Introducing PMI : the Post-Modern Portfolio Theory Index

Dear all,

We are very excited to submit a new methodology for an index product aiming to help non-professional investors - experienced as well as beginners - to automatically manage & rebalance their altcoin portfolio by optimizing its risk-adjusted returns.

Going by project name PMI (the Post-Modern Portfolio Theory Index), it has the opportunity to become the 1st index in crypto to be non-sector specific and optimize the downside volatility of its underlying assets.

An immense thank you to @Lavi & @verto0912 for respectively helping us building (with no intention to be included in the methodologist team) and finally reviewing this proposal.

1/ The Problem

When looking for diversification into altcoins, building a portfolio with long-lasting performance while minimizing its downside deviation is not immediate for all crypto investors.

Because getting your head around risk management and volatility is such a steep learning curve, placing bids on random projects based on external advice or crowd sentiment effectively feels more compelling than spending time and effort into fundamental research or data analysis.

This is where quantitative and rebalancing techniques get interesting : they can be combined to target the maximum growth while minimizing drawdown when price action becomes bearish - exactly like a professional trader or portfolio manager.

2/ The Product

The key performance indicator used to build PMI differentiates harmful volatility from total overall volatility by using the Sortino ratio, a key element of the Post-Modern Portfolio Theory (see this article for a detailed mathematical definition)).

The Sortino ratio is widely used in TradFi - yet passively - to track a portfolioā€™s risk-adjusted return. It has established itself as a more accurate indicator than its Modern Portfolio Theory counterpart, the Sharpe ratio.

Of course, being a good indicator of past risk-adjusted returns doesnā€™t mean that the Sortino ratio on its own can give an accurate prediction of any assetā€™s future performance. However, having tracked the Sortino ratio of a significant amount of tokens over more than a year, we were able to incorporate this metric into the following methodology.

2.1 - How it works :

PMI is based on the ranking of the best eligible tokens by annualized Sortino ratio, thus accounting for the time span difference over which performance is measured, and providing a projection of risk-adjusted return over a sensible period of time. The components are weighted in proportion of their respective score ratio to the overall.

2.2 - Market Opportunity :

PMI is a performance index targeted towards retail customers looking for diversification outside of Bitcoin, Ethereum or sector-specific indexes like DPI & MVI.

It aims to capture the early growth phase of a wider range of crypto projects and outperform market capitalization-weighted benchmarks, without introducing any subjective bias but relying instead on a rational and automated approach.

While this naturally points towards a customer base with higher crypto knowledge and risk acceptance than DPI or MVI, in our view PMIā€™s goal is also to help beginners to automatically manage and rebalance the riskier portion of their portfolio. In this respect, we believe that PMI has the potential to attract a wide range of customers inside and outside the current base of Index Coopā€™s.

2.3 - Value Proposition :

  • PMI is a self-driving smart beta strategy which provides exposure to any sector of the crypto industry (not just DeFi).

  • PMI clearly differentiates itself from the other indexes that have been launched by the Index Coop so far, or are in the product onboarding process.

  • PMI fully leverages the potential of the Set Protocol, with a high level of automation and computation.

  • If liquidity on the secondary market allows in the future, a portion of PMI could be allocated to LP or interest-bearing tokens presenting the best risk-adjusted intrinsic productivity : a selection based on the Sortino ratio of yield or fees would also work !

3/ The Methodology

Here are the various requirements foreseen before inclusion of a token into PMI :

3.1 - Market requirements :

  • The token is traded on DEXes with reasonable liquidity such as Uniswap and Sushiswap.
  • The token has a minimum liquidity of $3M (6M liquidity pool).

3.2 - Technical requirements :

  • The token is an ERC-20 token.
  • The token is not a stablecoin (neither backed by a reserve asset nor algorithmic).

3.3 - Safety Requirements :

  • The token has at least 1 month history.
  • The project must be considered to build a useful protocol or product. Projects with Ponzi characteristics, tokens with no utility, or projects that exist primarily for entertainment will not be included.

