Developing and Backtesting a Liquidity Provider Strategy on Uniswap V2

Figure 3 illustrates how two liquidity pools that trade the same asset (ETH) can still have very different behaviors from each other. The ETH_USDC pool has higher capital efficiency metrics compared to the DAI_ETH pool. The ETH_USDC pool's capital efficiency is about twice as high as the DAI_ETH pool. Assuming all other conditions remain constant, this suggests that for each dollar invested in stablecoins as a liquidity provider, the ETH_USDC pool could generate approximately double the profits compared to the DAI_ETH pool. This analysis shows us that if a liquidity provider or trader is looking to deploy funds and if other conditions remain constant, the ETH_ USDC pool will likely make more profits.

Now, let’s look at our second pool comparison. In Figure 4, we can see that the DAI_USDC pool in Uniswap v2 has a lower trade volume and volume than other pools. These fluctuations and very low volumes on certain days could reflect current macroeconomic conditions.

Figure 7 - Table showing liquidity price and total liquidity after pool events

Figure 8 - Root table of Liquidity

As shown in Figure 8, if we use Amberdata’s liquidity snapshots for every swap, mint, and burn event, we can analyze all pool activity and simulate how much we can collect as a liquidity provider. By building the root table above and aggregating the fees from USDC and ETH separately, we can easily trace the total profit earned by the Liquidity Provider and use it to backtest our trading strategy.

Please skip to the appendix for an optional, in-depth analysis of calculating fees.

Impermanent Loss Analysis

Now that we are able to calculate the fees earned by a Liquidity Provider, we can explore another important factor of liquidity provider returns: Impermanent loss (IL). Impermanent loss is the opportunity cost a liquidity provider faces when a token’s price changes relative to its pair, between the time it is deposited in a liquidity pool and when it is withdrawn. The loss is considered impermanent because liquidity providers can recover their loss if the token pair returns to the initial exchange rate.

The bigger this change is, the more an LP is exposed to IL. This loss is referred to as impermanent because as long as your position is not closed out, the loss is not realized. Impermanent loss only affects profits when funds are removed from a pool, which means the timing itself hugely affects our total profit. For more information on impermanent loss, please read the Amberdata and Blockworks Investor’s Guide to Navigating Impermanent Loss.

Let’s take a look at the mechanics of IL and how it fluctuates with price changes.

Figure 9 showcases the position value after the initial price (P1) and the final price (P2) change. Figure 10 illustrates the same concept; as the value of the asset both decreases and increases, the IL is equal to the change in price. Whether the price change is negative or positive, there will still be IL, meaning the liquidity provider value will never be more than the original liquidity provided.

This is why understanding how to calculate fees and impermanent loss is so important. In order to generate capital as an LP, the profit earned from fees must be larger than the losses taken from IL.

Calculating impermanent loss is complex, but the core idea is simple and revolves around how the token ratios and values change while providing liquidity. When an LP deposits tokens in a pool to add liquidity, the ratio of fees earned will change.

Take Figure 11 for instance. Initially, there are 100 ETH and 40,000 DAI in the pool, but after a liquidity provider deposits their assets, the positions change to 105 ETH and 38,095.24 DAI. By recording every position’s change in prices (P1 and P2), we can calculate impermanent loss as the total value of ETH and DAI at T2 absent from the pool subtracted by the amount in the pool. In Figure 11, the IL would be $76.281-$76,190.48 = $90.52.

The first chart in Figure 12 illustrates the steady decline in the price of ETH in USDC from late March to early June of 2022. When we examine the second chart in Figure 12, we can see that our liquidity provider strategy had a more positive performance than if we were to use our fund to buy ETH and hold it in the allotted time. The third chart similarly illustrates the profit difference between a liquidity provider and an ETH holder. When the line is below zero, the ETH holder outperforms the liquidity provider, and when it is above zero, the liquidity provider outperformed the ETH holder.

While this is the first way many think to compare a liqudity provider strategy to the performance of a coin like ETH, this does not accurately represent the efficacy of providing liquidity. To provide liquidity in a pool on Uni v2, a token holder must split their tokens 50/50 - in this case, into half ETH and half USDC. Because of this, it is most accurate to compare liquidity provider returns to a simulated half-holder, or someone who holds an even split of ETH and USDC.

The first chart in Figure 13 shows the backtested performance of our liquidity provider strategy, which requires funds to be split into half ETH and half USDC, compared to if we had held onto half ETH and half USDC. The liquidity provider performance is composed of the asset itself as well as fee collections within the pool. The second chart in Figure 13 illustrates the difference in the liquidity provider strategy from the half/half holder and shows that the LP is usually outperforming the half holder.

