STRIPS AMM Deep Dive 1/3: STRP-USDC LP Staking Collateral
Concept Warm-up
Staking liquidity is the amount of liquidity that stakers add to the Automated Market Maker (AMM) , denominated in STRP-USDC LP Token. The purpose of choosing STRP-USDC LP Token, instead of USDC as our AMM collateral, is to capture fundamental value for STRP tokens. STRP tokens can be perceived as an entry ticket to participate in Strips AMM and Insurance pool. In order to receive STRP-USDC LP token, stakers must stake USDC and STRP on SushiSwap. For discussion, the formula for Staking liquidity is calculated as per below:
Perpetual IRS is different from fixed term IRS. For an annualized IRS: the interest rate swap contract expires in 1 year. A perpetual IRS has no expiry.
A 1 year IRS has a duration of 1 year. By contrast, the duration level of perpetuity is (Y_i-Y_t)/Y_t. For example, at 10% yield, the duration of a perpetual IRS that pays $100 annually will equal be to 1.10/0.1=11 years. However, at an 8% yield, it will be equal to 1.08/0.08=13.5 years. This principle makes it obvious that although the maturity is perpetual, but duration may change based on the APY% level. The maturity of the perpetual IRS is infinite, while the duration of the instrument at a 10% yield is only 11 years. The cash flow early on in the life of the perpetual IRS, aka the trading PnL dominates the total position PnL. Given the duration together with leverage, the actual amplification of perpetual IRS can be (leverage * duration) of the trading position.
LP Token Collateral
In this section, we discuss in depth how STRP-USDC LP tokens are used as staking collateral. We will also analyze the impact of LP token price on stability of staking liquidity of AMMs.
Mechanism
- Staking PnL (in USDC): are collected from accumulated growth of trading fees and Realized PnL from traders’ closed and liquidated positions. It also captures the changes in Unrealized PnL marked to market when staking and unstaking.
- Staking PnL (in LP tokens): are collected from penalty charged when a user unstakes within a short period of time. This penalty is implemented to disincentivize slippage-reduction attacks. The longer you stake in the liquidity pool, the more LP tokens you may collect from other unstakers who don’t behave.
- Incentive Rewards (in STRP tokens): stakers and traders (with or without staking) will all receive STRP tokens as reward. More details about this will be released in the future.
Self-reinforcing Process & Fundamental Value Capture
- Staking on Strips AMM provides passive income denominated in USDC and STRP-USDC LP tokens. In addition, staking on Strips will also receive free STRP tokens as reward. In order to evaluate the self-reinforcing effects of using LP tokens as collateral on the token price and AMM robustness, we have backtested our AMM using historical data from protocols of Aave, Cream, Compound, Definer and dydx from Feb 15, 2021 until Aug 3, 2021 under 432 different market condition scenarios. We will release more details about how to construct such simulation backtesting in our AMM Deep Dive series in the future.
- In order to receive ROI% at 300–400% levels with Sharpe ratio at 3 and above, stakers will have to buy STRP on Sushiswap in order to get STRP-USDC LP tokens. Based on simulation statistics, base APY% is 314% on average and the staking reward APY% is 122%. Such demand will push up STRP price, as well as increase the liquidity of STRP-USDC pair on Sushiswap. A higher STRP price leads to higher LP token prices, and hence further increases the staking profit, since stakers will receive extra LP tokens sporadically together with STRP tokens as staking rewards. Higher staking rewards will encourage more stakers to buy STRP and stake on Strips, which also improve slippage for traders. Therefore, using LP token as staking collateral introduces reinforcing feedback loops and upward spiral of STRP price.
In simulations, we mocked the dynamics between Sushiswap and Strips: we can see that STRP price is driven by demand to stake. Higher STRP price will encourage more STRP demand in order to receive staking reward.
Our models also included simulated selling pressure from two external sources: (1) at monthly releases from investors’ unlocked tokens, and (2) some beta correlation of STRP to BTC. We found that despite these external selling pressures, the STRP recovers quickly as the high ROI% and Sharpe of the Strips AMM would attract stakers to buy STRP tokens and stake on Strips.
If STRP price overshoots, then the profit from staking would not be sufficient to compensate for impermanent loss such that STRP price appreciation is much faster than LP token price appreciation and the self-fulfilling process will be suspended. People will sell STRP to take profit or hold, hoping of more people to buy STRP tokens to stake, but demand of STRP purchase will shrink due to lack of staking demand, and eventually STRP would consolidate its rally. Then STRP price would drop faster than STRP-USDC LP token price, and the self-fulfilling process will be restarted again. In addition, if too many people stake, AMM’s slippage will decrease fast enough such that AMM might experience lower profitability or lower predictability of profit. This might also discourage demand to stake. Therefore, the self-reinforcing process will “self-correct” if it goes too far.
