Proposed Protocol Settings for Atlas Release

This proposal aims to introduce the RP community to some protocol settings that need to be established as part of the Atlas release. Below, we present the concepts and some recommended values for discussion. These initial proposed values are based on discussions in the #trading and #research channels over the past week and reflect the initial consensus for Knoshua, Valdorff, and Ken.

This post aims to initiate an official debate and formulate an RPIP to set these settings as part of the planned Atlas release.

  1. A smaller CollateralRate value (a) for the amount of ETH retained in the rETH contract. This parameter is entered as a percentage of the TVL attributed to rETH holders. This value acts as a liquidity pool for rETH holders to redeem their rETH tokens at their true value. All ETH rewards and MEV earnings attributed to rETH holders are returned to the rETH contract first, and any excess ETH is placed in the deposit pool (dp) for staking, which increases the yield for rETH holders. Currently, (a) is at 10% (~18,000 ETH), but we propose setting it to 1% (~1,800 ETH).

    Further justification.
    The MaximumDepositPoolSize serves as a liquidity pool for rETH holders to redeem their rETH tokens at their true value. All ETH rewards and MEV earnings attributed to rETH holders are returned to the rETH contract first, and any excess ETH is placed in the deposit pool (dp) for staking, which increases the yield for rETH holders. The value for (a), which is the maximum allowed in the rETH contract, can be kept relatively small to minimize the negative effects on APR while still serving its purpose of providing fair-oracle price redemption for both peg-protection mechanisms (the incentivized NO voluntary exit and the arb bot opportunities).

    There are two mechanisms that can use the ETH in the rETH contract to maintain the market peg of rETH:

    The NO voluntary exit mechanism adds ETH to the rETH contract balance, arbitrages the profit by purchasing rETH on Uniswap, and then redeems (burns) it for full-value in the rETH contract. Without the rETH collateral (a) value in the rETH contract, the exiting ETH would spill over into the dp. If, for example, there is a queue of NOs waiting to exit, they would not be able to access the ETH in the dp as it would be held or assigned to the minipools in the queue. This removes the incentive for the NOs to exit and restore the rETH peg.

    An arb bot can purchase rETH at discounted rates in the market and redeem it for true-oracle prices in the rETH contract. These arb bots, combined with frequent ETH deposits from minipool skimming and node distribution interactions, will help maintain the rETH peg.
  1. An increase to the MaximumDepositPoolSize value (b), expressed in ETH. This sets the maximum amount of ETH that can be unstaked in the deposit pool. ETH in the deposit pool reduces the APR for rETH holders, but having a too-low maximum limit can prevent large purchases of rETH and slow its growth rate. Currently, the maximum dp size is 5,000 ETH (5% of TVL), but it is currently full. We propose setting (b) to 18,000 ETH This value is currently ~10% of current TVL but was selected to target 5% of expected TVL in the months following Shapella.

    Further justification.
    Having too much ETH in the deposit pool or a maximum cap that is too large can reduce the APR for rETH holders. On the other hand, having a maximum dp value that is too low can prevent large purchases of rETH and potentially slow the growth rate of Rocket Pool overall. Most estimates suggest that there will be several solo node operators who want to convert their validators to RP minipools, as well as a large number of existing half16 minipools that want to convert to LEB8s. Shapella is also expected to attract more institutional investors who want to purchase liquid staking tokens for rewards, and there may be an increase in existing Ethereum holders who want to stake their ether in a liquid staking option. All of these factors are likely to increase the demand for rETH. Since LEB8s increases the liquid staking capacity of the Rocket Pool deposit pool by 3x, it is necessary to increase the deposit pool capacity to meet the expected demand for new minipools.
  1. The time (c) when to increase the dp maximum as determined above. ETH in the dp reduces the rETH APR. Still, we expect a significant increase in staking demand after Shapella, so the RP needs to capture this market opportunity and grow the protocol quickly during the post-withdrawal period. We propose that (c) be T-10 days before Shapella.

