Introduction

Terra Oracle Prices are the backbone of the Terra ecosystem. It dictates the exchange rate of Luna to other stablecoins such as UST to maintain UST's peg to the US Dollar. Since UST is used in almost every protocol in Terra, like Anchor and Mirror, UST's depeg to the US Dollar for an extended period of time will be catastrophic to the Terra Ecosystem and if it's depegged long enough, it will mean the collapse of Terra, much like the whole Iron Finance fiasco.

However, unlike Iron Finance, Terra handles the pegging system much more professionally. Oracle prices, or Luna exchange rates, for the stablecoins are voted and changes are implemented every 5 blocks, where validators that votes for untrue exchange rates risk being slashed. The final oracle price is obtained by using a weighted median of all the votes, with the most voted oracle price as reference.

To be included in the rewards distribution, validators must be within the reward band, or in other words, voting a price within the reward band means the validator agrees that the reference oracle price is true. Voting outside of the reward band would mean a "no" vote was casted to the reference oracle price while a non-positive vote means abstain and will neither be slashed nor obtain any rewards.

The default value for this reward band is 7% (-3.5% to +3.5%) of the final oracle price, however, this was changed in proposal 32. This proposal increased the reward band to 12% (-6% to +6%). The data for this dashboard is calculated using this reward band.

Since there are no abstain votes since 2 months ago, the abstain votes will not be visible in the graphs.

Oracle Votes (UST)

To calculate if the vote is within the reward band, all of the exchange rates were obtained, the price votes before each of the exchange rates' block and after the previous exchange rates block were obtained. The price reward band is calculated by using the current exchange rate's +-6% range. Prices outside of this range is regarded as a "no" vote. From the graphs below, we can see that most of the votes are "yes" votes. Although "no" votes started appearing since May 31, the oracle prices are mostly uncontested. Uncontested oracle prices are important to maintain stability of the stablecoins by eliminating potential malicious actors.

Note: The graph is split into 2 and only UST is calculated because otherwise the data will take too long to compute.

The graphs below show the average rate difference between the voted oracle prices for UST. It is seen that the rate difference increased since May 31, which coincides with the increase in "no" votes as seen above. Despite the increase in rate difference, it's still maintaining at a relatively low %, which means that most of the votes' values are relatively close to each other, even with a 12% tolerance.

Rate Difference by Currency (3 day data)

The graph below shows the average rate difference of all the currencies according to a 3-day data. From the graph below, it's seen that widely used currencies, such as usd and krw, tend to have a lower absolute rate difference than other less used currencies. This may be caused by the lack of sources to accurately determine the exchange rate.

Frequency of Voting Out of Bounds By Validator

From the graph below, Forbole, MoonletWallet, and Orbital Command have the highest frequencies to vote out of bounds, indicating that their source of UST and Luna pricing may be out of date or just inaccurate. Although unlikely, it could also mean that they are doing this intentionally to disrupt the oracle price. However, it is better to take caution and avoid delegating to these validators for fear of slashes and potentially delegating to a bad actor.

From observation, it's found out that the "no" votes come from medium range validators, where their voting_power is within 1M to 3M. The exceptions being Certus One and MoonletWallet, which have 12.77M and 73.6k voting_power respectively. This shows that medium range validators have a higher chance of submitting inaccurate oracle prices, although "big" validators do make this mistake from time to time.

Conclusion

The oracle price votes were obtained and calculated per exchange_rate_update event. The results show that although some "no" votes started appearing since May 31, "yes" votes still largely dominates the oracle pricing votes. The dominance of "yes" votes is essential to maintain the stability of the stablecoins by keeping the exchange rate swing to a minimum. Despite the "no" votes, the average rate difference between the voted oracle price remain at a low level, signifying that most of the validators voted for similar exchange rates. Finally, "big" validators do make the mistake of submitting inaccurate data from time to time.