🎁 Wrapping Up

We have looked at Uniswap V3 fees over the period of a month and seen that the fee volatility has increased over this period. We saw that the day-to-day change in fees seemed to be increasing over the month. We also discovered that fees were higher on weekdays compared with weekends, and this was a source of our volatility. Finally, we put our statistician hat on and looked at a measure of volatility called Z-score. The Z-score showed that volatility increased over the later part of July, confirming our earlier observations.

🤓 Measuring Volatility

We can look at a graph & say "yeah, the volatility looks like it's increasing", but this isn't the same as measuring it directly. So how do we measure volatility? In statistics there is a volatility measure known as a Z-score. This is a reasonably intuitive measure (making it rare in the field of statistics) which simply shows how many standard deviations each point is away from the mean of all the measurements. The formula is as follows:

• Z-score = (measurement - mean) / stddev

Where the measurement is the point in question, the mean is the average of all the data points, and the stddev is the standard deviation of all the measurements. You'll remember from school that for a normal distribution (bell curve) ~68% of measurements are within 1 standard deviation of the mean, and ~95% are within 2 standard deviations. We have plotted the daily Z-score of the Uniswap fees to get an objective view of the volatility, and colour coded measurements within 1 stddev, 2 stddevs and more than 2 stddevs of the the mean. Now we can see our suspicion that volatility was increasing later in the month was well founded. A handful of scores exceeded 1 standard deviation from the mean starting on July 24, and July 30 saw more than 3 standard deviations from the mean. This is objective evidence of increased volatility in Uniswap V3 fees towards the end of the month of July.

⛈️ Seasonal Factors

The ups & downs in the chart above look a little repetitive, leading to a hunch that there may be a time base factor in here. If we look at the average fees on weekdays versus weekends, we see 17% higher fee revenue on weekdays. It seems there is a day of the week impact to Uniswap fees and this is one factor driving day to day volatility. Are people trading on Uniswap for work, or trading on Uniswap at work? 🤔

Δ Change in Fees Day-to-Day

One measure of volatility of Uniswap fees is the change in fees from day to day. The graph below displays this, showing how much the fees increased or decreased relative to to the previous days total. Here we see differences of less than 1M for the first part of the chart, but later in the chart there are jumps of greater than 2M happening.

🦄 Uniswap Fees - What are they all about?

Uniswap is a decentralised token exchange on the Ethereum network. It pioneered the concept of the Automated Market Marker when it launched in 2018. Traditional crypto exchanges used offline order books to manage swaps - bids & asks and all that. Uniswap all changed that with a decentralised, permissionless & automated on-chain exchange mechanism. The concept has been copied & improved over many iterations, and is now the accepted standard for token exchanges. In March 2021, Uniswap launched V3 of their protocol. V3 had major improvements in capital efficiency, allowing liquidity providers to focus their liquidity around price points that maximise fees.

Uniswap generates revenues for the liquidity providers by charging a percentage of each swap transaction as fees. This percentage is either 0.05%, 0.3% or 1%, depending on the pool type. These fees can be calculated, in USD terms, from the very helpful Flipside Crypto Uniswap V3 tables. The graph below shows the USD fees generated across all liquidity pools in the Uniswap V3 protocol for a period of 1 month. You'll notice there are some big numbers in there - Uniswap V3 is generating over USD 1m per day in fee revenue. You'll also notice that the fees can vary significantly from day-to-day. It's this volatility that we'll look at it more detail.