    Further justification.
    **A conservative limit of 10 days was recommended to minimize the negative impact on the annual APR of rETH holders from an increased deposit pool. Most estimates suggest that the demand for LEB minipools will be such that the deposit pool will be depleted rapidly in the days following Shapella and will continue to be depleted for some time (weeks to months afterward). The 10-day limit was chosen to give some time for those interested in rETH to mint the tokens ahead of the Shapella hard fork while still being limited enough not to have a negative effect. There is already 8,000 ETH in unclaimed beacon chain rewards that have not been withdrawn and redeposited into the deposit pool. The increased effect will be similar to that experienced by the non-withdrawable ETH rewards. It could be argued that we could increase the deposit pool substantially beyond this proposed amount without a significant impact on the APR, but discussions indicated that the proposed value might be sufficient to generate enough demand both before the fork and during the weeks after when the NO queue processes with a higher frequency due to the interest in forming LEB minipools.
    Further justifications:
    Knoshua’s justification for 18K Valdroff’s justification for 18K

  2. Whether the profit from arbitraging the price difference when the dp maximum is increased should be captured by the RP team and given to rETH holders. We recommend doing this (true).

    Further justification.
    When the maximum deposit is increased, there will be the ability to mint rETH from the deposit contract. rETH can currently be sold at premium market prices, generating a profit. This profit could be automatically captured by the team when they deploy the setting changes and be returned to all rETH holders by depositing the earned ETH into the rETH contract.

    If the team does not capture this profit, it will be stolen in the next block by an arb bot. Likey, a comparative staking service will be the block proposer, and they will receive some of the arb profit as MEV.

    Current estimates indicate 20 ETH in profit in the arb.

1/20/2023 Update:
RPIP-16 has been created based on this DAO Forum post. The next step in the governance process is an informal poll to gauge interest and move it forward for official snapshot voting. Please take a moment to participate in the sentiment poll.

  • Yes, I think we should vote on this RPIP
  • I mostly like this RPIP except…; comment below
  • I only agree with a portion of this RPIP; comment below
  • I don’t think we should change parameters at this time

0 voters

11 Likes

Points 1, 3 and 4 look good to me.

Regarding point 2, the MaximumDepositPoolSize increase, I wonder if the proposed setting of 18,000 ETH / 10% current TVL isn’t a bit too aggressive. If filled, this means rETH APR will experience significant extra drag for a while until demand stabilizes post-Shapella. This would reflect poorly on the advertised rETH APR on Rocket Pool’s own website and the new MetaMask integration - it’s already below our largest competitor as is.

I do see the value of having some extra runway before Shapella hits, though. Plus, the current 5000 ETH cap is actually only 2,7% of current rETH TVL. So I’d propose raising the limit in multiple steps: first to ~8% current TVL (15,000 ETH) - a 3x increase of current capacity. Once we have more data on how NO / rETH demand develops post-Shapella we can always increase it further.

My very very rough guess is that 25-30k ETH would be ~5% after we’re through with migrations. That assumes a 3.5x growth between: LEB8, 16➡8 migrations, solo migrations, migrations from people that staked on CEX and became savvier later.

I expect just 16➡8 migrations to be a 2.5x or so, since it’s a 3x for folks that migrate and have enough collateral, and that could be done with half of minipools right now. I suspect some won’t migrate because unaware, but also expect some migrations from folks that do need to add collateral.

I’ll also note we have been trending roughly 2.3k ETH worth of rETH/week, and I expect that to ~double with Atlas to 4.5k ETH. That means if we miss by a little then we’ll “recover” from about 1k ETH of deposit pool overshoot per month assuming we’re aiming at 5% TVL.

Honestly - I don’t mind 15k (close enough), but I think we’re being conservative in any case.

1- Sorry, never heard about this collateral…

a) What is the contract address to this CollateralRate?
b) TVL of what?
c) Is this about LP on Balancer or Uniswap? If so, it’s earning fees from it?

2- If LEB8s increases capacity by 3x , being conservative , 3x5000k = 15 k is ok.

3- ok

4- Is it related to item 1 , c) above?

1: “Collateral” is one of my less favorite names in RP. It refers to how much ETH is in the rETH contract (or there + deposit pool). See https://etherscan.io/address/0xae78736cd615f374d3085123a210448e74fc6393#readContract.

The rate mentioned is just when ETH is allowed to be moved from the rETH contract (can only be used to redeem when burning rETH) to the deposit pool (first reserved for queued minipools and only then for rETH burning if there’s enough).

4: this is related to 1. When we create space, it’ll be possible to mint rETH at the oracle rate. If there’s a premium in any liquidity pools it can be arbitraged immediately.

Thanks @ken and @Valdorff

Points 1,3, and 4 also look good to me.

