Misunderstandings of Olympus DAO
Table of Contents
Did any of us really understand how Olympus DAO worked? I highly doubt so. In this post, I want to go over how Olympus DAO really worked, how the mechanism was really designed. Don’t get shocked: it’s quite different from how you would know.
We all remember the defi summer of 2021, where different defi and its mechanisms were being created here and there. Now we can’t talk about the defi summer without mentioning Olympus DAO can we? It was “the defi 2.0”, which also started the famous meme of (3,3). I remember also being hyped about the defi 2.0, so hyped that I forked the entire code base to cosmos SDK, won the hackatom VI 2021 with the fork.
So one of the thesis in blockchain that has been bothering my mind these days was that we do not review contracts. It is true that most of them are open source, and that contracts are mostly decentralized, but let’s be honest: we reallly don’t care much about the code that does the actual execution, but we highly depend on and trust the explanations on a twitter thread or medium posts.
I’ve also felt this when I was forking Olympus DAO myself, as all the explanations on Olympus DAO was different from what the code actually does, only explaining or repeating what the documentation says, or what a medium post says. The core concepts and the architecture of the code base for Olympus DAO was truly different from what the major explanations would go over: Thus today I decided to write about how it was different, how Olympus DAO really worked behind the scenes.
The typical understanding of Olympus DAO
Before I go into the explanation of what really was going on behind the scenes of Olympus DAO, I’d firstly like to go over the “typical” understanding of Olympus DAO to get a better comparison. Those who have a clear understanding of the typical Olympus DAO explanation, please feel free to the next section of this post, where all the fun comes in.
What’s an OHM? An OHM would be explained as a backed currency backed by DAI or other dollar pegged assets. But what does it mean to have OHM backed by DAI? It wouldn’t go below a dollar since the protocol would buy back OHM if it’s below a dollar. There’s a reason Olympus DAO was able to succeed in the overflowing defi scene: it presented an unique method of LP management. Conventional liquidity mining protocols or protocols with yield farming would “rent” liquidity from users. Why is this a problem? The liquidity is borrowed. Users could always potentially take their liquidity out and the protocol always has the vulnerability to become less secure in addition to removing the need to pay out high farming incentives to liquidity providers.
Olympus DAO doesn’t rent liquidity. It buys it. This is where the concept of bonding comes in. The protocol incentivizes users to sell their liquidity by selling OHM at a cheaper price than the market price in return. The protocol also benefits from selling OHM at a cheaper price than the market price since OHM only needs to be backed by a dollar. Thus the profit protocol obtained by bond would be discounted price — dollar per OHM. Well now the protocol “profited” from bonding, what do they do with the profit? Leaving some to back OHM by DAI and leaving some to maintain current reward for the high APY, the rest of the profit would go to the stakers. This is where (3,3) comes in. It originated from applying game theory to the concepts of actions that a user can do within the protocol. The famous (3,3) meme refers to “let’s all not sell: if none of us sell, we all profit”
How we should be understanding Olympus DAO
It’s not an exaggeration to say that money is “printed” in Olympus DAO. It’s an illusion to the stakers that their money is being doubled, tripled. Why is this an illusion?
Suppose 1 OHM is 100 DAI in the market. Suppose bonders could buy 1 OHM with 90DAI. Bonders would be able to profit by 10 DAI. Pretty straightforward, no Ponzi included up until this point.
The protocol sold 1OHM, they only need a DAI to back this 1 OHM they’ve just sold, but in the protocol’s hands they have the extra 89 DAI, [90 DAI (bought by the borrower) — 1DAI (a single DAI that is needed to back the single OHM that has been sold)]. Now that the protocol has extra 89 DAI, they’re able to mint 89 more OHM, which creates this massive illusion, such high APY, when in the end, it’s a chicken game between the stakers.
Another point that surprised me was that many of the important variables that decide the bonding rate, and the staking rewards are a fixed constant decided by governance. Most understanding is that the protocol provides staking rewards and bonding discounts based on calculation done using assets within the protocol, when in fact it was a human-controlled, fixed constant in the end.
The correct mechanism of Olympus DAO.
Well that’s what I’ve mostly seen on twitter back then: when that explanation actually includes less than 1% of how things are really happening in the contract. In fact, many of you would be surprised to see the actual “ponzi” mechanism behind Olympus. So here we go.
There are two parts of the mechanism for Olympus DAO. Bonding and Staking. Let’s go over how bonding works.
There’s three main parameters that control the bonding price of OHM: The actual price of OHM, the executing price, the risk-free price. The actual market price of OHM would always be greater than the executing price, while the executing price would always be greater than the risk-free price. So what meaning does these parameters have? Bonders are incentivized to bond, since their profit would be equal to (market price - executing price). The protocol is also incentivized and profit whenever an asset is bonded, as it benefits from (executing price - risk free price)
Whenever a user bonds their asset, debt amount would be calculated by the following formula:
where executing Price would be calculated using the following formula.
where the term debt indicates the amount of OHM that needs to be returned to the user after bond period has passed, and the control varaible and premium is set by governance. Then the profit of the protocol would be calculated by using the following formula.
For an easier understanding, let’s take an example to walk through how bonding works. Suppose the total OHM supply was 100 OHM , and 70% of OHM was staked. Alice, who has 20 DAI is trying to bond her asset to buy OHM at a discounted price. Here the debt that the protocol owns to Alice, that is, the amount of OHM Alice would be getting after the bonding period ended would be calculated using the `debt amount` formula mentioned above, which would be calculated using the formula collateral amount / executing price, which in this case executing price would be 1.1, with debt ratio being 10/100 = 0.1, Premium being 1 + 0.1 = 1.1, and the executing price being 1.1 * 1 = 1.1. Through these calculations, we get the result that Alice would be receiving 20 / 1.1 = 18.18 OHM after the bonding period. With Alice’s profit (=protocol’s debt) calculated, we’re also able to calculate the protocol’s profit using the formula above which would be equivalent to 1.818OHM(18.18 * (1.1–1) = 18.18 * 0.1 = 1.818. That is, 1.818OHM would be minted to the treasury, without any risk of OHM breaking its reserved currency.
Moving to how staking works, the concept of rebase is what makes the high APY possible and how rewards are calculated.
Every epoch (8 hours), rebase is triggered to adjust staking rewards and distribute the staking rewards. It’s mostly known that whenever a user stakes OHM, they would get sOHM in return. The important point here is that OHM and sOHM are not pegged, but maintain a peg-like being every epoch using the rebase algorithm. One of the most important concepts of staking is circulating supply, which is calculated using the following formula:
So how do we maintain the “peg” between OHM and sOHM? The protocol would calculate and mint extra sOHMs in accordance with the OHM supply using the following formula.
where rebase amount would be added to the total supply of sOHM. After rebasing total supply of sToken, the protocol iterates through all the staked accounts, distributing rewards on share base, where rewards are calculated using the following formula.
where reward rate is a fixed rate that has been decided by governance.
I’d like to finish this article by once more highlighting the importance of code base, the only decentralized answer. Not twitter threads, not centralized medium posts, but the code itself. As pioneers of the blockchain ecosystem, we shall verify, not trust.
I’m also open to any comments, questions and feedback or new ideas on cfmms and mechanism designs and researches: please feel free to reach out via twitter (https://twitter.com/mattverse2)
 Olympus DAO Medium. (Feb 2021) A Primer on Bonding. https://olympusdao.medium.com/a-primer-on-oly-bonds-9763f125c124