Proposal: the yETH-YFI-BULL strategy

Proposals Other Products (Labs)
41

@mattdw

Are you suggesting that this vault would act as an arbitrageur?

The vault knows what its total value in USD should be if it were just hodling the two assets. It has access to ETHUSD value through the various Oracles, and it can also deduce YFIUSD ticker as well.

I still think it’s a poor decision to create a new vault to cover the ass of a different one. If one vault bears all the risk so that another vault can reap rewards, why would anyone choose to put their ETH into the risky vault for weaker performance?

Remember there is no risk to yETH-YFI-BULL depositors at all, because rescuing CDP ETH doesn’t incur any loss (except opportunity cost for the duration of the black swan event). This of course assumes the size of yETH-YFI-BULL is so large that depositing its ETH to the CDP decreases the liquidation price massively.

Suppose for example yETH has 10 ETHs at yETH’s CDP and it’s currently at risk of liquidation because there is a Dai liquidity crunch. If yETH-YFI-BULL transfers 10, 20, or 30 ETHs to the CDP, that would push down the liquidation price by 50%, 66%, or 75% respectively.

Compare that to the yETH-YFI-BULL vault which would have (at current rates) have a 30% RoI from swap fees and all the risk of the yETH vault.

Recall that yETH-YFI-BULL is also generating yield from the ETHs/YFIs that is outside of Uniswap. In response to IL concerns I said we can increase that portion to 50% of the vault, which will still be earning yield as well through yETH and Aave or any other attractive ones in the future.

The only way I can imagine that this would be mitigated would be if we took a portion of the yETH vault that would be liquidated in the event of a black swan and gave that to the y-ETH-YFI-BULL market makers.

Recall that yETH can take of itself as long as there is no Dai liquidity crunch. It still has the plumbing to pay back its debt if necessary provided that there is no Dai liquidity crunch.

What if instead of an ETH:DAI pool in the yETH vault, this exact strategy as you wrote it is instead just rolled into that yETH vault?

It’s possible, although having the two strategies separate caters to different portfolio preferences.

I can replace it with a prettier graphic later

Could you clarify what “ybETH” is?

@benjaminmbrown

Is it fair to ask that you clarify your For/Against with meaningful rationale?

For: It automates a portfolio that constantly sells ETH pumps into YFI and vice versa so as to accumulate more of both, it caters to those bullish long-term on both assets.
Against: impermanent loss, it’s a “yolo” strategy, frontrunning of re-balancing, no analysis of performance yet.

2 Likes
42

Good question, thanks! It was essentially my shorthand for “2nd strategy.” I’m working on clarifying it and making it prettier in Photoshop. However, the more I work on my strategy it seems clear to me that it serves a different ethos than your proposal, so I’m going to draft up my own forum post sometime tomorrow.

Thanks. This wasn’t in your original proposal, so I forgot you added it in the update. Now that a portion of the fund is held outside the Uniswap pool a lot of my concerns have been negated, though I still think that this vault will not be able to beat out a dedicated arb bot to go back and fix the impermanent loss attained by the Uniswap pool - not that I necessarily think that’s a problem.

I tend to view Uniswap pools akin to hedging of bets. By staking equal value on both sides, I mitigate my losses (but also reduce gains). I would think that this BULL strategy would operate similarly, but with extra flexibility - I’d achieve higher gains with your vault strat then just putting funds into YFI:ETH Uniswap on my own.

Any chance you can create a quick mockup of rough APY numbers?

1 Like
43

I still think that this vault will not be able to beat out a dedicated arb bot to go back and fix the impermanent loss attained by the Uniswap pool

Totally agree. But the strategy is for the less fortunate who don’t have the time and/or expertise to express this particular portfolio preference. The motivation here is that there may be a lot of people out there trying to time bottoms and tops of the YFIETH pair, and this may be an option for them to allocate some of their portfolio into (saving a lot on fomo/fear energy and a lot of gas costs).

Any chance you can create a quick mockup of rough APY numbers?

Yes this what I want to do: a backtest against the YFIETH Uniswap pool since inception, with some average assumption around the yield on the outside portion from Aave/yETH vaults. Of course historical results are not indication of future retruns, but at least we’d get an idea. If anyone is interesting in contributing here and getting co-authorship to the strategy, please go for it!

1 Like
44

Given then UNI subsidy (and perpetual inflation), I don’t think it’s too crazy to just test-in-prod this, even as we do analysis to optimize AMM vaults generally.

The USDx/Dai will come from borrowing against YFI on Aave.

Screen Shot 2020-09-18 at 10.24.14 AM
Screen Shot 2020-09-18 at 10.24.14 AM1536×954 79.8 KB

2 Likes
45

Any more movement on this? Have you spoken with the team about what their plan is for the next yYFI strategy?

