Dynamic profit-sharing mechanism idea

Introduction

There has been some discussion in Discord about Vault’s risk-adjusted return a.k.a it’s risk/reward. I think there is room for improving the current profit-sharing mechanism (besides totally changing the equation)

To recap, the current formula is based on Gro’s documentation:

  1. If utilisation ration <= 80%, then: Vault total yields = Vault deposit yields + (30% + 3/8*Utilisation) * PWRD deposit yields
  2. Else: Vault total yields = Vault deposit yields + [60% + 2(Utilisation - 80%)] * PWRD deposit yields

When plotting Vault’s, PWRD’s yield and the amount of effective leverage Vault has we get this illustration:

Screen Shot 2021-11-21 at 23.01.41

You can check the calculator and various definitions here (make a copy)

The yield multiplier for Vault is based on dividing the above formulas by Vault deposit yield which gives:

  1. If utilisation ratio% <= 80%, then: Vault yield multiplier = 1 + (30% + 3/8*Utilisation) * Utilisation ratio
  2. Else: Vault yield multiplier = 1 + [60% + 2(Utilisation - 80%)] * Utilisation ratio

Essentially it is the multiplier of Vault’s yield against Vault’s base yield at a particular utilisation ratio

In fact, we can generalize the formula further by considering the minimum take, inflection point and utilisation multiplier:

  1. If utilisation ratio% <= inflection point, then: Vault yield multiplier = 1 + (Minimum take% + Utilisation multiplier * Utilisation) * Utilisation ratio
  2. Else:
    a. Define Inflection point's minimum take = Minimum take% + (Utilisation multiplier * Inflection point%)
    b. Vault yield multiplier = 1 + [Inflection point's minimum take% + (1 - Inflection point's minimum take%)/(1 - Inflection point%) * (Utilisation - Inflection point%)] * Utilisation ratio

Importance of Vault in Protocol’s Risk Management

I think it is important for the protocol to have a way to organically influence the utilisation ratio when needed. For example, the protocol may wish to target a lower utilisation ratio after an incident, like the recent CREAM incident, in order to manage risk. And as I see it, the most effective way to influence the utilisation ratio is through Vault.

There is a limitation to how much PWRD can be minted by the system as a function of the amount of TVL locked in Vault. At the same time, there is no clear incentive (currently at least) to make PWRD users exit their position since withdrawing PWRD leads to a lower utilisation ratio which leads to a higher yield. Essentially, organically adjusting the utilisation ratio by focusing on PWRD is going to be restrictive

So as of now the only realistic tool in the protocol’s belt is to (dis)incentivize the market such that the supply and demand Vault increases or decreases through adjustments on Vault’s yield. Of course, adjustments to Vault’s yield will lead to adjustments to PWRD’s yield on the opposite direction. So by focusing adjustments on Vault, we indirectly get the benefit of adjusting market participation on PWRD, but with Vault’s greater flexibility in TVL inflation/deflation

Yield entropy

To effectively change market participants (dis)interest in Vault we should look beyond high yield and also look at the risk-adjusted return. As we’ve experienced now, a leverage of 1.3x in Vault would result to a 1.3x loss (relative to what Vault without leverage would lose) in the event like CREAM hack. But a natural roadblock in optimizing for Vault’s yield is this idea of yield entropy which is an artifact of the profit-sharing mechanism with PWRD.

If you play with the calculator I linked above, you’ll notice that GVT’s yield multiplier is never really equal to the leverage (which is really just 1 + utilisation ratio). Ideally when you leverage up, your return multiplier is exactly multiplied by the amount of leverage you take i.e 1.5x leverage leads to 1.5x returns. But because of the profit-sharing between Vault and PWRD this is less possible, which leads to yield entropy for Vault (lost potential yield within the system for Vault)

Dynamic variable adjustments

But the savings grace for Vault is that the profit-sharing mechanism can be dynamically adjusted based on several variables as I mentioned above. Here is what the same graph looks like with minimum take = 0.3, inflection point = 50%, utilisation multiplier = 0.75

Screen Shot 2021-11-22 at 00.04.51

As we can see, Vault’s yield (returns) becomes more effective in contrast to the amount of leverage taken. As a side note, I’ve found the inflection point to be the most effective at changing the Vault’s yield effectiveness

Allowing a dynamic/variable profit-sharing mechanism would let the protocol be more nimble when adjustments do need to be changed. To achieve the greatest speed while also being transparent I think these kind of variable adjustments need not have to go through a DAO vote, but would still be good to be announced in the community (manually or by a bot). This approach is similar to Olympus’ approach in their BCV (bond control variable) adjustments (in fact they don’t even notify BCV changes I think)

I think a dynamic profit-sharing mechanism will be an effective tool for the protocol management and ensuring longevity going forward among other tools

4 Likes

Oh my original diagram had a glaring mistake on the PWRD yield part. The correct one is down below (nothing changed on the Vault side)

Screen Shot 2021-12-18 at 23.28.09

Thanks @Slacking - and thanks for the video chat we had to discuss.

To summarise for others benefit:

  • We want to first focus on increasing the overall yield
  • This will be primarily through allowing increased exposure to Curve strategies which have higher yield
  • The cost of this would be less stablecoin protection, but it seems this isn’t valued highly by the community as most PWRD is staked in the PWRD-3CRV pool.

I will share a post on this in the coming week. Then once this is executed we can revisit the yield sharing curve as well as potentially how the HODL bonus is shared.

1 Like