bet against them and take their money - they won’t be able to keep it up for long.

Market Resolution and “Redemptions”

When a market has been concluded, each winning token is redeemed for 1 ztg (or aUSD - depending on what the market was originally denominated in). So, based on our FIFA World Cup example, if you purchased ten “ARGWIN” tokens at the very beginning of the market for 0.5 ztg each, you would have spent 5 ztg. If Argentina then went on to win the match, at the conclusion of the market, you would receive 10 ztg in your Zeitgeist wallet (1 ztg for each ARGWIN token purchased), making a 5 ztg profit.

If you had purchased these “ARGWIN” tokens later on in the market, say when the prices reflected the more expected odds (67/33), you would have paid 0.67 ztg per “ARGWIN” token, spending 6.7 ztg, and then still have received 10 ztg for making the correct prediction. Making a 3.3 ztg profit.

If you purchased ENGWIN (the losing outcome), then you would receive nothing.


There are four primary types of a prediction market:

  1. Binary
  2. Categorical (also known as “Multiple Options”)
  3. Scalar (also known as “Range”)
  4. Combinatorial

In this Soccer World Cup example, we have used a very basic “Binary Market”. These are markets where the options are Yes and No, or in our example, ARGWIN or ENGWIN.

Categorical (Multiple Options) markets are like Binary markets but with more than two options.

We used a Categorical Market in our Kusama Derby campaign, where users could predict which projects would win the first Kusama Parachain Slot Auctions. This had multiple possible outcomes, but the principle remained the same.

Here’s how a Categorical Market would work if we expanded the above FIFA World Cup example into multiple options…

Let’s assume you wanted to predict who would win the tournament half way through the competition, after a number of teams have already been eliminated and only eight remain in the Quarter-Finals - we could ask the same question, but this time with eight possible outcomes:

Who will win the FIFA World Cup 2022?
Team 1
Team 2
Team 3
Team 4
Team 5
Team 6
Team 7
Team 8

When the market opens, each token (TEAM1, TEAM2, etc.) would be worth one divided by eight (⅛), which equals 0.125 ztg (1 / 8 = 0.125). But based on the same redemption principle, the winning token gets redeemed at 1 ztg each, while all other tokens get redeemed for nothing.

So if you predicted Team 3 to win the Cup, and bought ten of their tokens early on in the market at 0.125 ztg each, you would have spent 1.25 ztg.

Then if Team 3 actually did win the Cup, you would be redeemed 10 ztg for your 10x TEAM3 tokens.

This redemption formula applies no matter the price of the market. If Team 3 are the outright favorites, and their price rises to, say for instance 0.8 ztg - each TEAM 3 token still only gets redeemed at 1 ztg when the market concludes after the FIFA World Cup Final, with all other TEAM tokens receiving nothing.

In our testnet campaign The Kusama Derby, we designed the UI (User Interface) to reflect a horse race, with each token/horse moving along a slider based on the price of their token.

As you can see in the screenshot below, the DeFi project “Karura” was the clear favorite to win the first parachain slot, and its price was 0.7269 ztg. Ultimately, this means the “self-regulating market” was predicting Karura to have a 72.69% chance of winning that first Kusama parachain slot.

As per the above screenshot, we provided ten possible options in this market, with nine of those options representing a specific Polkadot/Kusama blockchain project, while the tenth was an “Other” token, allowing us to cater for any late-coming projects that entered the parachain slot auctions. (If you don’t know what a parachain slot auction is, read this).

The tokens available as options in The Kusama Derby market were as follows:

None (Meaning “none of the above).

Each token, just like in our FIFA World Cup example, would have started the market at an evenly distributed 0.1 ztg (1 ztg / 10 options = 0.1 ztg).

As users bought the tokens they thought would win, each token’s prices changed accordingly; this is thanks to the “Constant Product Market Maker” (CPMM) algorithmic formulae - as explained by our Lead Blockchain Developer Harald Heckmann here.

So with Karura being purchased by numerous users, its price appreciated significantly to 0.726 ztg. This meant that, in essence, the remaining 9 parachain projects would have had the price of 0.274 ztg (1 ztg - 0.726 ztg)  split between them.

Moonriver’s token had quite a few buyers, so its price was the highest of the remaining 9, at 0.14 ztg. So the remaining 8 tokens were split across 0.134 ztg (0.274 ztg - 0.14 ztg), and you can see our User Interface reflected these prices accordingly (the horses moved along the “track” based on the price of their token).

As you can see with these examples, Categorical markets can be incredibly insightful tools for the forecasting of events or outcomes.

Scalar (Range) Markets

The third type of prediction market is a “scalar (range) market”, which really is what it sounds like: A market where an outcome could be anywhere on a sliding scale or within a certain range. One of the best ways this can be applied is when trying to predict the price of an asset.

When predicting the price of something, there are a myriad of options. In a categorical market, there are only a few dozen potential outcomes at least, but in a scalar market, there are hundreds, if not thousands.

For example… If we wanted to predict the price of the Polkadot token ($DOT) at the end of the third quarter 2022, we could have anything from $0.01 to $20. This is far too many categories for a categorical (multiple options) market, but perfect for a scalar one!

Of course, it is up to the creator of the market to set their parameters, but the potential of parameters is myriad.

