PIP - YPIE Yearn Pie

pseudoPete · February 9, 2021, 2:42pm

Hi all! Working to make my first index for PieDAO. I was told that anyone can be the manager for a new Pie, so someone mentioned an Andre Index on Twitter so I thought I’d try to put something together and we can discuss the best way to do it.

Here’s a link to the google doc if you want to comment.
And here’s a link to the google sheet.

Overview: This doc is exploring what it would look like to have an “Andre Index,” a basket of tokens that somehow track the success of projects by Andre.

Background: Mike Demarais said “i feel like @PieDAO_DeFi or @indexcoop should create an “andre index”” on twitter (link), and PieDAO responded that anyone can manage a new pie

Considerations:

Definitions: What counts as an Andrew project? Things he’s launched publicly (YFI, KP3R), things he’s pseudolaunched (EMM), things he’s advised on (FTM), things he’s shilled (lots of stuff)? Unclear.
Timing: When is the right time to add something to the index? It is obviously helpful to be early but don’t want to get rekt. There needs to be some delay between contract live and inclusion in the index. Perhaps 7 days (can be shorter than DPI’s 30d limit because of trust in Andre).
Weighting: I think this should be some combination of: market cap, time in existence, sentiment, and category (ie founder, advisor, shiller).
Method: It’s also weird because he’s still experimenting with distribution. YFI was liquidity mined. KP3R was seeded in a UNI pool and given to those who do work. Who knows how upcoming projects (like Options) will be distributed

Proposal:

See this chart for a draft of how to weight the index. Note that it is not a complete list. I still need to do more research.
Need to figure out how to do the weighting… open to insights here! It’s hard to come up with something that doesn’t make it just like 90%+ YFI
I also don’t know how things get added, need to ask in the proposal channel on discord.

gabo · November 22, 2020, 7:53pm

Love the proposal! here’s a spreadsheet that can be used for further modeling

The model includes a max allocation applied to each underlying asset, that can be adjusted as per need (now =50%).

Furtermore, we could also adopt the sentiment score that was previously introduced in our DeFi indices (DEFI+L and rebalanced version of DEFI+S - wip), to weight in for all underlying assets the community shared sentiment according to the 3 following pillars:

Innovation (Solution & Roadmap/Timeline)
Functionality to the DeFi Ecosystem (Uniqueness & Composability)
Growth Potential (P/E Ratio, Volume & Outstanding Supply)

Each of the above items was graded on a scale from 1 to 12, and the resulting Sentiment Score for each asset contributed to the definition of a Sentiment Weight.

The final allocation of underlying assets would then be calculated as below:

Computing for each underlying asset the 30 Days Average of its Circulating Market Cap.
Using the 30d MCap to weight in each asset allocation, applying a x% Cap to their allocation (exceeding allocation iteratively redistributed among other assets), getting an Adjusted Market Weight
Calculating the Overall Weight as weighted average (weight tbd, usually 20%) among the Adj. Market Weight and the Sentiment Weight calculated from the Sentiment Score.
Computing the number of #tokens for each asset considering its live price (Coingecko price feed), and its resulting Overall Weight allocation.

masonnystrom · November 29, 2020, 11:45pm

While an Andre Index sounds interesting, a Yearn index would probably make more sense as everything that Andre does has some relationship to Yearn. A Yearn index could combine everything from $YFI to new recent partnerships/mergers including Cream, Cover, Pickle, etc. Finally, this would help provide an investment vehicle to get exposure to everything from the ever-increasing Yearn ecosystem.

pseudoPete · December 2, 2020, 5:41am

Wow that model is awesome Gabo! Digging in now… lots to process.

I like this idea masonnystrom – and coingecko actually just came out with a Yearn list that we could use as our baseline for which cryptos go into the Pie (I have been loathe to put more effort into this b/c I didn’t know how we would keep up, but using Coingecko as a crutch would help with the research!)

That said, it would miss something like KP3R (until it partners with Yearn I guess).

Also, what about things that are essential to Yearn vaults? Like 3CRV or yUSD? Would that have a place in the yearn pie in your opinion?

gabo · December 2, 2020, 2:52pm

Yeah I think the Coingecko list represents a cool baseline for this index, except for the current inclusion of CVP which is not really connected to the others except through meta-governance (same as PieDAO…).
I would swap that with KP3R. Gonna update the model accordingly to serve as a base for further discussions.

pseudoPete · December 2, 2020, 3:26pm

Thanks! Agreed, I think this list is now basically good enough to launch once we get the weights down.

Are there some tabs missing in the model? There are a couple items with #REF! errors, including on the sentiment score pieces, which I think keeps it from fully calculating.

Lui · December 3, 2020, 8:49am

The model looks great guys, the latest version of the weights seems sensible and YFI is not dominating too strongly.

In terms of next steps, how are pies usually turned into a reality from this point?

Cryptouf · December 4, 2020, 12:31am

are you also planning to add shishi ,pickle and co ?

williambrealey · December 4, 2020, 5:40am

As we’ve seen a lot of discussion and buzz about this Pie on discord I’d give it my strong support, and push for YPIE ticker (as long as we get the all clear from the Yearn guys, don’t want to seem like this is a publicity pie)

EDITED:
As I understand it would include:
YFI
SUSHI
COVER
K3PR
CREAM
PICKLE
AKRO

I think the weighting methodology proposed from Gabo is looking really attractive.

Also important to add that the pie will allow metagovernance, giving those interested in governing the Yearn ecosystem an easy way to get involved and raising participation.

gabo · December 4, 2020, 4:49am

Here some additional details on the proposed index allocation, as you may find in full on the spreadsheet

The current max allocation proposed (cap) is 40%.

One of the reasons to set a high cap is to provide higher scalability to the Pie also with higher liquidity levels. According to this scenario also the smaller allocations won’t likely represent a bottleneck even supposing the Pie would scale to $100m liquidity.

From an extended risk analysis performed on the period Nov 20th to Dec 4th (i.e. COVER data not available before), it’s evident how the majority of assets presented a very-low to low cross-correlation, while the remainings had a mid-to-low correlation (with the sole exception of the pair COVER-KP3R, scoring 0.8). The uncorrelation of assets is something very sought-after to minimize the portfolio risk, and not so common among crypto-portfolios.

Screen Shot 2020-12-04 at 05.10.46

On the other hand, over the (very short) period under analysis, most assets scored a sensibly high standard deviation. Only projects like YFI, HEGIC & SUSHI scored lower values of standard Deviation among the sample, indicating a lower price variance.

Below a visualization of the % risk contribution for each asset of the index, compared to their % allocation.
Risk Contribution (%) (4) Allocation (%) (3)

Some risk concentrations can be found around YFI & SUSHI, given both their higher allocation and higher returns scored by SUSHI in the period under analysis.

This picture shows backtesting of the index over the past 15 days, well summarizing the high volatility that occurred during such period.
Returns (Nov 20th - Dec 4th)