Regarding point 2 - I also think that the max deposit pool size should be increased in stages. I agree that the deposit pool does need increasing but I am very mindful of the rETH APR - especially considering we are being compared more and more. There is a lot of uncertainty around when node operators will migrate, how many solo stakers will sign up, etc. I feel it is better to respond to the TVL as it increases rather than try to preempt. I do agree that we should increase the DP before Atlas though I just think 18k and even 15k is quite high initially. One idea is that we agree TVL targets and DP amounts (based on target ratio) - that way we don’t have to vote and we can be responsive. As Pieter mentioned the current ratio is 2.7% of TVL it would be ideal to be lower than that to lift the rETH APR but I know there is a balance.

Bear in mind that what is likely to happen is the following:

  1. DP full
  2. Atlas released
  3. Loads of node operators migrate pools
  4. DP emptied and premium gone
  5. Arbs stop
  6. Minipool queue forms
  7. Max deposit size increases anyway due to minipool queue
2 Likes

Yes, this implies that there is no damage to rETH APR after 2 and no benefit to stages.

I think this makes it harder to reach consensus on a result. Could @Pieter or you instead suggest an alternative number for what we should set DP at Atlas release and then we can vote on that?

It’s not a given things will play out like this. After the initial wave of LEB8 conversions we might just as well sit at the DP limit again for a while, if rETH demand remains persistently high. My point is that we don’t know in advance, and a staged DP increase is a way to mitigate this risk. Especially since further increases to the DP limit can be done at any time without issue, but you can’t effectively reduce the DP limit if it’s filled to capacity.

IIRC, a target max DP size of 5% rETH TVL has been mentioned (informally) now and then. I don’t know if we need / want to formalize it into this RPIP right now, but if it’s still applicable as a rule of thumb that would put us around a 10k limit right now as a baseline.
As I said before, I think overshooting the baseline a bit in the upcoming period to capture more market share is defensible. It’s a tradeoff of being able to capture more peak rETH demand vs. the risk of sitting at a larger full DP, thus hurting rETH APR.

All this proposal is, is an attempt to preempt the initial wave of LEB8 conversions by a few days. I believe it is a very conservative attempt at that, which reflects the uncertainty you mention.
I believe this proposal has enough complexity as is and I would prefer to keep it focused on a single value adjustment. Can we get an alternative to the 18k suggested by us? Sizing and timing of additional stages as well as potentially formalizing the link between DP size and rETH TVL could be a different RPIP.

The CollateralValue that is part of this proposal represents a significantly bigger drag on rETH APR than the proposed MaximumDepositPoolSize. I also believe that arbitrage will have a bigger impact on rETH. So I think it’s important to not hold up these important changes over relatively small differences regarding the dp size.

Agree this is an important proposal that shouldn’t be held up for too long. Single value adjustment is fine by me for sure. I’ve proposed 15k in my initial post, which langers still mentioned as ‘quite high initially.’ So there’s definitely some difference in perception about the conservativeness of the increase.

How about the 10k limit I proposed in my second post then for now, if @langers is on board with that? Still a sizable x2 increase, and close to the current 5% max informal target.

1 Like

I believe 18k is conservative because currently rETH TVL is 187k ETH. With Shapella in about 2 months and some continued growth until that point we can expect say 210k ETH by then. Completely ignoring solo staker migration or new node operators attracted by <16 ETH minipools, we need 35% of existing minipools to convert to support rETH TVL of 360k ETH, which would mean that 5% of TVL and appropriate DP size is 18k.

Using the same logic for 15k, this would be appropriate for no solo stakers, no new NOs and 21% of existing minipools converting.

10k limit would likely be below the 5% target just from growth until Shapella with 0% existing minipools converting. Keep in mind that this increase is proposed to take place 10 days before Shapella, so most likely at some point in March.

Maybe @ken can add a forum poll with these options to gauge community sentiment?

3 Likes

Some independent math – just wanted to get a second take on this from @knoshua’s.

Grabbing data from https://dune.com/drworm/rocketpool.
From 10/03 (after the steep slopes from whale marriage were complete) to today, we’ve gone from 138.29k rETH to 178.45k rETH. In that same time the peg has gone from 1.0399 to 1.0551. That means rETH TVL has gone from 143.8k ETH to 188.3k ETH in 107 days or ~416 ETH per day. At a guess, we’re 50 days from Shapella, which would have us at 209.1k rETH TVL assuming the same rate (ignoring LEB8s being available for the last part of that period, and ignoring folks waiting for LEB8s to launch).

18k would be 8.6% - a bit higher drag than desireable, but nothing to write home about.

