Amoveo’s First Futarchy Market

Tallak Tveide
5 min readAug 25, 2018

--

Amoveo is just finishing the first futarchy market with the question

X = retarget branch gets merged into mainnet before block height 29000; Y = price of veo around height 30000 is > $300; (X AND Y) OR (!X AND !Y)?

The predictions based on this market was used by Amoveo’s creator Zack Hess and the Amoveo community to determine if Amoveo should hard fork. The fork would alter the difficulty calculation algorithm. Futarchy is the notion of governing based on prediction markets, and Amoveo is perhaps the first blockchain to implement this mechanism.

To understand markets in Amoveo, please look at my other blog post. The futarchy market used by Amoveo is a normal binary prediction market. These markets are used to bet on a outcome that has two results, either true or false. The structure of the question used for futarchy though, warrants some discussion.

In essence, the Amoveo community needed to answer the question:

Is it better to fork at this point?

You can’t present this question to participants in a bet because it is totally subjective and cannot be answered by an oracle. The prediction market is not a voting mechanism. The betters are participating in the market placing bets hoping to make money. If they are only concerned about making money and not the outcome of the prediction, the prediction will be more accurate.

Making an objectively answerable question for futharcy is not straightforward. One simpler option is the recursive:

“Did Amoveo fork on block X?”

You could hope that the participants in the market thought that the community will act wisely and thus vote for the wisest option. They would know that the community would most likely react to the outcome of the prediction. Thus if forking was a good thing, it would be more likely to happen. This is ok, but is quite vague and does not intencivize the betters in predicting whick option is better. Also, such a market could be bought by someone who did not care about the forking issue. So it is not cryptoeconomically secure it seems.

You can make the question better by coupling two questions: A— Did Amoveo fork and B — The price of Amoveo is more than $300 at a certain date. Note that the price of VEO at that point was somewhere around $200. This generates four separate options. Lets call the outcomes:

A is true — A
A is not true — a
B is true — B
B is not true — b

Four options is not what we want. It is more difficult to analyze the outcome of such a market. Both AB (fork and price increase) and ab (no fork and price decrease) suggest that a fork is a good thing. Also there are no four-way markets in Amoveo yet.

We make the four options into a binary market by grouping them. The oracle was formulated so that you could bet for either ‘AB or ab’ — or — ‘aB or Ab’. At this point, the participants were left with four concrete discrete probabilities that could be analyzed to make a profitable bet in the market.

Being a participant in the market, we assume that the probability of forking has the probability 0.6. Furthermore, if no fork is done, the probability of the price rising above $300 is 0.3. If the fork happens, the probability of the price rising above $300 is 0.5. This gives the probabilities:

AB — 0.6 * 0.5 = 0.3
ab — 0.4 * 0.7 = 0.28
Ab — 0.6 * 0.5 = 0.3
aB — 0.4 * 0.3 = 0.12
AB or ab — 0.58
aB or Ab — 0.42

It is clear from this analysis that you should participate in the market, buying ‘AB or ab’ shares at a price cheaper than 58 (probability 0.58, or the decimal odds than 1.72), and ‘aB or Ab’ at a price cheaper than 42 ( decimal odds greater than 2.38).

After doing this analysis, you might assume that other people are also able to do these calculations, and the probability of forking is actually a bit higher than you first thought. So refining the probability of forking to 0.75, the price and odds change to 55–1.82 and 45–2.22. After taking the ‘metagame’ into account the odds are a bit worse than we initially thought, showing the effects of an efficient market in action.

This is an example of how these questions often take a recursive nature. The outcome of the prediction may also have consequences for the prediction itself. In these cases, you might want to look at the initial prediction before placing your own bets as a helping tool.

The first futarchy prediction market was run autumn 2018 and the voters were still inexperienced with the way these votes worked, so it ended up with a massive odds for ‘aB or Ab’ suggesting Amoveo never reaching $300 without the fork and that reaching $300 with the fork was likely to happen. This was probably not an accurate prediction.

Once the software improves and the users gain experience in playing these bets to their advantage, the quality of a prediction is sure to improve.

In the aftermath, the community decided to fork based on the prediction market. At that point, the market was still opened for trading even though the nature of the question had changed. As everyone realized that A had happened, the market was really now just betting on B or b. In a bear market for crypto, the probability of Amoveo price going higher than $300 was quite unlikely.

At one point the market flipped. All open orders in the order book were matched, even at very small odds. And th eprediction changed. After the fork was a fact, VEO seemed unlikely to be worth $300 before the due date. The people who bet on the fork early and did not react to this lost their VEO.

A recursive braintwister — perhaps losing money in the futarchy process didn’t matter much as VEO price was probably higher because of the fork

This is an important reminder to anyone participating in prediction markets. Any external circumstances or events that affect the probabilities of outcomes should immediately warrant revision of standing orders and even cancelling matched orders to maximise profit or minimize loss.

As a personal observation, you may make many oracle questions coupling two clauses (this happens) and (price is such and such). When the first clause is implemented, you should try to make the orderbook look somewhat similar after the fact. This would normally mean setting the price clause as close as possible to the current price. As we can’t predict the future, using todays price when writing the oracle question is probably the best you can do. Also, if possible, decide on (this happens) officially only after the market has closed. In a market where early participants have a disadvantage, they are not acting on their own interests by participating in the market.

Thanks for reading!

The picture of King John was taken from Wikipedia at this page.

--

--