Providing Liquidity in Varying Market Conditions

Now, we will explore how profits earned from providing liquidity may shift and vary with macroeconomic conditions.

Uptrend Market Analysis

From mid-March to mid-April of 2022, ETH slowly increased in value from around $3000 to $3400.

As we can see from the charts in Figure 14 above, as the ETH price slowly goes up, the full-hand holder outperformed the liquidity provider. As the price of ETH increases, the liquidity provider’s ETH position is gradually being swapped for USDC, diminishing its potential profitability. As also shown in Figure 14, when ETH prices rise about 10%, the LP’s profit only increases by 5% - around half of the full-holder’s profits. This is due to the structure of Uni v2, where half of the LP’s fund is invested in USDC rather than only ETH.

In order to quantify the impact of the pool’s swap events, we must exclude the impact of half of the LP’s assets being invested in USDC. To understand exactly how providing liquidity impacts our profits, we compared our liquidity provider strategy to a holder with half USDC and half ETH to mimic the liquidity provider’s assets.

The green line in the second chart of Figure 15 above shows the profit difference between our liquidity provider strategy and a holder of half USDC and half ETH. With the same slow uptrend market conditions as the full holder graphs, we can observe that there are no significant differences between investing funds into a Uni v2 pool versus holding ETH and USDC. This indicates that there is no potential alpha to be earned under slow uptrend market conditions.

Downtrend Market Analysis

Now, let’s take a look at how our LP strategy performs in downtrend market conditions. Within a two-month test period, the price of ETH dropped 33% from 3,000 USDC to 2,000 USDC.

If we look at the difference between our LP strategy and a half-holder in Figure 16, we can see that out of the sixty-day test period, our LP strategy performed 0.5–1% better than the half-holder 95% of the time. When the liquidity provider strategy does worse than the half-holder, it is due to extreme market conditions and bounces back once the market has stabilized slightly. Overall, the liquidity provider strategy performs better than holding tokens in most downtrend market conditions.

Although a common perception is that providing liquidity during upward trending markets is not as profitable as holding the assets, our backtesting results show that there is no strong difference between the two.

In fact, in every scenario but extremely downward trending markets, our liquidity provider strategy performed better than a holder of half ETH and half USDC. Additionally, when the market is steady and shows no trend, liquidity providers have profits about 2% higher than the holder since a percentage of trading fees are being collected and there is little to no impermanent loss.

During a slow-moving downtrend market when all profits are decreasing, liquidity providers and half-holders still have similar performances. However, when the market plummets rapidly, our LP profits decreased, and our strategy performed worse. When bear markets like this emerge, it is important to close out LP positions fast as the resulting impermanent loss would heavily impact returns.

Overall, providing liquidity requires patience and trust that the market (and returns) will recover once volatility is back within a reasonable range.

Conclusion

In this research paper, we have discussed the factors to consider when picking a pool on Uni v2, such as the total value locked (TVL) of a pair, the number of trades (trade volume) per day of the pair, and the characteristics and volatility of the asset itself. After comparing a few Uni v2 liquidity pools, we examined the structure of liquidity provider fees and impermanent loss and how the two interact with one another. We then backtested a trading strategy on the ETH_USDC pool and examined our LP performance in varying market conditions. Our code can be viewed here.

Our final takeaway is that providing liquidity on Uniswap v2 may be a profitable strategy for those who are looking for a safer way to hold crypto assets. Liquidity provider strategies perform well in choppy market conditions and do not have significant differences from holding the assets during upward and downward trends.

Amberdata has full liquidity snapshots for decentralized exchanges like Uniswap. We also offer Uniswap v2 and v3 DEX data that shows movement from the wallet, pool, and protocol levels of the exchange. You can leverage Amberdata's DeFi offerings to track liquidity providers, identify patterns and trends for specific pools, mitigate risk, and generate profits.

Citations:

• Impermanent Loss and Risk Profile of a Liquidity Provider 

• https://uniswap.org/blog/uniswap-v3

• https://blog.amberdata.io/the-investors-guide-to-navigating-impermanent-loss

• https://blockworks.co/the-investors-guide-to-navigating-impermanent-loss/

• https://github.com/amberdata

DISCLAIMER: The information provided in this research is for educational purposes only and is not investment or financial advice. Please do your own research before making any investment decisions. None of the information in this report constitutes, or should be relied on as, a suggestion, offer, or other solicitation to engage in, or refrain from engaging in, any purchase, sale, or any other any investment-related activity. Cryptocurrency investments are volatile and high risk in nature. Don't invest more than what you can afford to lose.