Futarchy 101: Organizational Use

In this brief overview, our Researcher Francisco Manuel de Pablo looks at some practical outworkings of the "Futarchy" form of governance.

Futarchy 101: Organizational Use
Futarchy is a powerful form of governance based on prediction markets.

In previous blogs, we looked at the virtues of futarchy and the governance solutions it offers. Despite the benefits and solutions, there is still a considerable lack of practical examples that would cause organizations to take Futarchy seriously as a governance and decision-making method. We need workable use cases. So in this post, I will outlay a general but suggestive guide on how a decision-making model based on Prediction Markets can be implemented.

If you haven’t yet done so, I’d advise you to read my previous post on Futarchy, as well as our guide on Prediction Markets.

1- What Do I Measure?

First, remember that the logic under which assets are bought in prediction markets and their corresponding results are comparative in nature. That is, when one is working under a Futarchical regime, what one seeks to achieve is a substantial improvement to whatever it is their governance involves (a better protocol, a better application, a better work culture, etc.). In Robin Hanson's original papers on Futarchy, what defines the quality of state variation is called a “welfare metric”. (Admittedly, this concept can get rather vague at times.).

Considering this from the vantage point of a company, we can refer to the “North Star Metric”. For practical purposes, let us assume that this previously defined metric will be the one taken as a measure of well-being. IE. Each of the prediction markets that we run, will be aimed at optimizing this specific metric. To give an example, let's say Zeitgeit’s “North Star Metric” is 'Last 7 days Average Trading Volume'.

2- What Is The Time Period To Be Evaluated?

This is another important issue to evaluate because the organization’s valuation time must align with its plans, usually linked to a roadmap. These metrics must be stipulated by the Growth Team or team in charge of setting the specific metrics in question.

I would suggest that the objective period is divided into at least two sub-periods where the corresponding prediction markets are run. For example, a company has as its north star metric, 'engaged subscribers', and its goal is to increase this by 20% per quarter. For this, two prediction markets (at least) would need to be run, each of a month and a half duration in order to adequately evaluate the partial status of the KPI.

3- Operation of the Prediction and Audit Market

Keep in mind the Futarchy mechanics: once the market is resolved, the winning asset will represent the measure that will be carried out during the stipulated period. Once that date in question has arrived, the welfare metric will be audited and, based on the result it reflects, the people who predicted correctly will be compensated. Here, we have to consider several important deadlines:

  • Market opening phase: In this phase the market opens and allows traders to exchange information, thus determining prices and the correlative distribution of probabilities on what will be the result that will most likely help achieve the desired result. Then the market will be resolved, not allowing new trades, and leaving time to implement the winning measure and evaluate its result.
  • Production phase: Once the market is resolved, a phase will pass in which the company's team must fine-tune what is necessary to carry out the winning strategy.
  • Implementation phase: After preparing what is necessary, this phase will begin where the measures will be applied, trying to achieve the expected result.
  • Audit and reward phase: Once the implementation period has passed, the current state of the welfare measure will be evaluated versus the expected one. Based on this balance, users who predicted accurately will be rewarded.

It is important that the duration of these phases is stipulated prior to the launch of the market and that they are mentioned as a variable to be considered since the implementation periods of the measure have a direct influence on the result in question.

With this in mind, let's look at an example with temporary measures in order to see it in a practical way. Let's go back to the example of our company wanting to work on the number of engaged subscribers: The company established, according to its SMART objective, an increase in this KPI of 20% for the next 3 months. For this, a first prediction market will be established that will last a month and a half (6 weeks). At the same time, it will aim to improve the metric by at least 10%. Within this period, the times of the phases will be divided as follows:

  • Market opening phase: 1 week
  • Production phase: 1 week
  • Implementation phase: 4 weeks
  • Audit and Reward phase: 1 day

After sending the rewards, a new iteration can be started with another process of the same duration, which will end at the same time that the proposed objective gets evaluated.

4- Without Metrics, There Is No Futarchy

The title of this section could actually be named “without metrics, there is nothing”. The establishment of “welfare metrics”, KPIs, or growth strategies is crucial not only for the future but also for any healthy and sustainable project. In the specific case of futarchy, we need an objective that allows us to monitor and compare states against the ones that we are trying to achieve in order to generate a governance system based on rewards for the provision of valuable information.

How to create a market based on futarchy

Ok, so how do we put into practice what we have just described? The first thing to do is select the metric that will serve as a reference for decision-making. This metric must not only be identified by the company but also needs to be understood by the users who will exchange the information that, when aggregated, will constitute the best path to take (IE. Understood by the market participants).

A reference isn’t required when metrics are used that are already included in the frontend of the site (such as Volume Traded, Floor Price, Average Price, etc.), but other times there are more specific aggregate indicators, which need to be provided to users so that the audit process is more transparent (‘last week Engaged Users’ can be an example of that).

