Treasury portfolio - A suggestion for Diversification

I remember reading a few tweets saying that the Sushi community was looking at ways to diversify the treasury away from 100% Sushi. I’ve seen that you have been having some discussions.. I’ve been thinking about portfolios for long term holding for individuals / DAO’s and thought I would share an idea that I’ve previously proposed to the INDEXcoop as a potential product aimed at long term passive holders both individuals and DAOs.

The idea would be for Sushi to invest part of the treasury into a diversified portfolio with a market cap structure:

  • 50% ETH
  • 15% wBTC
  • 15% DPI
  • 15% yDAI
  • 5% Link

Component justification

  • ETH is really the core of crypto and the future of DeFi, so being focused on ETH makes sense for any initial position.
  • BTC is the largest cap with room to expand into CeFi. Some exposure is essential, but a straight market cap match of 60% seems pointless. Having a good chunk in BTC with monthly rebalances captures some of the benefits of the volatility vs ETH.
  • DPI DeFi is growing and a carefully curated set of projects has massive potential, and the use of monthly rebalances captures the volatility.
  • yDAI gives the portfolio a way to capture crypto pumps and buy the dumps from the other crypto assets. Use of yDAI provides some income compared to a stable coin.
  • Link part of the crypto and DeFi infrastructure and a key sector going forward. Its price movements are not highly correlated to ETH / BTC so the monthly rebalance captures this.

Backtest charts

There is not a lot of DeFi data to backtest, so I’ve looked at a backtest using ETH as a replacement for DPI:

ICP Backtest
ETH 50 65
wBTC 15 15
yDAI (2% pa) 15 15
DPI 15
Link 5 5

For the backtest calcs, I’ve assumed that cDAI would capture 2% interests pa on an ongoing basis.

Starting from 1st Jan 2018, I’ve calculated the value of a $10,000 portfolio with and without monthly rebalances. Then repeated the calculations at 3 monthly intervals (I can do monthly, but the chart would look a mess)

65:15:15:5 portfolio backtest 2018 to 2021

The key observation for me is that the rebalance (red or green) is better than the corresponding hold (Blue or yellow) over the longer time periods.

65:15:15:5 backtest 2019 to 2021

Monthly rebalancing with 5% Link is / was absolute rocket fuel under some conditions.

Obviously, this is cherry-picking a portfolio based on past performance. No one could have predicted that Link would do what it has done. The question is will it be correlated to the other components going forward.

If we look at the last 5 months, it doesn’t look so good (because we would be selling BTC and ETH as they pump)

Date Hold value Rebalance value Additional profit
01 Oct 2020 $10,000 $10,000 0.0%
01 Nov 2020 $10,966 $10,966 0.0%
01 Dec 2020 $16,024 $16,021 0.0%
01 Jan 2021 $19,472 $19,097 -1.9%
01 Feb 2021 $31,069 $29,415 -5.3%

However, If we include DPI at the 15% weight and reduce ETH to 50%, then the rebalance becomes loss-making as we sell more ETH for DPI.

Date Hold value Rebalance value Additional profit
01 Oct 2020 $10,000 $10,000 0.0%
01 Nov 2020 $10,286 $10,286 0.0%
01 Dec 2020 $15,021 $14,174 -5.6%
01 Jan 2021 $18,110 $16,117 -11.0%
01 Feb 2021 $25,575 $20,692 -19.1%

Correlation calculations

I’m not an expert on correlation calculations and what they mean. Obviously, we want to avoid 1.00’s, I’m not sure that we need -1 values. I’ve looked at the correlations over different periods:

Monthly correlation 01 Jan 2018 01 Jan 21
BTC ETH Link yDAI (2%)
BTC 1.00 0.74 -0.57 -0.16
ETH 1.00 -0.51 -0.01
Link 1.00 -0.01
cDAI 1.00

Monthly correlation 01 Jan 2019 to 01 jan 21
BTC ETH Link yDAI (2%)
BTC 1.00 0.79 -0.48 -0.16
ETH 1.00 -0.52 0.03
Link 1.00 -0.04
cDAI 1.00

Monthly correlation 01 Jan 2020 to 01 Jan 21
BTC ETH Link yDAI (2%)
BTC 1.00 0.77 -0.33 -0.20
ETH 1.00 -0.89 0.10
Link 1.00 -0.05
cDAI 1.00

Adding DPI from 01 Oct 20 (Determining correlations over 4 data points feels wrong to me, but I’ve not constructed a pseudo-DPI backtest.)

Monthly correlation 01 Jan 2020 to 01 Feb 21
BTC ETH Link DPI yDAI (2%)
BTC 1.00 0.34 0.20 -0.56 -0.14
ETH 1.00 -0.71 -0.86 0.34
Link 1.00 0.97 -0.18
DPI 1 0.67
cDAI 1.00

How to manage the fund

I can see three possible wats to implement and manage such a treasury

  1. Sushi use a standard multi sig to manage the funds - This is easy to set up, but requires ongoing management form Sushi
  2. Deploy a Set protocol smart contract that allows others to co-invest in the same strategy - This is more work to set up, and potentially manage, but the generation of the pool token may open up opportunities to use the fund capital as collateral in other protocols (Cream etc). This is similar to the approach YAM are using for their treasury.
  3. Invest the assets in a fund managed by others and effectively outsource the management of the treasury to all sushi to focus on other activities (As I said, I’ve already suggested this idea as a proposal for a future product to the INDEXcoop. However, without a large partner on board, they seem reluctant to work on such a diversified product).

Notes on construction

  • This is pretty close to my current portfolio so I’m obviously biased (I have less stable coins in crypto and I don’t rebalance due to tax implications)
  • I’ve looked at a few other mixes, but this one seems to hit a sweet spot. (and I like the symmetry)
  • There is not much back test data for DeFi, so it’s basically an assumption that DeFi will grow and not be a perfect correlation for ETH.
  • Link appears to add some uncorrelated return
  • We could use DAI as the stable coin, but I think there is enough liquidity in cDAI / others to allow easy issuance / redemption, and it makes adds some yield.
  • We could replace the yDAI with others (aDAI, cDAI…), but I’m not sure there is an easy way to identify the best candidate.
  • I considered going 5% cDAI, 5% aDAI and 5% yDAI. It would diversify the interest rate a little but adds complexity.

Potential changes we could consider:

  • Change stable coin (yUSD)
  • Replace (part of the) ETH with an income-generating ETH (based on staking / yETH)
  • Replace (part of the) wBTC with an income-generating wBTC.
  • Swap wBTC for a different ERC20 BTC.
  • Reduce BTC (if ETH flips BTC)

Use of income-generating forms of ETH, BTC etc probably make sense for a treasury as it adds more value (and more risk), but if you want to tokenise and make it available other, then it risks looking like a security.

I realise that this is a long, post, so thank you for getting to the end, and I welcome your thoughts.