After careful deliberation and analysis, the Umami team has determined that it is in the best interest of its depositors to pause the USDC Vault’s current asset allocation strategy and reallocate deposited assets to $USDC.
Despite accumulating substantial $ETH yield from its $GLP position, Umami’s USDC Vault has been operating at a net-loss during a majority of its rebalancing cycles since its launched on July 27th.
Umami will reimburse depositors for the net-losses at its own expense to restore the value of $glpUSDC to its original value of 1 $USDC at the time of its first deposits.
The Umami team is conducting a thorough post-mortem analysis to identify the exact causes of the Vault’s underperformance. Meanwhile, Umami will be winding down its positions in $GLP and Mycelium’s Perpetual Pools and reallocating the Vault’s assets to $USDC.
The Umami team is accelerating plans to develop its “v2” USDC, ETH & BTC Vault products. They will feature improved strategies informed by the real-world data collected from the “v1” USDC Vault.
In keeping with Umami’s commitment to transparency, the team has outlined the thinking that guided its decision and its planned next-steps below.
Trust and Transparency First
Since shortly after the USDC Vault launched, the team has been working on a series of upgrades to the Vault contract and hedging model.
However, the team concluded that even with the upgrades, performance of the current USDC Vault is unlikely to live up to Umami’s stated expectations for the product.
Umami’s commitment to transparency, trust and accountability is absolute. It will not accept deposits or operate a strategy for a product in which it does not have complete confidence and will reimburse depositors at its own expense.
A Detailed Post-Mortem
The Umami team is presently undertaking a detailed review of the Vault’s real world performance data to identify the root causes of its underperformance.
However, the team has already clearly observed certain shortcomings in the current USDC Vault strategy that it plans to correct in its “v2” USDC, ETH and BTC Vaults.
Underperformance of Mycelium’s 8 Hr SMA Perpetual Pools
The Umami team knew from the start that Mycelium’s 8 Hr SMA Perpetual Pools are complex instruments that would require careful management in a delta-hedged strategy. It also understood that “skew” on the Pools would be a significant constraint to Vault scalability and performance.
Umami structured the USDC Vault’s hedging model accordingly and made clear to its users that the Vault would not be completely delta neutral. It also launched a campaign to line up millions of dollars in TVL to support liquidity in Mycelium’s Pools ahead of the launch. In addition, Umami reviewed all of the known risks in its strategy in its docs.
Unfortunately, the actual cost of hedging using Mycelium’s 8 Hr SMA Pools proved to be higher than Umami had anticipated in it’s models. Per discussions with Mycelium, Umami understands that Mycelium’s 12 Hr Spot Pools may exhibit overall lower hedging costs.
Keeper Bot Malfunctions
On July 20th, the Umami team immediately recognized and corrected a small bug in its hedging model. A detailed post-mortem from the team can be found here.
Subsequently, on Friday, August 18th, an abrupt change in the effective leverage on Mycelium’s Perpetual Pools triggered a last resort circuit-breaker the team had built into its keeper bot. Concurrently, a sharp drop in the price of $GLP created a condition that the bot was not equipped to handle. The result was that the bot’s circuit-breaker failed to execute for six hours during the rebalancing cycle.
The first bug was corrected the day it was found. The logic that caused the second keeper malfunction was also correctable. However, the fact that an event Umami considered to be a tail risk occurred within 3 weeks of launch prompted the team to reassess the Vault’s broader hedging strategy.
The Umami team concluded that the additional technical complexity required to further optimize the performance of the 8 Hr SMA Perpetual Pools would only increase execution risk for its already complex strategy.
Umami is in the process of withdrawing deposited $USDC from $GLP and Mycelium’s Pools. Once that process is complete, Umami will send the required USDC from its own treasury to offset the net losses to $glpUSDC relative to its starting price of 1 USDC.
Withdrawals from the Vault are permissionless and can be queued by users at any time. In the coming days, Umami team will publish a detailed guide on how to handle withdrawals.
Umami will share further details about adjustments to its Vault strategy that will inform its upcoming “v2” products as soon as it completes the winding down of the current USDC Vault’s positions.
The team remains fully committed to Umami’s vision and will be working hard daily to develop and launch an improved suite of Vault products.