AMM Performance Matrix
In this section, we will compare the AMM’s profitability and stability between fixed-term IRS and Perpetual IRS under 216 combinations of AMM parameters over 170 days for both historical and stressed market conditions. In total, our AMM’s statistics are generated from more than 70,000 datapoints, and over simulated 7,344,000 transactions of staking and trading actions using artificial model agents with different risk preferences drawn from different probability distributions. Overall, Strips’ economy mean Sharpe is 3.275 and median Sharpe is 3.47. Strips’ net ROI mean over 170 days is 314.7% and net ROI median is 286.2%.
- First, we compare the orange(perpetual IRS) and blue(fixed term IRS) marks, they all centered around 3–4 range for Sharpe, and especially under the high yield high volatility condition (cross marks highlighted in pink area), we can see the orange cross (perpetual IRS) marks concentrated at the range where ROI% is within (300%,800%) with Sharpe remained stable at 3–4 level. By contrast, the blue crosses (fixed term IRS) are less dispersed with lower Sharpe ratio at levels below 3 (cross marks highlighted in green area). It is obvious that perpetual IRS, due to its concave nature of trading PnL, helps the AMM to remain robust even when markets experience high volatility at higher levels of APY%.
- Second, under the low yield low volatility condition, blue dots (fixed term IRS) are tightly concentrated around Sharpe of 3.5–4 while orange dots(perpetual IRS) are more dispersed. Therefore, we can tell that perpetual IRS doesn’t necessarily hinder Sharpe(risk-adjusted return), but just increase returns (of stakers) higher together with higher risk. Therefore, the trade-off is that perpetual IRS sacrifices predictability of profits at low yield low vol condition for higher robustness at high yield high vol condition.
- Last, the orange efficient frontier also shows that the concave shape of perpetual IRS trading PnL indicates that beyond Sharpe of 4, it is hard to further increase Sharpe ratio based on current model design.
- First, as highlighted in blue area, AMM can run 1yr IRS with Unrealized PnL remaining close to 0, but Realized PnL is also capped below $10m, with average at $5m after 5 months. Instead, perpetual IRS AMM runs with higher Unrealized PnL above -$10m, but will be able to accumulate more than $30–35m realized profit after 5 months.
- Second, there is hardly any difference between dot marks and cross marks for perpetual IRS (highlighted in orange area), which means perpetual IRS AMM is insensitive to yield and its volatility level, while 1yr-IRS AMM’s risk profile will change drastically based on underlying APY market. Even in conditions of high yield high vol, 1yr-IRS’s profits are more widely dispersed than perpetual IRS, which indicates less predictability of AMM’s profitability when underlying APY is volatile.
- First, given the low duration, 1yr-IRS AMM’s profitability is not too sensitive to initial pricing of the fixed rate (highlighted in blue area), but perpetual IRS AMM’s profitability is highly dependent on initial pricing. Highlighted in lower bottom yellow area, when initial haircut set too high at 0.9 while market is skewed to short or initial pricing set too low at 0.5 while market is skewed too long, then both ROI% and Sharpe ratio will be worsened. Overall, setting 0.5–0.7 for perpetual IRS tends to achieve higher ROI% for the same level of Sharpe ratio at 4.
- Obviously, 1yr-IRS AMM suffers when underlying APY is higher with higher volatility while perpetual IRS AMM remains the same performance level. Again, when setting the initial haircut at 0.7 for perpetual IRS, AMM tends to achieve higher ROI% while remaining the same Sharpe ratio at 4.
In our next medium release, we will discuss about how our AMM sets its parameters. One of the critical AMM parameter is to limit its net exposure(duration) risk by adding integrity checks on traders when they open new positions. In summary, integrity checks areused to prevent traders from putting on positions beyond AMM’s max capital capacity and risk. Based on simulation, we observed 50109 rejected trades, and 283 liquidated trades among 1–3m trades initiated in simulation. Looking at the data, it is obvious that integrity check plays a significant role to improve AMM’s risk-adjusted return and protect AMM from taking unfavorable and large positions beyond its risk capacity in any individual market.
- If we jump to observation, as expected, perpetual IRS AMM will definitely observe more liquidations, but the cluster (highlighted in orange area) is not far from 1yr-IRS, considering we have up to 3,672,000 trades initiated (open+close+liquidation+hold) in simulation. This means, regardless of underlying APY level and volatility, perpetual IRS AMM will have similar liquidation frequencies as that of 1yr-IRS.