Once this metric has been determined, the markets can then be carried out. The creation of the prediction market will have to evaluate a measurable updated state within a specific period of time. This means that, for example, a market cannot be created under the question 'What is the best measure we can take?' because in such a case there would be no specific measurable to evaluate. The concept of 'best' is entirely subjective, so the information provided can’t be adequately aggregated, and the prediction market becomes a simple survey, where incentives to make joint decisions are created (hiding their own preferences) and Liquidity Providers experience significant loss [1].
An example of a good “Futarchical” market would be:

Which of the following measures will increase the 7-day average traded volume by at least 20% within the next 60 days?

increase the 7-days average traded volume: here we are already talking about the North Star metric that we highlighted previously and the direction of the variation (it should increase)

at least 20%: reference to the magnitude of the variation

within the next 60 days: time reference to be able to measure the result

Already our prediction market pins the result to a specific event, which guarantees a reward structure not necessarily tied to the majority opinion.

With this in mind, we are able to roll out the required steps for futarchy-based decision-making…

Step 1: Create a Market

The first thing we are going to do is follow the steps previously described to create a market. Seen field by field:

  • Market name/Question: Here we enter the question formulated with the tips that we just provided.
  • Market ends: Here we will only put the date when the event ends (following the example question, at the end of the 60 day period)
  • Outcomes: All the possible outcomes get placed here.
  • Oracle: Ideally someone who fully comprehends the welfare metric.
  • Market Description: In this section, you can delve into useful information for decision making, such as where you can see the metric in question, what you want to achieve with this market, how much time users have to make predictions ( more on this later), etc.
  • Permissionless vs. Advised: Ideally there should be a prior evaluation of the market premise, but it can also be run permissionlessly.
  • Liquidity pool: This step is not mandatory, but remember that developing a liquidity pool takes additional time, which reduces production and implementation time.

Step 2: Choosing The Market Opening Period

This step determines the moment when the production of the winning outcome begins. It is important to clarify the market opening period in the market description so that users know later if they are in a trading period in which their decision will influence the decision to be made (prior to this date), or if they are simply exchanging information about the possibility that this measure has to achieve the result in question (after this date).

Although this date must be determined by the market maker, Zeitgeist will provide tools that help optimize decision-making, determining if the market stage is reliable enough, or if we have to wait a little longer for the condition to be met.

At this point you may be asking "Why keep the market open even though the decision has already been made?" Good question, and the short answer is because it's an additional dynamic auditing mechanism. By keeping the market open, users can continue exchanging information, but now the market acts as a guide for the developers of the organization as if it were a “hot and cold market”.

If one takes a picture of the state of the asset allocation when the majority measure production decision is made, one can use it to compare the current situation with this: if the allocation becomes even more polarized, it means that the market 'approves' the performance of this measure in pursuit of the KPI; if the results tend to even out, it is a signal from the market to review the direction the organization is taking with respect to that particular goal.

If we relate this to our example: suppose that before starting our market it is decided that the first 15 days will be taken to form a probability distribution around the event in question. At the end of these 15 days, the price distribution will indicate which option should be carried out. During the remaining 45 days, traders will trade assets on a measure that is already in implementation, so these trades will indicate how confident the market is about the fulfillment of this objective based on the application of the measure decided during the first 15 days.

Step 3: Produce And Apply

Here, the vision of the company against the achievement of the objective must change. From the moment the market was opened to the date when it is decided to implement a certain action, the vision must be focused outwards. This means that the organization must be aware of what happens in the prediction market and not of possible debates that may take place internally about that event. After the winning result is defined on the given date, the focus changes and turns almost completely within the organization to manage everything required to implement the outcome. Once implemented, an integrated approach is adopted, controlling performance with internal information (organization data) and external information (variation of prices in the prediction market).

Step 4: Audit and reward

Once the market closing date is reached, we will have both the values of our metric in question and the distribution of the corresponding prediction market. When the oracle reports the final result, users who provided the correct information (by predicting accurately) are rewarded.


Hopefully you now understand how to define a Prediction Market in such a way that it refers to an event that is as specific as possible to have a good prediction. Furthermore, it is important to understand the different periods that have to be established from an organization in the form of a time window to analyze the results, as well as how the assets of the market can change their purpose when the market is kept open during the implementation of the predictions. This changes the predictions from predictive signals to becoming thermometers of how well an organization is carrying out the implemented measures.

This, in a broad nutshell, is how Futarchy can work. At Zeitgeist, we believe that the use of the Futarchy represents a substantial improvement in the decision process of any organization: This is because Futarchy not only has a better incentive system but also allows a company’s stakeholders to participate in more relevant aspects of decision-making. Our hope is that this would ultimately lead towards more participatory and transparent form of governance.


[1] Let us remember that the worst case of loss in the case of a Market Scoring Rule occurs when the probability of a certain result is equal to 1. Here, as the winning result will be given by a factor internal to the prediction market and not external to it ( that is, the determining factor of the 'victory' of an asset is given by the simple majority of it), a negative feedback phenomenon occurs where the market self-induces the situation.