Liquidity providers across DeFi have come to accept the plague of impermanent loss as unavoidable. That narrative is incorrect, and now disproven. It’s time for the DeFi industry to rethink how it evaluates DEX yields. But before we dive into what exactly this looks like, it’s important to understand how IL came to exist in the first place.
The Rise of CPMM DEXs
For context, roughly $91B worth of crypto was traded through decentralized exchanges (DEXs) this November alone. DEXs use a technology called Automated Market Makers (AMMs), which allows traders to pay a small fee (e.g. 0.3%) to exchange one asset for another by drawing from a standing “pool” of crypto assets (commonly referred to as “liquidity”).
As of November 30, the top 10 DEX pools hold over $12.2B of assets on behalf of Liquidity Providers (LPs), who receive a cut of the DEX’s trading fees. These fees as a proportion of the liquidity provided is known as the “Fee Yield”. Of course, an AMM may make bad trades, in which case it realizes losses. Obviously, LPs only want to provide liquidity if they believe the fees they collect offset these losses as well as the market opportunity cost of allocating their capital to the liquidity pool. This net trading gain, or “Profit Yield” is the LP’s actual return on capital.
AMMs were first comprehensively explored by Shipyard co-founder Abe Othman in his 2007 dissertation, “Automated Market-Making: Theory and Practice.” Several years later, Ethereum cofounder Vitalik Buterin proposed an architecture for DEXs using a special type of AMM called a Constant Product Market Maker (CPMM). CPMM was chosen for its extreme computational simplicity and thus gas efficiency. Shortly after, Hayden Adams implemented a CPMM DEX named Uniswap. In 2021, Uniswap utilized Abe’s research (see Uniswap V3 whitepaper citations) to enable concentrated liquidity using a leveraged CPMM architecture. As of November, Uniswap accounted for roughly 60% of total DEX trading volume, with $3.5B in its pools. The advent of Uniswap has since set an industry trend, with multiple DEX projects implementing a similar CPMM model or outright forking Uniswap’s code. As a result, the vast majority of DEXs today are structurally similar, for better and for worse.
CPMMs’ Critical Flaw: Impermanent Loss
Unfortunately, CPMMs suffer from a critical flaw, namely impermanent loss. Essentially, every CPMM protocol earns revenue for LPs in the form of trading fees, but loses money for LPs if the market price of the assets in the pool change at all from any given LP’s initial deposit. This is a huge problem because crypto prices are constantly changing. The losses incurred by these price changes are called “Impermanent” because the loss could be zero if prices happen to revert. However, there is no inherent reason why crypto prices will naturally snap back to their “starting” value. Thus, a more accurate term might be “Unrealized Loss”. In any case, it takes an incredible, almost unrealistic amount of trading volume for fees to offset IL when providing liquidity for a CPMM, because CPMM IL is so enormous. As a result, the majority of DEX LPs have actually lost money, because the majority of DEXs are CPMMs.
IL is a constant headache for yield farmers and the DeFi community as a whole, both because it exists and because it is rarely displayed by CPMM protocols themselves. IL is also omitted and intentionally obfuscated by many DEXs because it almost always makes the CPMM look bad. Over the past several years, no one has effectively solved IL and so it has increasingly been treated as a necessary evil. But it is not. In fact, there is an enormous, unexplored design space for other AMMs, which do not bear IL like the CPMM. It’s a shame that little innovation took place in the years after Uniswap first launched, but that is no longer the case.
Clipper Has Beaten Impermanent Loss!
Clipper is not a CPMM, rather, it is a Formula Market Maker (FMM). That allows it to run much more sophisticated AMMs than the CPMM. And after a suite of recent upgrades to ban bots, Clipper now has empirical evidence that it does not experience impermanent loss.
To be clear, this does not mean Clipper LPs aren’t exposed to any risk. It is still possible to lose money while LP’ing on Clipper, relative to simply holding your initial assets. But that loss will have relatively little to do with crypto prices upon withdrawal vs. their original prices when first deposited into a liquidity pool, as is the case with IL.
