• Open Talk

  • How to design vaults with such high APYs that you cant pay for the liquidity

The way that the current model works is that someone else adds the liquidity required to then increase the borrowing limit afterwards. But with high APY vaults (e.g. OHM, we cant afford to pay people to provide the liquidity) there is not enough profits being generated to grow the liquidity on its own whilst keeping the collateralisation ratio "good enough".

Instead of the funds going into the transmuter to boost yield (which is already crazy as it is) any excess alAsset in the LPpool is removed and matched with the corresponding Asset and then that paired liquidity is added back into the LPpool, increasing the liquidity and more actively managing the peg. The pool would need to be seeded for this to work but I'm sure that the OHMies wouldn't mind donating some OHM to get it started for them.

Diagram:

for example; take a curve pool of alOHM and OHM and there is 60:40 alOHM😮HM, you could remove 20 alOHM and pair that with 20 OHM from the new transmuter and then add that back so now you have 60:60 in the pool, growing your liquidity and keeping the peg and this LP position is owned by the transmuter. The debt cap can be a set multiple of the amount of LP owned by the transmuter (to avoid manipulation) so it can dynamically increase over time. Because these function calls are not affected by price this can all be called and maintained by the users using the vault. for example if they want to borrow then they can also pay to draw down the funds and do the balancing (which is not affected by price).

This means that for high APY assets we can grow the pool ourselves a own the liquidity and take our 10% performance fee on top the high APY assets which we can then stake back into the vault. I've been looking over the V1 code and it seems do able whilst avoiding having to edit the base alchemist contract, just the vault router and the transmuter. I have no idea how long it would take to implement as I'm not sufficiently familiar with V1, but since OHM only has one main strategy that people would want to use any way, we could launch alOHM in a V1.

What do people think?

Ok, Just want to break this down to make sure I'm understanding properly.

  • Olympus provides Alchemix with OHM to seed the pool (50% transmuted to alOhm)
  • The alOHM/OHM pool is created with 50/50 balance
  • People sell their alOHM for OHM to (9,9) and either redeposit/fold in Alchemix or Olympus
  • This leads to the pool being a higher ratio of alOHM to OHM (60/40 in your example)
  • Enough alOHM is automatically removed from the pool to establish 50/50 balance (assuming Alchemix owns enough liquidity to do so)
  • instead of the OHM vault yield going to the transmuter - it is used to pair with any alOHM surplus and add back to the pool automatically

Am I understanding correctly?

Honestly it looks like quite a bit more work for both the Alchemix (changes to the transumter logic / automated pool balancing / etc) and Olympus (yield is paid via rebasing) devs. I haven't done any coding in yeeears but it seems like a more complex implementation than alBTC for example. I doubt there's any chance of this in v1 TBH.

The initial rush to the vault/pool would also put a lot of early sell pressure on alOHM (would need to have a super limited ceiling). Not insurmountable I guess.

I'm not sure how viable it would be in practice, I'll happily defer to others with more experience. Creative thinking though!

    hekmat In terms of the complexity:
    it requires a new vault to be plugged into and 2ndly it requires a new transmuter to be written be it V1 or V2 the amount of work stays relatively similar regardless. OHM only has 1 strategy that people would want to use anyway.

    But yes this is more theoretical as there are a lot more low hanging fruit that we can go after, mby its something that's revisited later on I suppose, post V2. I'm just looking at the high APYs and the TVL that we could get from it, and once the APYs drop we could point the vault to an old transmuter once we have a large ownership of the LP pool unless this is able to keep the peg, if we are able to keep a good peg with this and retain all of this value then why not keep it around?

    Yes you get the idea, so if we wanted to close the pool we could pay back the OHM to Olympus or keep it depending on what deal we struct. We would take ownership of the rest of the pool which at the rate of OHM emissions could be a sizeable amount if we where able to capture the high APYs. But this could be applied to any high APY strategy.

    Thanks for your feedback!

      12 days later

      Let me simplify one of the steps. Really what's happening is we provide a 50/50 LP pool where users sell alOHM to OHM, so effectively all of the yield generated by the OHM collateral should be sent to the LP pool - there's no need to remove alOHM and pair it with OHM to redeposit. Just send OHM to the LP as necessary to maintain the balance.

      Thinking out loud about the numbers behind it, let's assume ohm APY is 10k% and we start with a 25% LTV to minimize recursion:

      1. Say we start with 1k alOHM & OHM in the LP, debt cap 250 alOHM.
      2. Users deposit 1k OHM and withdraw 250 alOHM
      3. Users swap 250 alOHM for OHM, pool balance is now 750 OHM/1250 alOHM (37.5%/62.5%)
      4. After about 1 day, the 25% loan is paid off (yeesh that's quick) for all users. There is 250 new alOHM available to mint.
      5. The protocol owns 250 OHM that it can send to the staking contract, which balances the pool to 1000 OHM/1250 alOHM.
      6. 250 alOHM is available for minting so users mint it and sell it. Now at 1000 OHM/1500 alOHM, (40%/60%)
      7. Process repeats every day for 30 days - new 250 OHM for the protocol, new 250 alOHM minted and sold. Results in 8500 OHM/9000 alOHM so close to 50/50.
      8. Raise debt cap to 500 alOHM, same cycle happens. So on and so forth.

      So maybe this is sustainable! But what happens if everyone wants to withdraw their OHM collateral? I suppose if they've all been taking the loan and selling, then only 1k OHM would need to be removed from the LP to make those users whole.

      This seems super interesting... trying to think of what would actually be able to blow this up. Main thing seems to be keeping debt cap low and LTV low. Now I'm also wondering if this would also make sense for alUSD and alETH.

