In the last article we have broke down our calculation of liquidity providing yield into four components. Here we will give more details as to the assumptions of each component.
This is pretty straight forward. For Uniswap it’s the 0.3% trading fees. To have a more accurate gauge of the yield, we take the average daily liquidity level for the last 7 days as the liquidity, and the average daily trading volume of the same period as the volume. This is gave a more accurate number than what uniswap.info offers, which is based on 24 hours numbers.
For other platforms, it’s sometimes confusing. Sushiswap charges traders 0.3% but only 0.25% is given to the liquidity providers and the rest to Sushi token holders who staked. Out of the 0.25%, only a portion of it is distributed and the rest is locked for 6 months.
Similarly, we take the 7 day average liquidity numbers as the more accurate reflection of the liquidity level in a pool. For pool tokens, we are taking its market price as is.
For some platforms, the mining rewards to each pool is not fixed but dynamic, e.g. Balancer. We will explain that in the last part of this article series.
It’s hard to estimate impermanent loss, as it’s a matter of predicting future prices. Impermanent loss cannot be hedged away, as the movement of price in either direction (up or down) will incur impermanent costs.
We estimate impermanent costs by looking at the standard derivation of the hourly closing prices of the tokens in the pool, similarly for the last 7 days. For pairs that have two non-stablecoins in it, we assume there’s 100% negative correlation of the two tokens, to be conservative in estimating the maximum impermanent loss. In reality, some tokens do move together, e.g. BTC and ETH.
Impermanent loss is calculated by the formula below, if there are two tokens in a pool with equal total value.
In a hedged situation, the % of impermanent loss is different, as we are calculating loss based on the original value at the moment we invest, and not the moment we exit the pool. In other words, when the price goes up, the actual % of impermanent loss is higher than the one derived by the above formula, as the amount of capital we use as denominator is less. Conversely, when the price goes down, the actual % of impermanent loss is lower. For instance, if we invest in 100 ETH and 50,000 USDT in T0, the total value is $100,000; when the price of ETH goes to 1,000, we can derive a 5.7% impermanent loss based on the then value of the liquidity of $150,000, or 8.6% based on $100,000.
As we are currently using Binance, there’s a positive hedging return, rather than costs. (It can be quite high sometimes, thus giving rise to another strategy.) https://www.binance.com/cn/futures/funding-history/1
Similarly, it’s based on the past 7 days’ actual data.
Again we assuming there’s 100% negative correlation between any two non-stablecoins, e.g. YFI and ETH, so that we hedge 100% of each of the token exposures. We also arbitrarily determine that we need to provide 100% Binancen deposits for hedging with its USDT future accounts for non-mainstream tokens, e.g. if we are hedging 10 YFI at $20,000 each (with a total exposure of $200,000), we need to deposit into Binance $200,000 as reserve to buffer YFI price surge of 100%. For mainstream tokens (ETH and BTC), we deposit 50% of the exposed value. We do not hedge stablecoins (you can hedge USDT if you want). This will affect our final computation of the ROI, as it affects how much capital we are using.
(Serenity Team, 27 Nov 2020)