Instead, Clipper’s pool closely tracks a specific benchmark called the zero-cost Daily Rebalancing Portfolio (DRP). The DRP is pretty intuitive – you may already know it as a simple diversified portfolio, where each asset has a target weight, and assets are bought and sold once per day to achieve that target. For example, a given DRP might be 50% ETH, 50% USDC. If ETH rises 10% on day 1, the DRP will sell ETH for USDC to maintain its 50%-50% target. If the price of ETH falls 5% the next day, the DRP will then sell USDC for ETH to maintain its 50%-50% target. Notice that in this case, price levels have changed but the DRP made money. By contrast, a CPMM would have lost money from all these price changes.
The DRP is usually, but not always, outperforms HODL’ing (i.e. the value you would have if you simply held your initial crypto assets instead of allocating them to a liquidity pool). This depends more on intraday volatility than it does on ending price levels. Typically, you’re better off holding a DRP if prices move a lot between daily rebalancings but ultimately revert. It’s possible for the DRP to make money if prices move 20%, and it’s possible to lose money if prices return to their initial level (although this is extremely unlikely). The worst-case price-series would be a sudden flash crash or pump followed by completely flat prices with no intraday price movements. In short, the ultimate profitability of a DRP depends on the volatility of the specific time-series of prices that occurs in the market.
It’s a mathematical certainty that a DRP would not have the IL specific to a CPMM. The question is whether Clipper meets that benchmark. Here is a visual overview of how Clipper has performed compared to the theoretical DRP following the merge.
As you can see, Clipper has effectively tracked the DRP, which means it has beaten impermanent loss! The importance of this achievement cannot be understated.
What Liquidity Providers Should Do Next
Naturally, this difference is useless if it can’t be easily factored into LPs capital allocation decisions. So if you’re an LP, here’s how we recommend thinking about your capital allocation:
First, decide whether you want exposure to a given benchmark, which is entirely dependent on macro market price series. In TradFi, this would be known as its “Beta”. Then, assess the fees that a specific DEX pool earns. In TradFi, this would probably be known as its “Alpha”. You should only compare “Fee Yields” against other DEXs with the same benchmarks. For example, Uniswap tracks the CPMM benchmark and can be compared against other CPMMs with similar assets, like Sushiswap. Clipper tracks the DRP benchmark (on mainnet, this constitutes a 30% BTC, 30% ETH, 40% stablecoin portfolio), and other future DEXs will track other benchmarks.
Given these differences, LPs can make better allocation decisions by building an income statement for each DEX. For instance:
This is a rather complex undertaking, and it’s a lot of work for individual LPs to evaluate a variety of different benchmarks. Until DeFi tools emerge to make this simpler (e.g. benchmark Indexes like the S&P 500 in TradFi or a 60-40 Stocks/Bonds portfolio), this may be asking too much of LPs.
For our part, the Clipper team has begun displaying a new “Comparable APY” metric on Clipper’s Data Dashboard, which shows the top-line returns Clipper LPs would have earned if our figure was benchmarked to a CPMM over the same period. In other words, this “Comparable APY” metric is the same type of figure DEXs like Uniswap claim their LPs make, even though it is misleadingly inflated since impermanent loss is not factored in. It’s an attempt to bridge the comparison between AMMs that track different benchmarks.
We recognize this “Comparable APY” figure is a counterfactual, but it gives Clipper LPs an easy and transparent way to provide a simple comparison between Clipper’s profit-based yields and the revenue-based figures other DEXs provide. And if Clipper’s “Comparable APY” figure seems a bit high to you, that’s because LP’ing on Clipper is indeed massively more profitable than on a CPMM-based DEX.
Stop Choosing Impermanent Loss
Ongoing concerns about solvency issues across a range of centralized crypto exchanges and lending platforms will likely continue to drive crypto users to DeFi-native exchanges and platforms. DeFi needs liquidity providers in order to function and grow. In turn, these LPs need DEXs to start providing more accurate, bottom-line yield figures in order for them to compare investment opportunities across different protocols and make better allocation decisions.
Hidden costs like IL are rarely ever accounted for in most DEXs’ advertised LP APYs, because the real return figures would be much less flattering, and oftentimes even negative – but Clipper is setting a new standard in DeFi fee transparency, while also demonstrating an entirely new approach to DEX design. Given Clipper’s empirical track record, there is now zero doubt that impermanent loss is avoidable, and the entire DeFi space will be better off if more DEXs abandon the CPMM model in favor of more efficient, value-enhancing AMM protocols.