Summary
This proposal suggests to change WETH pool parameters based on simulation results:
On Curve for pool 0x6e5492F8ea2370844EE098A56DD88e1717e4A9C2:
mid_fee 43000000 (0.43%) → 60000000 (0.6%)
out_fee 240000000 (2.4%) → 220000000 (2.2%)
fee_gamma 2300000000000000 (0.0023) → 1395000000000000 (0.001395)
On Yield Basis:
rate 317097919 (1% per year) → 513698630 (1.62% per year)
amm_fee 14000000000000000 (1.4%) → 15000000000000000 (1.5%)
As a result, TRD should be smaller on average while returns are earned in almost all market conditions, with a smaller pressure on crvUSD peg.
Current parameters of WETH pool
Currently, WETH pool uses a very low refuel rate - 0.5% per year (which corresponds to borrow rate 1% per year). This is made possible by fairly low amplification factor of the pool A=2.5. Also the pool has a dynamic fee ranging from 0.43% at minimum to 2.4% at maximum.
Some work on the cryptopool simulator, however, made this simulator much less noisy. And it appeared that these parameters aren’t necessarily always good.
What one should look at is growth of the pool’s value, ability of the pool to have price_scale keeping up with the current prices and imbalance factors (value_crvusd / value_pool).
Growth of the value looks rather good as it might seem:
But when we look at price_scale repegging, it’s not all that great:
The
price_scale (center of concentrated liquidity) tends to unpeg. This leads to large pool imbalances and high and prolonged temporary redemption discounts (TRD):
Clearly, we want the pool’s price_scale to follow the current prices better than that.
Methodology and simulations
First, do preliminary scans. We are looking for a global maximum of the difference between APR and refuel rate (boost) before working on removing IL.
This scan is done so that
mid_fee = out_fee at some (probably previously found) refuel rate. It’s done to get a sense where the fees and amplification factors A could be. Very high fees are not advisable b/c it would take too long for the pool to realize profits there, even if APR is seemingly high. Lowest fee (mid_fee) would probably appear around a bump on the graph around 0.6% (and we will find later that this is true).
We scan A vs refuel rate to get a sense of where A could be at the constant fee found previously (or one of those fees). We can see that A could be somewhere between 2 and 10 optimally.
Now, we pick some (large but not extreme) value of maximum pool fee (out_fee) - here we took 2.2% - and perform (boost_rate/fee_gamma) at different amplification coefficents A and mid_fee parameters. I’ll present several examples (you can find others in the repository).
Higher APR, higher liquidity concentrations, less smoothness
And taking the maximum here, we do releverage at that point which gives the following price/imbalance/growth graphs:
This configuration gives probably the highest APR on WETH, however it suffer from multi-months periods of the lack of pool value growth.
Lower APR but more steady growth, less concentrated liquidity
And graphs after releverage
One can see that growth is almost uniform which is preferred by both LPs and veYB holders.