Farming one of these pools would likely be decently profitable (barring ETH crashing in price), but this is my only pause for why it might not be considered a viable vault strategy– if ETH crashes, then you may end up with less YFI than you started with to pay off your debt…

1 Like
46

I’ll create a poll in the original post.

Im not sure what you mean by “debt”.

yYFI is still WIP, Andre is doing cost/risk analysis last I checked.

47

Beautifully written, however, it is worth noting that the rebalancing issue is not discussed in detail, and it comes back to impermanent losses. YFI always moves in terms of magnitude (has so far) and with ETH left behind, you’re giving up on the potential growth of your portfolio, merely for the balancing issue proposed here.

For this very reason, I am AGAINST this strategy.

1 Like
48

I was assuming that you were solely borrowing against YFI for your ETH and your DAI/USDx– but now I’m realizing this was still related to the ETH and YFI joint vault, not just focusing on YFI only– so the ETH comes in directly, and the stable coins are borrowed against the YFI, correct?

In this case the only small risk would be IL– if the price of ETH tanked, your LP would become ETH-heavy, and you would potentially need to sell off some of the ETH to repay the borrow against the YFI (if you needed to exit).

49

Yes.

No, the opposite actually. When ETH tanks the vault unwinds some of YFI and buys more ETH with it. The “unwinding” involves paying back some of the stablecoin debt to Aave so that a safe c-ratio (collaterlization ration) is always maintained. c-ratio maintenance is always checked with or without re-balancing (because, e.g., both ETH and YFI go down, so rebalancing is not needed, but de-leveraging is)

50

Yes this strategy is not for someone who is good at timing tops and bottoms of the YFIETH ticker. Keep in mind, if you think there is a catalyst for ETH or YFI to pump or dump massively, you could always withdraw ahead and ride that catalyst.

You could also only use this strategy as a hedge against your active top-bottom-timing trading, so you allocate to it, say, only 10% of your YFI and ETH.

Here is an example whale who is long two assets and just rebalancing the expensive one into the cheaper one contineously.

1 Like
51

Let’s poll this. If you’re not familiar with this strategy check the original post and the followup.

TLDR of the yETH-YFI-BULL strategy: a vault that accepts YFI and ETH deposits, 50% of funds are in the YFIETH Uniswap pool to collect fees and farm UNI, but also be exposed to IL. The other 50% is in Aave and yETH. YFI in Aave is leveraged the same way yETH leverages the ETH in MakerDao. The USD value of ETHs and YFIs are kept at 50:50% through incremental rebalancing: buying YFI with ETH when USD value of YFI AUM goes down and vice versa. In black swan event (Dai liquidity crunch) this vault moves collateral to yETH vault to avoid liquidation.

  • Yes, implement the yETH-YFI-BULL strategy
  • No, do not implement the yETH-YFI-BULL strategy

0 voters

1 Like
52

This strategy is not smart, we can do better.

No need to subject Yearn clients to IL.

Let’s focus on other non-directional, market-neutral strategies that provide consistent profits and superior risk-adjusted returns.

2 Likes
53

There are customers and opportunities for both approaches, which benefit the ecosystem, users, and the protocol. Important is to communicate the risks accordingly.

4 Likes
54

I agree that we should accommodate a wide variety of money-market, mutual-fund/ETF, and in due time even Renaissance Technologies type strategies (requires far lower tx fees to implement).

Impermanent Loss however is currently not risk-mitigated enough to meet the standards of the Yearn brand.

Set Protocol is currently deploying a UNI Farming strategy that could really come back and bite them in the arse.

1 Like
55

That’s a fair point. I’m advocating the have moderate risk-seeking strategies in the yearn economy. It might be smart to brand it separately (e.g. risky yearn or whichever branding bluekirby or other marketing experts come up with).

3 Likes
56

Longer term, the risk-adjusted return performances of Strategy Managers can be tied to their public reputations.

Impermanent loss is just way too unpredictable. As LPs, there are too many ways to lose and there is no guarantee that the UNI token price will be sufficient to make up for losses.

Statistical models (Monte Carlo analysis et al) could be generated as a case study to elucidate further the variability of probable investment results in IL strategies.

57

That may be true, but a Yearn Vault comes with a certain brand promise and that includes a promise of no IL. If this strategy would be subject to IL, that’s ok, we should just call it something other than a Vault and should explain the risks.

3 Likes
58

Agreed, the name Vaults inplies safety, so we absolutely need separate branding for any ‘vaults’ with IL
No idea what we could call it. MoonRooms?

1 Like
59

Launchpad. High probability of system failure.

2 Likes
60

Dragon’s Lair – pile of gold, but a chance you’ll get burned. Or maybe something boring like Xvaults.