      Def agree with hekmat this is a v2 thing though. It's a whole new system - more work than alBTC, so it's not gonna happen before V2. Gives us time to keep thinking about how this system would break, if it can be broken. Will have to model out the recursion scenario next.

        ov3rkoalafied Yea it would work just like that, and would stay collateralised, the reason why the current transmuter isn't built like this is because of the boost feature it has, but with OHM getting such high APYs you don't really need to boost it, you can instead use the yield to generate a bigger LP pool for free, but eth and dai don't have any where high enough APYs to be able to create an LP pool unfortunately.

        If we can get seeded with a large enough Pool of OHM then we can really quickly get up to very high debt caps and LTV, we could even use a curve TWAP to automate debt increases as the liquidity cant be rugged. Because of how this process can be hard defined in the contract it could allow for 0 upkeep from the devs and the users could pay to draw down the alOHM from the vault and re-balance the pool when ever it wants.

        I do think that there is a huge area in creating custom vaults with special properties, mby even undercollateralized loans one day...

            25 days later

            ov3rkoalafied Thank you for the example, I think it makes it much more simple to try to understand the process.

            I am a bit late with the reply but I am completely new to governance, which also means that I might be missing crucial processes, so please correct me if I am wrong.

            IMO the flow would work the way you described, however I think that the math needs a bit of adjustment, which might make the process unsustainable this way.

            1. 10k% APY would translate to a rebase % of 0.45% per 8 hours (roughly), which would make a 25% LTV repayment time 16 days. I will get back to why this is important
            2. When users take the 250 alOHM loan, and swap the alOHM to OHM, that changes the composition of the LP to 750OHM/1250alOHM like you stated and whenever the 250 OHM is repaid from the yield on the deposits that changes the LP to 1000OHM/1250alOHM, also like you wrote.
              However, after this I think the only change that happens in the LP for each new cycle is that the amount of alOHM increases by 250 and OHM decreases by 250 (as the users take it out when swapping the alOHM to OHM), which is then replenished from the deposit yields to go back to a 1000OHM, but this results in the amount of OHM staying constantly 1000, while the amount of alOHM keeps increasing by chunks of 250, making the LP more and more unbalanced.
              This means that twice the amount of OHM (500) would need to be put in per cycle to keep the LP balanced.
            3. Because of this, the debt cap can only be raised once the deposits generated 500 OHM, but that is twice the amount of time that the debt of users is repaid. Which means that 50% of the time the users would need to keep their OHM deposits in Alchemix while not having any active loans. I don't think they would do this because of the treasury 10% cut on their rebases. Also, getting back to my first point, a cycle becomes 16 days, which is plenty of time, so users will withdraw their collateral to wait for the next debt cap raise, which will empty the vault.
            4. An additional thing to consider is that if the OHM that is generated from the deposit rebases is funneled into the LP position, then that is not generating OHM staking yield, making the yield for depositors even smaller, on top of the 10% treasury cut.

            I actually like your idea on the Olympus forums about Olympus chiming in and replenishing the OHM/alOHM LP position themselves, but then Alchemix would not own any part of that pool which is not the best, but that could work in theory I guess.

            Again, it is very likely that I am wrong with my math somewhere, I just can't figure out where, so pls let me know 🙂

              5 days later

              Biddls Replying to both you and @ov3rkoalafied Correction on my part, had a few more days to think about it, I still believe my math is correct, however my conclusion is not and this would work indeed.

              Reason being is that I assumed that the yield from the staked OHM would need to be able to increase the amount of OHM in the OHM/alOHM LP, but as I previously described, I don't think it would, it would just keep it stable at a 1000 using the given example, while alOHM would keep on increasing infinitely.

              However, maybe this is not an issue! Having the cap on the amount of alOHM loans, that keeps the amount of OHM deposited at a soft max cap, which means that the LP position does not necessarily need to grow as long as the debt cap is kept at the defined level. (Maybe the excess alOHM can periodically be withdrawn from the LP and burned/kept in the treasury/whatever to keep the LP position balanced)

              As an option, to keep being able to increase the debt cap slowly, the OHM that the treasury takes as the 10% cut, could be funneled into the LP position, slowly growing the LP and the possibility to increase the debt cap.
              This would, on the flip side, take away the treasury's option to stake the OHM that it makes through these 10% cuts, but probably an acceptable price to pay to allow the establishment of these high APY positions in ALCX.

              Am I correct?

                  4 days later
                  5 days later

                  Sorry for spamming, but had some more time to think 😃

                  I think in this case it might be possible to set up a different "stability" LP than what are used now for DAI and ETH and the alOHM/OHM that has been discussed.

                  So I think the pool could be alOHM/gOHM(or wsOHM) which is the staked version of OHM, meaning it gets the yield from the rebases. This way, the OHM side of the pool would constantly be growing and it would possibly not need any additions over time from the treasury or anyone else, and the LP position would still keep on growing, allowing for a constant debt cap raise. (the alOHM side is not staked of course, but that side constantly grows because of the loans taken anyways, keeping the LP more than balanced)
                  Or alternatively, this could keep the OHM side stable, but would increase the returns for OHM depositors in ALCX as the rebases from the deposited positions would not need to be funneled into the LP.

                  I believe this would still fulfill its main goal, which is the stabilization of alOHM, as the prices of staked and non-staked versions move in tandem always.

                  (In theory, maybe it would be possible to do this with the alETH/ETH(->stETH) pool as well, to provide better returns for LPs and incentivize this way with less bribes in crv/cvx)

                  This would allow for ALCX to manage these high yielding positions itself, like the original post outlined.

                  This seems too good to be true if I am being honest, but again, not sure what I am missing.

                  Write a Reply...