I have previously done the work to find how many minipools could move over without adding any extra RPL (grabbed data from rocketscan and processed it with a python script). As of 12/19, it was 15,302 new LEB8s, which represents up to 244,832 extra ETH [small error fixed 1/19]. I’m not redoing the work, but this will obviously have grown a ton - especially given RPL appreciation against ETH. In order to get to get to 5%, we need 360k rETH TVL, which means we only need about 62% of the minipools that could migrate without adding any investment . Again - this is very conservative because it’s based of data from 12/19 with RPL/ETH around 0.016 instead of today’s 0.021.

On top of that, this conservatively assumes:

  • Zero acceleration in rETH minting from LEB8s
  • Zero solo migrators
  • Zero migrators that add RPL to migrate
  • Zero folks with money that’s freed up from (coinbase/binance/stETH regaining peg/etc)
  • Not counting on the whale marriage that has been hinted at

I think 18k is a very safe bet. I haven’t seen any model for how we realistically end up with:

  • A full deposit pool
  • 18k being more than 5% of rETH TVL

If you think 18k is aggressive, please share a model showing something along those lines.

2 Likes

Thanks for elaborating @Valdorff and @Knoshua - these numbers are really helpful to give more context to the 18k DP limit and why it’s presented as a conservative increase. To me personally, this is convincing enough to go with it for now, even though I still have some doubts I’ll explain further on.
@Ken could you add links to Valdorff’s and Knoshua’s posts above in the MaximumDepositPoolSize justification section of the OP?

For the record, my reasons for bringing up the topic of the DP limit specifically were these:

  1. This thread received little engagement before, but historically the topic of the Deposit Pool size has gotten differing opinions in Discord. It’s better to have a solid case nailed down now rather than it springing up when the RPIP is brought to vote and there’s more time pressure involved. This discussion and the poll proposed by Knoshua definitely help here.
  2. I think the MetaMask Rocket Pool integration is a big deal, especially after Shapella. But it does bring a very direct comparison between RP and Lido - in APR, and the ability to stake with us in the first place (since MM only uses the DP directly at the moment.) So we need to strike a careful balance here, in this crucial phase after withdrawals become available and the staking landscape is reshuffled.

I didn’t really doubt that we can hit ~5% drag again sometime after Shapella with an 18k limit. I think an even higher limit than 18k is well within reach later on. It’s more about the timing of the increase. Hence why I mentioned a staged approach initially.

Consider the following scenario:

  • DP limit is raised 10 days before Shapella and fills up immediately.
  • Shapella hits and major CEXes are well-prepared, immediately making withdrawals available.
  • Lots of marketing / attention around this fact in the ecosystem.
  • The mass wave of RP LEB8 conversions hasn’t happened yet - it will take a few days to a week for all validators to skim and the team to enable migrations for existing pools.
  • In the mean time, CEX users that want to move to LSD’s see Rocket Pool with a DP that’s still full and with a rETH APR that’s even lower than it is currently.
  • Even if the actual APR ‘damage’ is short-lived / temporary, initial impressions matter, and many of these users could end up going for a competitor rather than RP.

Regarding the ‘aggressiveness’ of 18k, having seen the numbers I’ll admit that this impression was biased by the current low DP size (5k) and the existing low drag of 2.7%.
Keeping in mind that the 5% target is an ‘acceptable max’ rather than a direct target. Of course having lower drag is better, but as we’ve discussed this does come at a cost.

Final point - to be fair, a staged approach isn’t free either. It adds complexity and friction in needing another RPIP / vote at a later time. So overall, I’m now on board with the 18k as proposed.

Curious to hear @langers / the team’s thoughts on the latest discussions.
(btw, as an aside - would the team consider sharing their propietary router with MetaMask so MM users are never hard-blocked from staking with RP either?)

3 Likes

The models that you both (@Valdorff and @knoshua) have put forward make perfect sense.

My only concern was how long the migration will take - I simply don’t know which is why I was proposing raising DP, as the TVL increases to keep rETH APR as tight as possible.

I do think we need to front load the deposit pool though. Based on the fact, that we would only need 60% of the minipools that could migrate without adding any further funds or 35% of all minipools - makes me confident that the TVL increase will be quick.

I agree that 18k is reasonable.

FWIW I am just expressing my opinion, not making decrees.

5 Likes

Yes we are on that. Not sure if they can integrate it but we are talking to them.

1 Like

This was very helpful. I agree that it is very unlikely we see a full DP without substantial TVL growth.

1 Like

looks all pretty good to me.

I guess the DP limit has to be set and manually adjusted as an absolute value because a dynamic percentage of ETH TVL wouldnt work since adjustments require a smartcontract update?