Every transaction on Ethereum needs to pay for gas. In the last year as demand for block space has ballooned, so too has the gas price. This has put pressure on consumers to decide whether or not to transact given the added cost.
In this analysis, we take a look at some of the broad impact gas has on Uniswap activity today as well as look back at a recent example of trader behavior when gas spikes.
Sections:
The average value per swap was flat between $13-14k for the bottom for quintiles of gas prices and then jumped over 2x to $31k for the top quintile. This suggests that small fluctuations in gas do not lead to the exclusion of smaller swaps until fees reach a higher threshold.
Median gas fees as a percentage of swap volume are roughly 1-2% across gas fee quintiles. The 75th-percentile however sees much more of a spike, going from 2.5% to 9%. Smaller traders who are still trying to get transactions done at higher gas prices are hemorrhaging more of their capital.
We looked at the impact of gas fees by day of week, hour of day, and minute of hour to see how traders might shift their behavior to account changes in cost. We didn’t find any interesting insights when looking at minutes of the hour so we focused on the two other timeframes.
Gas fees spiked on Thursday and week days more generally, which coincided with the days with the highest share of dollar volume swapped. Weekend dollar volume share and gas fees were lower while the number of swaps remained roughly the same. This pattern suggests that large, potentially institutional, addresses are more active during the week, while the weekend consists of more small dollar value retail swaps. Volume per swap peaked at $30k on Thursday, but dropped to as low as $16k on Saturday. (see dashboard for details)
When looking at the data per hour, we see a similar trend in that dollar volume and fees appear to be positively correlated while swap volume is more consistent across time. Volume per swap peaked at $33k at hour 13, while unique traders swapped at hour 21 at 29k.
At 1 UTC on May 1, 2022 gas spiked to over 7,000 gwei average with a max of 79k gwei, one of the largest spike of all time. To analyze this time event we looked at 24 hours of data from 12 UTC 4/30 to 12 UTC 5/1.
We saw swaps on Uniswap grind to a halt from 100+ swaps per minute down to three to four as most all traders were priced out. We saw swaps rebound as soon as gas prices subsided although they remained well below pre-spike levels.
Two potential hypotheses:
When analyzing the same temporal gas price data but with quartile swap values overlaid we see some choppy results. In this period we see some massive spikes, with the 75-percentile reaching up over $400k in some minutes. This trend wasn’t consistent, however, with some minutes having a 75-pct of less than $30k. Part of the noise in this is that there were very few swaps so these percentiles were being defined by only a couple data points.
Swapping in and out of APE coin were some of the most popular trading pairs around this event. Surprisingly, even at the height of the gas spike a quarter of swaps were coming from other pairs. As further research, I’d like to better understand what the motivation was for traders to be aggressively trading in and out of APE around this NFT mint.
We broke addresses down into quintiles by number of swaps through the observed period (1: <=2; 2: <= 8; 3: <= 24; 4: <= 100; 5 > 100). The results were rather surprising with first quartile “new” addresses increasing their share of traders to close to 50% in the first couple hours of the gas spike. Quartiles 2-5 all gave up share. Then the “power users” in the 5th quartile bounced back to account for close to 50% of traders once gas fees subsided.
Uniswaps dominance reached 92% at the height of the gas price spike, with Sushi making up the balance. Effectively the entire long tail of DEXs dried up as customers were being priced out.
For the full dashboard: Uniswap Gas Dashboard