3.4 - Maintenance / rebalancing :

  • The index composition should include minimum 10 / maximum 15 tokens in order to capture uptrend potential and dilute risk.
  • The proposed rebalancing period is bi-weekly to manage risk as efficiently as possible in choppy market conditions.

Following this methodology and based on data processed between December 2020 and March 2021, we were able to identify a shortlist of 12 tokens to provide an illustrative PMI composition :

Please bear in mind that the ranking of tokens constantly evolves with the market and liquidity conditions, therefore it is only presented here for indication.

Starting from 1st March 2021 (4 weeks after the first available datapoint for $POLS), here is how PMI would have performed in comparison with the benchmark :

Over both timespans, PMI would have outperformed DPI in terms of annualized Sortino ratio and average daily return. Over the 2 weeks timespan, the standard deviation of PMI downside would also have been optimum.

4/ The competition

We havenā€™t identified any similar index product in the crypto space as of today.

5/ The Fees

We suggest discussing fees during the product-onboarding process. We can assume that this methodology will require substantial effort from the engineering as well as methodologist sides.

6/ Background & Commitment

ā€œMonportefeuille.digitalā€ has been rebranded into bloom.fi especially for this occasion !

We are the 1st French speaking community based around a DAO and a community token called $CIRCL. We aim to promote the learning and understanding of all things DeFi, as well as a positive and rational approach to crypto investing.

My name is Julien, I am the creator of the community and also Index Coop contributor since the end of 2020. My background is engineering, but Iā€™ve been falling down the crypto rabbit hole since 2018 and experiencing an incredible journey so far - of which Iā€™m now keen to share the learnings.

Together with my ā€œother-Web3-halfā€ and fellow DAO creator Benjamin as well as our community members, we strive to create an environment where the French-speaking audience can discover hands-on the incredible potential of blockchain and DeFi - but in a safe, transparent and accessible way.

We aspire to build on PMI and our contribution to Index Coop to take our organization to the next level by :

  • Becoming an advanced antenna of the Coop in :fr:to boost its growth outside of the US
  • Developing the liquidity and regulatory framework to open the 1st fully decentralized crypto investment club / fund in France
  • Promoting our content under the form of blog articles (:fr:) or educational NFTā€™s (:fr: & :us:)
Do you think PMI should be progressed beyond this post towards a DG1 vote ?
  • Oui / Yes
  • Non / No

0 voters

4 Likes

Wow, great effort! I have felt that there is quite a strong argument for a kind of alt-coin index in the Coop and that the main concern is with one would be how it reflects on the Coopā€™s image. This nails looking good. But then again, I am, for my sins, a part-time Francophile. If I was unkind, it is degen and completely obliterates my half-baked and ill-conceived Gods of Defi, GOD token idea. But seriously, this is way better, it doesnā€™t just look good, ā€¦it smells good too.

I know how hard you have been working and I believe in the future of France. My only questions are how big does the market look and how careful should marketing be.

1 Like

Great! It is very interesting to work on a risk-based methodology and there is for sure a market for the kind of index you are describing.

However the timeframes you are using seem incredibly short (2 weeks / 1 month for the back testing and one month history for new tokens). Is the ratio really relevant for this short amount of time?

Also, were the tokens in the exemplified PMI composition selected using the exact same criteria as the ones you would use for selecting a new token? there were probably a lot of other tokens with the same risk profile using the Sortino ratio, so how these ones (GRT, POLS, ALPHAā€¦) were selected?

1 Like

Hey Julien thanks for the post Iā€™m excited to see what the community thinks of PMI but I certainly like the look of it! It seems to hit all the right points with an objective, automated approach, giving a differentiated product and with demand for something in this area.

On the last point, Iā€™m interested to know what research youā€™ve done around the demand for this product, even if only anecdotal at this stage? Are there any products you see as competitors that are already in existence, my mind certainly jumped to $DEGEN from Indexed Finance while reading through.

