This article will dive into the releases of Uniswap, mainly the switch from v2 to v3 and the effects we can observe. All data mentioned is about Uniswap on Ethereum can be found in this Dune Dashboard.
Uniswap launched its first ever product in November 2018, revolutionizing DEXs as we know it today using Automated Market Makers. What first started with a simple swap system has now grown significantly to other features, from AMMs to single side liquidity providing to different swap fee tiers. These changes have come about with 2 new version releases, v2 and v3. Version 2 at the time was a major upgrade when launched in May 2020. It offered swapping between ERC20-ERC20 pools, while in v1 only eth-erc20 pools were offered. Version 2 furthermore improved the user interface and experience to the inferior v1. It is no suprise when looking at the graph below that V2 captured over 99% of the volume in just 4 months between v1 and v2. Volume in that period also saw explosive growth, with barely $10m volume per day in early May to $200m daily volume at the end of August (Defi summer 2020).
https://dune.xyz/embeds/494487/936586/65107ff6-de23-44ab-8bc3-d2e00df4f0cf?display=iframe
V3 launched a year later in May 2021, and offered concentrated liquidity, fee tiers, and the ability to set buy orders. Uniswap designed a migrate function to easily swap your liquidity from v2 to v3. In total that feature has been used 11256 times, of which 10597 in the first 4 months, especially the first few days:
https://dune.xyz/embeds/494494/936596/30a34999-d526-463a-bc02-e86b598527b5?display=iframe
Interestingly, of the migrations most (82.7%) went to v3 pairs with 0.3% fee amounts. The same fee structure as used in v2. This might indicate that the fee was not import for those moving, but the concentrated liquidity advantage was.
https://dune.xyz/embeds/494498/936606/45ecb4c2-d49b-45e9-b907-1cc6452e647b?display=iframe
So there were plenty of people jumping on the newly released v3, however the graph below for volume shows that the switch from v2 to v3 was not as drastic as v1 to v2 was. There are still features that keep users on v2.
https://dune.xyz/embeds/494511/936636/f1075193-9a8d-49b5-b19e-9b83844e3670?display=iframe
Volume on v2 is still higher than 10% in 2022, although somewhat on a downtrend:
https://dune.xyz/embeds/494512/936634/1adcbc3b-7730-45ee-bdee-3df7fc04efa0?display=iframe
The usage and popularity of v2 becomes even more apparent when comparing the amount of swaps of v2 and v3, settling in between 60%-70% of the combined total after the first 4 months.
https://dune.xyz/embeds/494521/936652/7e072ecb-0b82-4b51-8663-f72d5c56305b?display=iframe
In fact, v2 is to this day still doing more transactions per day than v3, a share of 63% over the total of 2022, and has not moved much after the launch of v3 in 2021 and now.
https://dune.xyz/embeds/494524/936647/b08fffb1-ff29-40e0-9e96-2c80724790c2?display=iframe
Furthermore in 2022, v2 had a total of 424400 unique wallets that made a swap, while v3 had only 263263 wallets that made a swap. This is also apparent when looking at the user behaviour between the two versions, the average amount of swaps between v2 and v3 does not differ much. V2 has an average 9.35 swaps per wallet in 2022, and v3 has 8.69 swaps per wallet in 2022. The difference between total swaps between v2 and v3 can be explained by more users, not that those users are more active.
So v3 is considerably more popular by volume, but not by amount of swaps. This is also highlighted when looking at the average and median USD value of a transaction. The average value of a transaction on v2 is 5800 dollar in 2022, while in v3 this is 53000 dollar. Almost 10(!) times higher. The median value of a swap in 2022 for v2 is ~$1000, the median for v3 is $5500. These average and medium also display that high end users (whales) who swap larger values are more active on v3 as the pools have lower fee tiers, making trades cheaper.
When looking at number of pools, v2 also has the upper hand. In the last 7 days ~5000 pools had at least one swap on v2, on v3 on the other hand only ~1500 pools had a swap. Of course v2 is older, so might have had more pools created. However if we look at new pools created in 2022 we also see that v2 is dominating:
https://dune.xyz/embeds/494531/936658/310e4f57-ca12-4880-b423-f8370b04f808?display=iframe
The trading of low volume “exotic” coins are a main driver for using v2. The difference maker for v3 in volume are the top pools. The top pool on v3 is USDC-ETH 0.05% which did $55.69b volume in 2022, which is more than double(!) of total v2 volume ($22.96b in 2022). The USDC-ETH pool is on v2 also the biggest pair by volume but only did $4.52b over that period. The v2 USDC-ETH pool ranks only 9th on the combined volume list. The reason seems rather obvious that traders prefer the lower fee tier when making swaps. This is also noticeable when looking at the v3 fee tier volume for 2022, where 2/3 of the volume is made up of the lower fee tiers (0.01% and 0.05%):
https://dune.xyz/embeds/494532/936661/9255339c-34ba-4cab-9c2d-245a0c3aaa5f?display=iframe
Fragmentation
A risk with the launch of v3 was the fragmentation of liquidity pools. Not only can a pool for two tokens exist on both v2 as v3, it is even possible that on v3 4 pools exist (0.01%, 0.05%, 0.03% and 1% pools). Five pools for the same token just on Uniswap is not very capital efficient, however in the table below we can see that this is not for too many pools, mainly the big coins/stables.
Another interesting metric is looking at the volume for pools on v3, v2 or common on both. The common pairs, mainly on v3, are dominating this chart, as popular traded pools by volume will be made on both v2 and v3.
https://dune.xyz/embeds/494534/936664/860d9add-b524-46d7-bb13-383575f9adea?display=iframe
The graph below shows the number of swaps on v3, v2 or common pools. Here it’s interesting to note that more than 40% of swaps are done on pairs existing only on v2:
https://dune.xyz/embeds/494534/936670/ea97d997-c827-43c1-847e-82c714b78c7d?display=iframe
A look at Sushiswap and whether the launch of v3 had an effect on the market share of Uniswap in comparison with Sushiswap
Sushiswap is a clone of version 2 of Uniswap and launched in August 2020, gaining significant market share in the Defi space. As Sushiswap is a clone of Uniswap, it is interesting to see if the release of v3, and thus the launch of new features had a market effect on the two. In April, the month leading up to the release of v3, Sushi market share compared to Uni was hovering at around 20%-25%. At the start of June, this dips below 20% and does not recover:
https://dune.xyz/embeds/494543/936694/435e28ba-7d28-42ef-bf95-c9510ff35625?display=iframe
The decline for Sushiswap become even more noticeable when looking at 2022. We can see that in 2022 Sushi’s market share keeps declining in comparison to Uni, although other factors besides the added features of v3 can have an influence on this.
https://dune.xyz/embeds/494542/936687/e0154915-3b14-45ae-bfa5-aaa0adc09f96?display=iframe
Conclusion
In this article we explored some of the differences of v2 and v3, and saw that they complement each other quite well. Version 3 has become the dominant version for high volume trading, as the lower fee tiers offer lower cost to the trader. Version 2 remains the more popular version when looking at total number of swaps and users, and is mainly being used for the lower volume liquidity pools.