For the sake of this example, we (as the creators of this market trying to get an insight into the future price of $DOT) will set the upper limit parameter to being $20, and the lower end parameter to $0. Participants can then make their predictions for the price of Polkadot at the end of Q3 2022 to be anywhere between $0 and $20 (if participants believed the price was going to be above $20, they could predict $20 as the outcome, and still get redeemed at a full 1 ztg per prediction).

When the market resolves, the closer a participant is to the actual reported outcome, the closer to 1 ZTG they will receive per prediction token purchased.

Let’s run through an example of what this above market could look like:

  • A scalar market: What will the price of DOT be on 1 October 2022 UTC?
  • Lower parameter: $0
  • Upper parameter: $20
  • James makes a prediction that the price will be $5
  • In this specific example, let’s assume that a “$5” prediction is priced at 0.25 ztg per token at the beginning of the market, and James purchases 10 of these predictions for a total of 2.5 ztg.
  • IE. James pays 2.5 ztg for ten tokens that will be worth 1 ztg each if DOT is $5 on 1 October 2022.
  • Then, by 1 October 2022, the price of DOT is $6.
  • James was close, and according to the LMSR price formula, his tokens are worth approximately 0.9 ztg.
  • He redeems his 10 prediction tokens and receives a total of 9 ztg for making a pretty accurate prediction, making himself 6.5 ztg (9 ztg - 2.5 ztg = 6.5 ztg).

This is a key difference between a categorical market and a scalar market. If this was a categorical market, and there 20 options, from $0 to $20, and James selected $5 and the actual price turned out to be $6, James would have received nothing. Only predictors who purchased the $6 token would have been rewarded with 1 ztg for each winning token. In a scalar market however, the Logarithmic Market Scoring Rule developed by economist Robin Hanson (LMSR) allows participants to be rewarded depending on how close they were to the actual outcome, hence this kind of market is perfect for predicting the price of an asset, because the possible outcomes are so nuanced.

Combinatorial Markets

Combinatorial prediction markets are by far the most complex form of market, especially for our engineers to design! In essence, with combinatorial markets, users can combine a number of different prediction markets to create an accurate forecast of multifaceted possible outcomes.

IE. If the iPhone 14 only comes in red, it might sell less phones. But if it came with a free set of AirPods shipped in the box, it may sell a whole lot more, even if they all came in red. Add more variables to that equation, like removing fingerprint ID and an extra $150 to the retail price.

All of these variables can affect the success of the phone’s launch, and when combined they make for intriguing options. IE. You could have a red only iPhone, a free set of AirPods but it still had fingerprint ID and didn’t cost an extra $150, or, you could have any color iPhone but with no AirPods, still has fingerprint ID and DID cost an extra $150. Just those four variables (red only, free AirPods, fingerprint ID, $150 extra) alone make for a myriad of possible options, thereby increasing the prediction alternatives exponentially.

By combining many solo categorical prediction markets, we can create a whole that provides for fascinating forecasting data that cannot be achieved when these markets operate independently.

Combinatorial prediction markets are especially helpful in weather insurance, where brokers enter multiple variables, and come back with a single policy price.

We’ll do a devoted article to Combinatorial prediction markets soon, as the specifics of how they function can be overwhelmingly complex, especially if you’re new to the concept of prediction markets in the first place.


Hopefully by now you’ve been able to comprehend that prediction markets and their unique economic market-making math allow for some incredible forecasting data. They give us an apparatus whereby we can create our very own forecasts without having to poll countless people and ask for their opinions. Prediction markets let the incentive of profit attract people to participate, and let those incentives do the work for us. All we need to do as “truth seekers” is create the market.

Why is this important?

It’s important because the labor required to poll countless individuals is intense, and ensuring those individuals provide accurate information when responding to a questionnaire (which in itself is laborious) is almost an impossible task. Contrast this with a prediction market in an open economy, and the nature of human instinct will ensure overpriced shares are quickly compensated by the buying of underpriced ones, and vice versa - the natural order of the most likely outcome almost always prevails when financial incentives are involved!

As you would have seen in this blog post by now, prediction markets are fantastic for predicting almost anything:

  • Sport results
  • Asset prices
  • Corporate product decisions
  • Political decisions or candidates
  • Weather

Those who have important insight and know their stuff will always be incentivized to correct any over or under priced market, and wild card punters are disincentivized from participating in large positions because the risk of loss is too great.

Our goal at Zeitgeist is to create a smooth and extremely well formulated platform where prediction markets are easy to create and even easier to participate in. This means attracting liquidity to all markets across the platform in addition to a responsive UI and rapid response times. All of these factors are significant contributors to our mission of launching the best prediction market application available anywhere. Add to that our decentralized nature and permissionless participation, and you get for a powerful ideal.

We believe prediction markets have the power to provide truly revealing data about our world, and it is our mission to uncover that data.

In the next post, we’ll explore Futarchy (governance based on prediction markets and not one-man-one-vote democracy - in the meantime read this great post by our Economic Researcher). We will also have a few step by step guides to help you get started in using our platform.

If you have any questions or comments about what you’ve just read, feel free to ask in our Discord, or mention us on Twitter!

Here’s to more accurate forecasts!