- On the opposite side, we can see clearly 2 clusters of liquidation counts for 1yr-IRS, for which on the left-hand side must be low yield low volatility markets, while high yield high volatility market concentrated on right hand side. We don’t see obvious uptick in rejection counts between directional vs. non-directional integrity check (more details to be covered in next medium release). Given we have more than 1m trades opened in simulation, the rejection rate should be less than 5%, which is reasonably low while protecting the AMM.
- In addition, non-directional integrity only rejects about 4% more than directional integrity check. In future releases, we will discuss whether directional integrity check would necessarily re-install AMM’s profits faster than non-directional method, under momentum market, by allowing more trades initiated to lower net exposure of AMM.
- The left blue circle shows that 1yr-IRS for low yield low volatility market is insensitive to liquidation level at 3.5% or 5%. Apparently, 3.5% would increase the room of higher profits for Strips in general, at the cost of higher chance that trader’s position might be liquidated with negative net equity. The reason why all marks align on 45-degree line is because, at the end of day, the liquidation level only decides the profit split between AMM and insurance pool, the higher liquidation level, the lower profit base to be shared between AMM and insurance pool while insurance pool will have higher certainty of positive income from a smaller base. The lower liquidation level, the higher profit base to be shared between AMM and insurance pool while insurance pool will have lower certainty of positive income from a larger base. This is the reason why dots tend to have higher net return which includes PnL from insurance pool than total return.
- However, under high yield high volatility condition, perpetual IRS AMM tends to acquire higher net return than 1yr-IRS AMM with similar total return, regardless of liquidation level at 3.5% or 5%, and this is because under high yield high vol market condition, perpetual IRS will dampen the trading PnL when liquidated,compared to 1yr-IRS.
Vista of Strips’ ecosystem — Performance Time Series
Low Yield Low Volatility
Perpetual IRS Strips Net income under various market conditions and trader preferences → 5 month, 4–8m net income denominated in USDC.
If we zoom in on the Balanced market condition with neutral trader preference, and focus on the daily change in the orange line in chart above:
- Trading fee growth is non-negative.
- Realized PnL growth is a random walk with mean at profit growth of $200,560 every day.
- Unrealized PnL growth is a random walk with mean at unrealized marked to market loss $114,872 which is simply accounting ledger rather than crystalized loss. Realized PnL growth daily range surpasses Unrealized PnL by 2x.
Not surprised to see fixed term IRS has more stable profitability when underlying market has low yield and low volatility, but it also means AMM sacrifice trader’s room of upside which can be realized by higher duration. Strips wants to empower traders with more upside, and we will discuss in details about how traders can benefit from：
- capital efficiency
- flexibility of IRS to trade on interest: any trade can be constructed cross-chain, cross-protocol, on any term-structure, for any underlying asset.
- power of duration * leverage
in future use case examples.
- Trading fee growth is non-negative.
- Realized PnL growth is a random walk with mean at $19,566 (1/10 of perpetual IRS) every day.
- Unrealized PnL growth is a random walk with mean at $2,557 per day.
High Yield High Volatility
Perpetual IRS Strips Net income under various market conditions and trader preferences → 5 month, 10–20m net income denominated in USDC, when underlying market is high yield and high volatility. Better performance with less volatility on performance is due to the dampening effect of perpetual IRS when underlying yield is high.
It is obvious that AMM definitely *CAN* lose money and suffer from large drawbacks. If underlying market has high APY% and high volatility, then the 1yr-IRS Strips (collective AMMs) can suffer large peak-to-trough drawdowns of almost $4m when *A LOT OF* participants in the market would like to short until some short squeeze happen (yellow line). We do want to highlight, there are indeed cases when AMM can lose money, but it doesn’t deny the fact that over long-run on average, it is still a money making operation with positive Sharpe above 3.
People might wonder how does the AMM adapt itself when the underlying market APY is one-way traffic and heading lower. We can take compound as an example:
- When compound APY was trending lower from 10% until finally stabilized at 2%, fixed rate also resets its market level from 7% to 3%.
- Once Oracle APY stabilizes at reasonably low level in June, fixed rate can even trade above Oracle APY, as there is no necessity to hedge against a adequately low and stable rate, then fixed rate is only an tool to bet on directions and divergence/convergence between oracle APY and market fixed rate.
- As stakers accumulate their staking profits paid in LP tokens, together with higher LP token price, total staking profit will increase with accelerated pace(blue line), which is much faster than staking profit paid in USDC only (red line).
Disclaimer
Any past performance, projection, forecast or simulation of results is not necessarily indicative of the future or likely performance of any investment.
The information and publications are not intended to be and do not constitute financial advice, investment advice, trading advice or any other advice or recommendation of any sort offered or endorsed by Strips Finance
