Trading on Uniswap

Uniswap is a recently launched project built on top of Ethereum which has garnered a lot of interest over the last few months because of its novel economics.

In short, Uniswap is a kind of decentralized exchange – a crypto asset exchange which operates without a central authority – which utilizes an Automated Market Making (AMM) algorithm¹ aiming to guarantee liquidity for any order on a listed trading pair at a given volume. This research note does not aim to fully explain exactly how Uniswap works because there have already been a number of good explanatory articles written about it which we recommend the reader looks at. This note – split into two articles – will look at two key agents within the Uniswap economy and analyze the relevant data for each to ascertain the viability of the Uniswap AMM model.

Key Agents

The agents are as follows:

Traders who can submit orders to buy X amount of ETH and must pay Y amount of a given token as decided by the AMM or vice versa. In general, a trader aims to get the best available price for their trade (less fees) – defined as the lowest possible value of Y/X for a given order size when compared to alternative exchanges.

Liquidity Providers who provide liquidity for a given trading pair which the AMM then uses. For this, they are rewarded a proportional share of the 0.3% fee taken on trades carried out on Uniswap. In general, a liquidity provider aims to maximize their returns which is measured as the increase in terms of dollar value of their proportion of the liquidity pool – which is a function of: (1) the liquidity they committed (in tokens X & Y) as a proportion of total liquidity of the trading pair; (2) the dollar value of tokens X & Y each over time; and (3) the amount of fees they receive from those who use the given liquidity pool.

Trades

In order to ascertain whether traders, on average, received the best available price on Uniswap, we narrowed in on a specific trading pair – ETH/DAI. The below chart shows the derived exchange rate (on a trade-by-trade basis) for ETH/DAI and DAI/ETH trading pairs over time:

As expected, the ETH and DAI prices have an (almost perfectly) inverse relationship over time. This fact is not surprising given that it can be said to be a corollary of the function of the AMM.Compare this instead to the trading pairs for ETH & DAI on another popular order-booked based exchange called EtherDelta. The below chart shows the price of the ETH/DAI & DAI/ETH trading pairs on EtherDelta:

There is a stark difference in price stability and liquidity in EtherDelta when compared to Uniswap. This can potentially be put down to the effectiveness of the AMM thus far when judged solely on those two aforementioned factors. However, there are a few caveats to this kind of analysis: (1) liquidity on EtherDelta is generally quite low for ETH & DAI trading pairs; and (2) if one is trying to make an argument for the AMM’s ability to provide traders with good prices it makes more sense to compare Uniswap to the most popular decentralized exchanges. Thus, the chart shown below compares trading volume across the most popular decentralized exchanges (or relayers) since shortly after Uniswap’s launch to mid-March (2018-11-14 to 2019-03-14):

Analysis

As the above chart shows, Uniswap’s trading volume for most of the timeframe paled in comparison to some of the more popular decentralized exchanges such as Paradex, Radar Relay, and Kyber (as well as DDEX). This can be mostly explained by the exchange’s youth, as well as the nature of the use of a Constant Product AMM where the possibility of a ‘best price’ gets worse as the volume of the trade increases. This fact disincentivizes large trades (as a proportion of the liquidity pool) from happening, although this fact isn’t necessarily important given Uniswap’s ostensible niche as a user-friendly decentralized broker-dealer service.

Moreover, all exchanges looked at for this analysis show surprisingly small (non-weighted) average price intergroup variance which demonstrates that arbitrageur behaviour on DEXs has been relatively efficient despite potential issues with their early-stage infrastructure. Plotting the volume-weighted average prices (VWAP) of these DEXs, however, paints a slightly different picture:

As can be seen, the daily VWAP of Uniswap is more responsive to changes in price than the other DEXs; this is expected as large quantity orders (which the VWAP metric gives more weight to than a naïve average price calculation) receive a worse exchange rate than on other exchanges with similar liquidity. This fact extenuates price shocks as measured by VWAP.

Conclusion

This article has served as a brief commentary on the exchange data behind Uniswap to help readers better understand how the exchange’s mechanics affect traders. This commentary series will continue next week with a look at the data behind Uniswap’s liquidity provision (the second key agent).