Now that things are cooling down a bit in the #DeFI land, it seems like a good time to explore some of the core concepts and evaluate them from the first principles perspective.
Uniswap has been at the core of the #DeFI phenomenon from the beginning. The market making formula that powers it is very simple, but its economics and long term viability are quite complex. Lets try to dig a little deeper and try to get a better understanding of its value proposition.
If you don’t know anything about Uniswap, you probably have to at least skim through its whitepaper before moving on. This key quote from the whitepaper summarizes Uniswap protocol quite well:
Each Uniswap v1 pair stores pooled reserves of two assets, and provides liquidity for those two assets, maintaining the invariant that the product of the reserves cannot decrease.
For analysis, we will just use ETH-WBTC pair as an example. For initial analysis, we will also assume that Uniswap protocol is hypothetically running on an efficient network that has negligible transaction costs (i.e. not Ethereum). We will also ignore all the governance token shenanigans and just look at the core of the protocol itself at first.
LPs: Liquidity Providers
For Uniswap protocol to work, there needs to be set of users who provide liquidity for the trading pair. So in case of ETH-WBTC pair, users needs to provide both ETH and WBTC to the pool to allow other users to trade between ETH and WBTC.
The way Uniswap formula is designed, LPs essentially buy a share of a fund that maintains allocation of ETH/WBTC at 50%/50% by value. When Uniswap pool is out of 50%/50% balance, other users (i.e. arbitrageurs) have an incentive to rebalance the pool, but in doing so, arbitrageurs pay a 0.3% fee per each trade.
So, if we want to have an apples-to-apples comparison of Uniswap to a potential alternative, we should compare putting funds in Uniswap ETH-WBTC pool, to putting funds in an “index fund” like thing, that maintains 50%/50% allocation of ETH and WBTC that does its rebalancing on its own.
Before we go on, to reiterate, it is important that LPs understand that they are essentially buying into a fund that is maintaining 50%/50% allocation for the token pair. That may not be the best allocation for maximizing long term returns to begin with, but that is irrelevant for this analysis. The allocation split is more of a macro decision, i.e. maybe the best allocation for long term is 80% BTC, 20% ETH instead of 50%/50%, but that is a macro level decision and doesn’t allow us to do a proper apples-to-apples comparison of the protocol itself.
Uniswap vs an Index Fund
First, lets try to gain some high level intuition for Uniswap profitability. Lets look at two scenarios for our example ETH-WBTC pool:
Scenario1: Uniswap is the primary place where all the regular users/traders go to trade between ETH and BTC tokens. Thus, Uniswap is what is determining the “true price” of the ETH/BTC pair.
Scenario 2: Majority of the actual regular trading of ETH/BTC pair is happening on the other exchanges. Most of the trades that happen on Uniswap is due to arbitrageurs, since every time price changes significantly enough in other exchanges, there is a profit to be made by making a trade on other exchange and making counter trade on Uniswap.
At a high level, in scenario 1, Uniswap should do quite well against any competing “Index Fund” option that tries to maintain similar 50%/50% ratio. Because “price discovery” happens on Uniswap itself, in most cases it should stay as the “best place” to trade. Thus, with additional 0.3% fee that it collects for each trade, it should definitely be positioned as one of the best options to execute and maintain 50%/50% asset allocation.
On the other hand, in scenario 2, Uniswap will most likely do worse than an Index Fund that does its own rebalancing. This is because arbitrageurs will only trade on Uniswap if it is profitable to trade even when 0.3% fee is included. So a fund that maintains 50%/50% allocation by doing its own rebalancing using best price from other exchanges, should have much better and more efficient execution of the same asset allocation.
Of course, its never fully one way or the other as described in these scenarios. The reality for different trading pairs would be somewhere between Scenario 1 and Scenario 2. If it is more in the direction of Scenario 1, i.e. more of the trades are from the actual users who are deciding price of the asset pair, the better for Uniswap profitability.
Uniswap vs CEXes vs Other DEXes
As discussed in the previous section, the key to Uniswap being a good option for maintaining 50%/50% allocation for a particular token pair, is for it to become the dominant exchange for that token pair. Lets see what are high level Pros vs Cons of Uniswap as an exchange:
Fundamental Pros vs Centralized Exchanges (CEX):
Decentralized. No KYC, no AML requirements. Any token pair can be listed.
Easily programmable with other smart contracts. Can make trades in an atomic transaction, unlocking possibilities of previously impossible new features for financial operations.
Potentially deeper liquidity even compared to largest CEXes due to its simple setup (and current governance based incentive systems).
Fundamental Pros vs Other Decentralized Exchanges (DEX):
Higher liquidity due to its simplicity (at least as other DEXes exist today).
Generally lower and more predictable slippage for higher volume trades due to deeper liquidity.
Fundamental Cons vs Centralized Exchanges:
High transaction fees. Even built-in 0.3% is much higher compared to most CEXes. Per trade fees are particularly worse on Ethereum due to high gas costs of the network itself.
Much higher trading latencies. This is also due to Ethereum right now, and should potentially be improved in the future. However, it is likely that CEX trading latency will continue to be better compared to any type of decentralized setup.
So what is the verdict, can Uniswap be a dominant Exchange? Is this simple formula the future of market making?
In my opinion, there are enough issues with it, that it makes it hard to buy into the Uniswap dream for the long term:
First, the asset allocation of 50%/50% is very inflexible for the LPs. It seems hard to believe that so many LPs will continue to want to maintain that exact type of asset allocation once other (ahem: Governance token) incentives are over. Balancer is an alternative protocol that tries to fix this asset allocation limitation, however it introduces other issues with the overall liquidity depth. On the other hand if LPs will try to do some rebalancing operations on their own to maintain different asset allocation, that would significantly increase LPs complexity. At that point they could probably do more standard type of market making and maintain much more flexible overall asset allocation.
Second, the trading fees are simply too high. Both the built-in fee of 0.3% and Ethereum fees are too high to properly compete with the alternatives. There is also a catch-22 with the Ethereum fees:
If Ethereum transaction fees decrease significantly and network throughput increases, it is very likely that the alternative regular style DEXes will become much better. As long as those regular DEXes are able to attract standard market makers, they should have no trouble to compete with the liquidity depth of the Uniswap. If Liquidity depth is matched, standard DEXes will be much more advantageous with much lower trading fees. So Uniswap will lose all of its advantages, and there would really be no reason to trade on Uniswap instead of a more regular DEX.
If Ethereum transaction fees don’t decrease, then Uniswap will continue to stay significantly more expensive compared to CEX alternatives. And at some point, DEX options should appear on other higher throughput networks and should also take away all the advantages that Uniswap has right now.
Before we wrap up, we should also touch upon another complexity that Uniswap dominance would bring. If Uniswap has the deepest liquidity for a particular token pair, that creates a really weird economic setup where very large amount of the value for those two tokens is held in 50%/50% allocation. If the two tokens aren’t actually correlated (like they are in curve.fi, where they only trade like-for-like assets only), this type of setup will create an unnatural correlation between the two assets. This would be somewhat equivalent if the index funds that track SP500 exactly, held significant portion (i.e. >70-80%) of the total value of SP500 companies. At that point, those index funds would create unnatural correlation between different companies that exist in SP500, which would no longer make much economic sense.
To wrap things up: simple formula, complex consequences. There are of course different ways to look at the future of the Uniswap protocol and many would disagree with this particular analysis. If there is at least one thing you should take away from this blog post is that, if you plan on being an LP for Uniswap, make sure you understand that you are committing to a 50%/50% asset allocation for the pair that you are providing liquidity for.