In the last weeks we noticed an increased demand in cUSD and cEUR that caused cUSD and cEUR to range above the price of 1 USD and EUR, respectively. In order to respond we have two parameters, that can be adjusted: spread / fee and reserve fraction. The spread / fee parameter was adjusted in October via CGP 40 .
In a second CGP 44 it is now planned to increase the bucket size of Mento by increasing the reserve fraction. An increased bucket size will result in a decreased slippage.
Intuitively, the relation between the reserve fraction / bucket size and the slippage is the following: The less you hold of an asset the higher your price will be if someone likes to buy from you. To give an example with numbers: Currently the Celo bucket size of Mento (cUSD) with a reserve fraction 0.5% is approximately 550000 Celo. Buying 4000 cUSD would result in 0.25% fees and 0.12% slippage, in total 0.37%. By increasing the reserve fraction to 1% the bucket size would increase to 1.1m Celo. Buying 4000 cUSD would result in 0.25% fees and 0.06% slippage, in total 0.31%.
USD/cUSD price from coinmarketcap
EUR/cEUR price from coinmarketcap
In the last two months cUSD and cEUR are faced with an increased demand. As a consequence the prices are ranging above 1. From the charts it can be observed, that cUSD/USD and cEUR/EUR prices tend to stay above 1.005. The reason for this threshold is, that below this value the accumulated fees + slippage are higher than the profit that can be made by trading the arbitrage cycle.
To give an example, with the current Mento fee and reserve bucket size buying 4000 cUSD would result in ~0.37 % in fees + slippage, with the larger part being due to the Mento fee. We did not choose 4000 cUSD randomly. It is approximately the average cUSD amount minted per bucket update based on the cUSD supply of approx. 54mn cUSD between 20.09.21 and 04.11.2021.
A full arbitrage cycle would look the following:
The total fee + slippage will be around 0.5-0.75%
Below is an illustration of the net demand / net cUSD outflow per bucket update. We have chosen rolling non-overlapping 6 hour windows and calculated different percentiles for the sample of net demands. Per 6 hour interval we ~6 x 12 =72 samples if we are assuming a bucket update to happen every 5 mins.
<aside> ❓ Why have we done this analysis: We wanted to understand whether there is a demand threshold per bucket update at which trading is less likely to happen. Ideally, after a parameter change we would be able to see an increase of the threshold.
</aside>