The other question I have is how often do you expect projects to drop in and out? With a 2 week rebalancing period and potentially a lot of change, it could be a fair amount of work! It looks like something that might need more than 1 month of backtest dataā€¦

I feel PMI is the definition of a ā€˜sophisticapeā€™ which allows users exposure to altcoins but with a data-backed methodology. I couldnā€™t help but think of this meme as I read the post.

Screen Shot 2021-04-15 at 8.46.22 PM

6 Likes

:pray: Thanks a lot @mrvls_brkfst !

1/ Evaluating the market size accurately is not an easy task since there is no equivalent product. The closest we could find are Indexed Financeā€™s DEGEN with āˆ¼ $9M market cap according to Coingecko (actually their best product) & Spice DAOā€™s SCIFI with āˆ¼ $2.6M (fully diluted) market cap according to Etherscan.

However both of these indices rely on a narrower scope of elligible assets and / or a more subjective, less data-driven methodology : I believe this is where PMI has got the edge and can target a bigger marketā€¦

2/ ā€¦ Which brings us naturally to your interesting point regarding marketing : yes, I think we should be careful and avoid branding PMI as a degen index, but rather as a smart-beta altcoin index designed to capture growth as early as possible AND find a better compromise with downside volatility. @DarkForestCapital has already found the best way to illustrate this below :point_down::joy:

I hope I was able to answer your questions !

1 Like

:pray:Hey @trx314 thanks !

Maybe I should have given more insight about the choice of the backtesting period in the post : because I believe the ideal rebalancing period is every 2 weeks / maximum every month, I decided to provide some data focused on a single, maximum 2 rebalancing periods. Of course I can work to provide backtesting data over a longer duration.

I agree the choice of 1 month history for new tokens is quite aggressive, but is is the result of some research weā€™ve done on the evolution of the annualized Sortino ratio week on week, which shows that the metric tends to stabilize after 3 to 4 weeks for the best performing assets : you can see this post for more details.

And finally the exemplified PMI composition relies on a ranking we froze on 1st March in order to isolate a suitable dataset for the backtesting. Since then, we have automated our tool to be able to process many more tokens on a much more regular basis as explained in this post. So it is exactly as you say, a lot more tokens that could also be elligible have appeared on our radar.

1 Like

hey @Monportefeuille, thanks for all the work on this.

I gave you most of this feedback privately but I think it makes sense to also surface it publicly.

In terms of the target market, I donā€™t think there are many people in DeFi that want smart beta products, yet. It was quite noticeable in the lack of engagement on TTI, for instance. I think the market is still dominated by retail degens (outside of BTC and ETH). This leads me to the same conclusion as most others here, this is a DEGEN+ index. I trust @DarkForestCapital to meme the hell out of this, but is there room in the market for 2 DEGEN products? :slight_smile:

The original DEGEN index is marketed by redphone, I believe, who & how is going to market PMI?

What is your feel for the timing here? This looks like it might be better deployed privately, as a Set, so we can see how it performs for a couple of months, how the rebalances are managed given potentially meaningful changes in weights, etc.

Also, if you are picking up projects 1 month after launch, a lot of them will be pumping a lot of incentives out to create liquidity for their token. How would you assess that? If liquidity dries out after the initial incentive period, it might be challenging to exit the position.

4 Likes

@Monportefeuille this is an awesome proposal. I will admit that I didnā€™t fully appreciate it until a second read through and absorbing this conversation.

This is the most important sentence of this entire discussion. We differentiate and build our mote by releasing brand new and revolutionary products. Index Coop is not just re-branding the S&P 500 for DeFi, we are making incredible new products with the goal of transforming every aspect of modern finance.

Because our products are highly complex the onus is on us to build narratives and memes around these products. This product will succeed and fail based around our ability to communicate a very complex strategy in a simple and exciting way.

I am hugely in favor of this project. But we need to create -

  1. A good name for it. Iā€™m going to be honest - I wonā€™t be telling a date or my parents about the Post-Modern Portfolio Theory Index. Not because they wont be interested or able to understand the concept, but because the name will turn them off right away.

