We’re currently voting on a change to the LP rewards discussed in this thread. When revising reward rates a frequent concern is the sudden jumps and the concerns that these jumps are too significant. I agree and want to add a bit of context why this was built this way originally. The answer is quite simple, we built the simplest rewards mechanism possible when we originally designed this. At the time this was a simple fixed rate that would be updated continuously with governance to ensure that rewards are lowered as TVL grew. It is not however how most DeFi projects designed their rewards mechanisms and it’s probably time for Centrifuge to follow suit and modify the mechanism.
To better illustrate the differences I’ve created a graph with example numbers to illustrate how we’ve been going about it.
The current formula for rewards is quite simple:
Total Rewards = TVL * Reward Rate
And the parameter we update via governance is the
Reward Rate. This leads to the steps in the rate every time the community votes on a reward rate update. It requires us to either frequently update the rate or try to predict TVL growth to know how to set the rate. If growth is faster than expected the issuance goes above the optimal rate and the protocol thus overspends:
If we changed the model to instead setting a global daily reward that is then split across all users we would solve the problem of having to adjust the rate depending on TVL and the reward rate would go down gradually as TVL grew.
Reward Rate = Total Rewards / TVL
I believe this is a more reasonable model and would suggest to the community that we go ahead and implement this along with the changes in the ongoing poll. If the feedback is positive I would follow up with a poll soon.