The Umami team is pleased to present a review of the first two weeks of real-world performance of its USDC Vault. Umami will be regularly updating depositors on Vault performance in keeping with its commitment to transparency.
Since going live on July 27th, the USDC Vault has performed largely in line with expectations. Prior to its launch, Umami worked with partner Balance to generate 30 days of backtesting data for the Vault, which is available in Umami’s docs. The data forecast a steady upwards trend in the value of deposited assets. It also forecast some volatility along the way due to the impossibility of perfectly hedging against all market risk.
In the first 48 hours post launch, there were some minor deviations from expected performance that were promptly corrected by the team. Since then, the Vault has performed very well. In fact, over the past 6 days, a conservative model shows that Umami’s USDC Vault has been averaging 25.59% APR!
Moreover, the Umami team is working on several improvements to the Vault model that could result in performance further exceeding expectations.
Measuring Vault Performance
The most relevant metric for assessing vault performance is the Price Per Share (PPS) of the Vault’s fungible $glpUSDC token. The PPS of $glpUSDC compares the total asset value in the vault to the number of $glpUSDC tokens outstanding, which are minted when $USDC is deposited into the Vault and burned when it is withdrawn. (See detailed data here).
The PPS of $glpUSDC is updated at the end of each of the Vault’s 9 hour “cycles” for rebalancing its Tracer hedges. It is important to note that the Tracer Perpetual Pools that the Vault uses to hedge against market volatility are priced based on an 8 hour SMA (Simple Moving Average) leading to some inevitable delta between the value of the hedges and the value of GLP, the Vault’s primary yield-generating asset.
Further, depositors should note that the $glpUSDC is not “pegged” at a 1:1 ratio to $USDC. It varies dynamically for incoming depositors based on the Vault’s asset value. That means that new $USDC deposits may receive $glpUSDC at a ratio greater or less than 1:1 depending on recent Vault performance.
Early Adjustment To Hedging Model
In the first 24 hours post launch, the Umami team discovered that the Vault was not placing its hedging positions correctly on Tracer.
Specifically, on July 28th, the Vault placed over-allocated assets into hedging positions by about 3.5–4.5%. This resulted in a loss of around 0.9% of Vault asset value during the cycle ending at 8:00 PM UTC on July 27th. As a result, the PPS of $glpUSDC decreased to 0.99044 $USDC per $glpUSDC.
Within approximately 4 hours of that event, the Umami team corrected the error. The team determined that it had been caused by changes to the Vault smart contract made in late stage live testing, and therefore did not show up in backtesting, which had been completed earlier.
On August 1, Umami returned 13,734.49 USDC to the Vault from its treasury to offset losses to asset value and return performance fees collected during that period. However, the amount returned did not fully close the gap in PPS because additional $USDC had been deposited into the vault since July 28th. Incoming $USDC deposited at a PPS of <1 $USDC does not dilute earlier depositors but does increase the amount of USDC that would need to be added manually to bring PPS back to 1 $USDC.
Vault Performance Since July 28
Now, with nearly two weeks of live data collected on the Vault over the course of 27 rebalancing cycles, Umami is able to offer a clear picture of overall vault performance. The team can confidently state that, subsequent to the correction of the hedging misallocation, the Vault has been performing in-line with Balance’s model.
Vault performance has exhibited the same steady upwards trendline, with mild cycle-to-cycle volatility, that was illustrated in Balance’s backtesting data. In fact, based on a conservative “smoothed” model that dampens the effect of cycle-to-cycle volatility and actually underestimates total APR, Umami’s USDC Vault has been averaging 25.59% APR over the past 6 days.
The charts below show the 6 days / ~30 cycles since Umami corrected its hedging model. The chart on the left includes the PPS at the end of each cycle. The chart on the right shows a 5 cycle moving average, taking the PPS from the 2 cycles before and after and computing the average. This is a way of reducing the noise from the pricing snapshots and focusing on the true trend.
The team is pleased to say that, after observing the USDC Vault function for two weeks “in the wild,” it has identified several opportunities to further improve Vault performance. Umami’s devs are working on these improvements now and expect them to be live in the coming weeks.
Externalizing GLP Mint & Burn Fees
Each time GLP is minted or burned, depositors pay a variable fee (currently around 0.22% for mints and 0.29% for burns using $USDC). At present, the cost of those mint and burn fees is borne collectively by all depositors in the Vault, creating a modest drag on performance. However, the team is adjusting the vault’s fee structure to shift the burden of those fees to the user whose deposit or withdrawal is actually triggering them. This will not result in a net increase in Vault fees but will protect long-term depositors from dilution from other users’ deposits/withdrawals.
Currently, the USDC Vault has some unavoidable exposure to delta during the 8 hour rebalancing cycles for Tracer’s Perpetual Pools. Essentially, the Vault “locks in” a hedge for 8 hours. If the market conditions change dramatically during that time, the hedge may prove to be too much or too little. However, the team is developing a method for moving small amounts of $USDC in and out of $GLP to balance out the Vault’s total delta exposure during these 8 hour windows, which should significantly reduce cycle-to-cycle volatility in the PPS of $glpUSDC.
Potential Future Adjustments
The Umami team is also looking into the prospect of switching its hedges from its current Tracer Pools, which rebalance every 8 hours and track an SMA of the price of BTC and ETH, respectively. It may begin using new Tracer pools, which rebalance every 12 hour and track BTC’s and ETH’s spot prices. The longer rebalancing period should reduce “drag” on the Pool’s performance from rebalances.
More importantly, the 12 hour Spot pools allow for hedges to be minted minutes before the next rebalancing cycle begins. That should greatly reduce the volatility of the Vault’s intra-cycle delta, further smoothing out performance.
Faster Than Expected Scaling
As those who have been following the launch of Umami’s USDC Vault are well aware, maintaining liquidity for Tracer’s Perpetual Pools is crucial to successfully scaling the Vault.
In essence, if the Vault were to increase its own TVL faster than pace at which TVL was scaled in its Pools, the cost of its hedges could become prohibitively high. Fortunately, that has not been a problem. The team is pleased to say that Tracer’s ability to scale TVL (Total Value Locked) in its Pools has exceeded all expectations!
Following a joint education and outreach campaign by Tracer and Umami, there has been an outpouring of investor interest in “skew farming,” Tracer’s term for providing balanced liquidity to both the long and short side of its Pools. Within the two weeks after Umami’s Vault went live, TVL in Tracer’s 3s-ETH/USD Pool, for example, has increased from around $600,000 to more than $1.7 million.
As a result, the USDC Vault has been able to scale its TVL faster than expected. In fact, Umami ended its gated whitelist period for deposits 3–5 weeks ahead of schedule. The USDC Vault is now open to deposits from the general public!
Strong Overall Performance
The Umami team is proud to say that its USDC Vault has been a success. With >25% APR and faster-than-expected scalability, the Vault has brought sustainable, risk-hedged yield to the DeFi ecosystem. As the Umami team proceeds to scale the Vault’s TVL and develop additional Vault products, the future has never looked brighter for Umami.