  2. A good narrative around it - this is the kind of product that you need to show the market why they need it and get people excited about it. The more catchy the name and the story behind it the better it will catch on.

  3. We should not underestimate the demand for this kind of product. I am about as deep into the alt coin scene as you can get - and I get overwhelmed all the time and can barely keep up. I promise you that I am not the only sophisticated crypto user who feels this way. Unless you are in the Matrix 24/7 no-one can keep up.

4 Likes

:pray: Hey @DarkForestCapital thanks a lot for your feedback and questions !

1/ To be fully transparent, the biggest part of the research so far has been done on optimizing PMI as a methodology. As mentioned to @mrvls_brkfst and as you correctly intuited, the closest product we could find is the DEGEN index ā€¦ even though as we speak the PMI composition would probably have nothing more to do with DEGEN !

But, without pretending that PMI is THE solution to it, I like to see it as a proposal for a wider problem that ā€œhobbyā€ investors seem to have. Especially those days I see a LOT of beginners posting screenshots of their portfolio on crypto forums, asking precisely this question to a bunch of perfect strangers : what do you guys think, what can I do better ?

And this just drives me nuts :nerd_face:ā€¦ because these guys can so easily fall into FOMO or subjective (sometimes also sponsored ā€¦) advice, I do believe there is a real need for a rational / automated / transparent / unbiased / risk-optimizing approach to this !

2/ Regarding your 2nd question I agree the 2 weeks rebalancing period is also quite aggressive : as shown by the (short) backtest above, allowing a 1 month time window increases exposure to downside volatility. Iā€™ll make sure I explore both options when I present the longer backtest, so we can put figures on the potential performance loss that this represents.

2 Likes

:pray:Thanks for your guidance and for sharing this transparently @verto0912 !

My brain is trained to initially approach problems from an engineering / product perspective, thatā€™s why I probably wasnā€™t able to answer all your questions in the initial post, and thought we would continue the discussion here :wink:

I really like the example that you take of redphone marketing the DEGEN index ! As much as I understand this approach and respect his work, I keep asking myself the questions (thinking out loud) :

  • Is DeFi marketing to the outside world really ā€œcondemnedā€ to be driven by :clown_face: figures ?
  • If DeFi aims to revolutionize finance, why wouldnā€™t a product which leverages its full technical capabilities to actually help / teach people investing find its space on the market ?

Iā€™ve got your point regarding initial liquidity incentives on my bucket list for a backtest, do you already have one or several examples in mind from the last months that we could use for a case study ?

1 Like

:pray:Hey @BigSky7 I think thatā€™s a great summary of what our actual point is !

I donā€™t think any one has the intention to get a ā€œPost-Modern Portfolio Theory Indexā€ tattoo on his back yet :wink:

Therefore Iā€™m naturally open to build a stronger branding around it, starting with a good name ! Letā€™s keep brainstorming and ideas comingā€¦

So I think itā€™s more a question of timing. In the short-term ie this cycle, I think the answer is yes. In the long-term, I think the answer is no.

Which is kind of my whole point of the market potentially not being ready for the product. At this stage of the cycle, investing based on memes is as good or better (DOGE) than investing based on rational and data-based strategy. People currently in the market either understand this or donā€™t care.

Not of the top of my head. But you could look at some of the Indexed products, like CC10 and DEFI5. I know that they had lots of liquidity when incentives were high which then trailed off. Perhaps it gives you something to model liquidity impact.

Dear all, some week-end reflexions to get the ball rolling again on branding / storytelling : while we were preparing this proposal together with @Lavi, he also mentioned the ā€œDegen with an IQā€ flavor of PMI - using these exact words.

Thinking more about it, marketing PMI as an ā€œIQ indexā€ could be quite an interesting stance to take :

  • On the one hand, because this message could indirectly resonate in a degenā€™s head :exploding_head:
  • On the other, because ā€œIQā€ stands for quality, technology, robotization ā€¦ which is exactly what this product intends to bring to crypto investors, like a robo-advisor for altcoins with intrinsic productivity potential.

