Black Thursday for MakerDAO: $8.32 million was liquidated for 0 DAI
- Maker DAO liquidations on March 12 and 13 resulted in protocol losses of 5.67 million DAI
- This happened due to the opportunity to win liquidation auctions with zero bids, which was 36% of all liquidations
- The greatest Vault has lost ~35 000 ETH whereas the most successful liquidator has had a profit of 30 000 ETH
- $8.32 million was withdrawn through zero bids auctions in total
The drop in Ether price along with the blockchain congestion led to emergence of the negative MakerDAO Protocol’s System Surplus (a debt to the platform), which appeared due to 5.67 million DAI being uncollateralized. This problem arose not because of a sharp drop in price and the lack of collateral, but due to manipulations of initiative keepers (liquidators).
What actually happened:
- The price of gas for getting into the block increased dramatically under the congestion on blockchain (partly due to the need to update the Chainlink oracles).
- This considerable leap resulted in transactions with a price update not being sent by the MakerDAO oracles for a long time, despite the fact that the price had already increased to $30. Also, the oracle had received the market price later due to the OSM (Oracle Security Module).
- By the time the price in the protocol was updated, the magnitude of opportunities to liquidate the positions had appeared. It should be mentioned that liquidators use scripts from MakerDAO that are not programmed for adjusting high gas prices. This led that the number of liquidators who could participate in auctions had reduced.
- One of the users came up with the idea that problems with the network create a unique opportunity to profit. They began sending minimal DAI fraction in bid at the auction. As there were no competitors at that time, they have received lots up to 50 ETH for their near zero-bid DAI.
- Over time the network’s state improved and other liquidators adopted these tactics.
- Moreover, position owners have lost not just collateral demanded to return the position to the required level but all Vault collateral.
As a result, Vault (position) owners, the system, as well as MKR holders (more on this later) suffered losses. Only few liquidators benefited from this event, as well as DAI holders, which price has been higher than $1 due to the lack of liquidity.
People, concerned about MakerDAO governance, quickly noticed a problem, dubbing this day as a Black Thursday, and gathered on the Risk Call to plan future actions for eliminating system debt. Instead of Emergency Global Shutdown, a less uncompromising option for the ecosystem was chosen, namely the launch of Debt Auction. At this auction, users will be able to buy freshly minted MKRs for DAI. Auction will dilute the token share of current MKR holders. The purchase price is expected to be 200 DAI, and the lot size is 250 MKR, which makes participation in the auction quite difficult for small players. Dharma created Maker Backstop Syndicate to help them. The Paradigm crypto fund has already gathered to compete with them.
At the same time, new system parameters were selected at the MakerDAO forum, which should eliminate the possibility of repeating such liquidation scenario. The maximum lot size was increased from 50 to 500 ETH, and the duration of the auction rounds was prolonged as well. It is difficult to say how this will affect the system, because such parameters create additional risk and require more capital, but they certainly make it more complicated to carry out fraudulent manipulations in the future. At the moment, these parameters have been implemented, but it took a whole day. The reason of this lag is GSM that was activated three weeks ago, — a module that creates a delay between receiving the required number of votes and the proposal execution. Rune mentioned on the forum that the lack of GSM would solve the problem earlier, but security is always a trade-off.
To build up a coherent picture of the situation, the data on liquidations of the past two days is required. We collected this data from the Ethereum blockchain.
Following is a brief summary on collected data:
First, let’s take a look at the liquidations for these two days as a whole. We see two peaks associated with sharp changes in the price of the oracle: in the range of 15–16 hours (hereinafter the time in UTC) and after 00–01 hours of the next day.
The question may arise: Why didn’t many liquidations happen after the price change at 14 hours immediately? The reason is the fact that the liquidation means the bet that won. After the advent of the zero-bid strategy, some of the liquidators also decided to use it, which led to the presence of competition. This saved a certain part of the Vaults.
The auctions where a winner receives 100% discount on collateral are considered as liquidations with zero bid. Cumulative losses from auctions with zero bids amounted to $8.325 million. Their distribution corresponded to the previous graph of the price and number of liquidations:
Out of 3994 liquidation transactions, 1462 (36.6%) were realized with 100% discount. Consequently, more than a third of all liquidations were practically free for keepers:
Having examined the data, we found four addresses that used a strategy with zero bids. Together they have earned 62 892.93 Ether:
Let’s take a look at when each of the accounts won the first auction with a zero bid. It is obvious that the approach was successfully adopted by competitors, after the transactions had included in blocks:
We examined Vaults and the graph below indicates the top 10 positions for lost collateral, on which the liquidators earned:
It may be noticed that one of the addresses has lost near 35 000 Ether. Others lost smaller amounts, but they posted about it on social networks: Vault #849, Vault #4458 and many other users on Reddit.
To sum up
As we can see, the case with liquidations had a perceptible impact on the community and the reputation of the Maker ecosystem. We hope that our research will help to better understand the situation, and the Maker team will overcome the current difficulties.
We believe that the community should be open, and all information accessible, therefore, we provide the data that we collected and used: here it is.
This story was written by Frank Topbottom.
You can reach me on Twitter.