Let me know what you think, in any case we will keep digging to bring more ideas out :wink:

3 Likes

I liked Post-modern, but partly because it resonates with my like for post modern punk. IQ has its merits but might sound a bit generic, it needs teaming with something.
Separately, while re-caulking the bath, I was trying to find an acronym for it. What about SOrtino Ratio Tokenised Ethereum DeFi Index ā€¦The SORTED Index, or SRTD. I half like it, half not. Whatā€™s anyone else got?

This is a great discussion!
I also liked the Post Modern Theory aspect, but I can see that there is potential to improve this from a marketing perspective. And itā€™s definitely very important to get a good story and brand for this index.

I really like the sophisticape meme, and while the degen+ / degen with an IQ story could resonate well with potential customers, Iā€™m not convinced that we should have this index associated with the Degen index. As Julien mentioned above, the methodology is fundamentally different and I think itā€™s much more powerful if the IC community creates its own moat around this product!

In my view, some of the key words (other than Post Modern Portfolio Theory) that best describe the PMI are:

  • Performance
  • Altcoins
  • Sortino Ratio
  • Risk adjusted
  • Automated / Robo advisor
  • any other points?

We could highlight the performance aspect of the product, i.e. trying to outperform other indices and provide a better alternative to what is described above as ā€œhobby investingā€. Some ideas:

  • Sortino Performance Index (SPI) pronounced spy :eyes: :mag_right:
  • Sortino KPI Index (SKI) pronounced sky :milky_way:
  • Automated Altcoin Performance Index (API) :electric_plug:

Unfortunately, some of the names already exist, but I hope to further spark the conversation nonetheless. Keen to hear what others think!

1 Like

Dear all,

Great to see so many ideas coming in to actually build a brand and a story around this methodology : thanks for taking the time to reflect on this, while weā€™re still early in the product onboarding process.

As @mrvls_brkfst & @Lavi pointed out, there are some key words that best describe the methodology, and weā€™ve also tried to brainstorm on these to be able to build a strong narrative and great memes around it.

We chose to keep the focus on the robo-advisors segment and spent a couple of hours digging in the list of the most famous robots from the movie or comics historyā€¦ out of which R2-D2 came on top. After digging a couple more hours in the dictionary, we figured out this proposal could actually tick all the boxes.

What do you guys think about R2D2i : Risk-Reward Digital wealth Development index ?

r2d2-1

2 Likes

I think you are onto something with the robot index and I like the R2D2 visual- may I suggest that you go even simpler and call it simply the Robot Index :robot:. This is a really easy thing to remember - it jumps off the screen on Uniswap or Sushi and its even easier to build a narrative around.

The trick is to get people curious enough to learn more without being to obscure. When staring at a screen full of numbers and obscure token names something like iRobot or Robot Index alongside the :robot:emoji will really jump out at people. From there a little research will lead them to the understanding that it is an automated index built around trading ratios.

The goal is to not overload people with complexity right off the bat but to draw them in. That is the beauty of a protocol like Badger the name and the meme is so strong that it draws people in - imagine how much less successful they would be if they had called it WBTC Yield Vault Protocol.

4 Likes

Thanks so much @BigSky7 : with such great input from everyone I definitely agree weā€™re getting there, step by step !

tumblr_oucx98dIVm1ue779no1_400

Besides the narrative around the methodology, this positioning as iRobot / robo-advisor gives us more landmarks from TradFi to take a crack at evaluating the opportunity size :

  • According to Statista, assets under management in the TradFi Robo-Advisors segment are projected to reach $1.4 trillions this year (3x higher than in 2018).
  • This market is expected to show an average annual growth rate of ~ 20% resulting in a projected total AUM of ~ $2.5 trillions by 2025.
  • Vanguardā€™s Personal Advisor Services take by far the biggest share of this market with $212 billions AUM as of year end 2020.

Funnily enough, there are also some recent developments in this direction in the crypto world : Makara for example just announced the closing of a $2M seed round :point_down:

In terms of delivering further input for the methodology itself, I am preparing a 3 months backtest from 1st december to 1st march 2021 as well as an update of the token selection with our latest screener version (more than 400 assets in there at the moment ā€¦) . Letā€™s keep pushing ! :rocket:

3 Likes

Hi @Monportefeuille ,

A few thoughts from me.

Itā€™s fantastic to see the application of data-driven methodologies to component selection. Itā€™s a step up from the simple market cap weights (which Iā€™m a big fan of for Trad Fi and DeFI) not to mention the community sentiment selection criteria used by other protocols.

I think the name will benefit from some more brainstorming.

1 month inclusion time neglects the impact of initial product pumps etc. I realise that the early life gems offer more upside (and better Sortino ratios ???, but they are more risky.

Itā€™s not clear to me whether the methodology accounts for liquidity mining. For example, Alpha is currently offering staking rewards on the native token which I assume is based on token reserves (unsustainable) rather than protocol income (sustainable). Does this LM impact the ratio? Should it? (The methodology may well allocate to tokens during LM and exit before the rewards become ineffective).

Provision of more backtest data is essential. Seeing how it could have worked will build confidence that the model / application of the model works in Crypto. It would also be good to see a record of recommended weights going forward with the actual returns calculated as time passes - then we can see how the prediction works vs the market.

I would also be interested to see how the recommended allocation changes over time. [One of the issues we identified with the TTI methodology was that large swings in weights / additions / removals as tokens get repriced by the market resulted in significant rebalancing. This adds friction to the management of the product. Particularly when the trade size becomes significant compared to the token liquidity.

One problem I have is that while the model is great for a rational / efficient market, my worry is that crypto markets just donā€™t behave in the way the model holds [Maybe I should just convert my holding to a total market cap portfolioā€¦]

I think that the biggest challenge is whether the market is looking for a product like this. The coop has experience of launching products that are statistically sound but donā€™t match what the market is doing ($GCI is conservative in a Bull market), and more data-driven products ($TTI) have been deferred.

Can we convince ourselves that people will choose to hold this product with long term conviction compared to others? Selling the use case for $DPI and $MVI is pretty straight forward ( You need broad DeFI exposure, The metaverse is early but has the opportunity to be high - if you pay attention).

In the current market, I think the only downside strategy that our (current) customers are considering is reducing their leverage. So, Iā€™m not seeing huge demand for smart Beta (Yet). [Which I think is similar to @verto0912 's comments]

I know that the investor relations team have been talking to a few more traditionally-minded investors and I think hat the challenge is convincing them that tokens outside BTC / ETH / etc have value / uses cases. Then that DPI is a good strategy. Iā€™m sure they will succeed in this, and that smart beta products will eventually gain traction (we want to be early to capture these people, but are we too early?).

What assessment have you done on customer demand over the next 12 months?


It really is an impressive strategy, I think that the outstanding challenges are:

  • showing that it out performs
  • Estimating the size of the market available in the medium term.

Thank you :pray:

4 Likes

Dear all,

Dear @overanalyser :owl:: thanks for looking into this methodology & product, always nice to discuss around a good bunch of data :wink:

Iā€™m finally able to share a 3 months backtest of the Robot Index, as well as a performance comparison against DPI / ETH / BTC ā€¦ and even a ā€œvirtualā€ BED index.

This backtest was run on data gathered between 1st December 2020 & 1st March 2021 : this was the quickest solution since we already had the Sortino ratio processed week on week for this period, in order to prepare the original proposal above.

On the other hand, this is the most pessimistic approach we could take to evaluate the Robot Index performance : liquidity on DEXā€™s has been roughly multiplied by a factor 5 since then, but was still marginal at the end of 2020.

Bearing all this in mind, here are the results we were all waiting for :

- Composition :arrow_right:as mentioned above, the reduced liquidity observed at the end of 2020 compared to now has delayed by a few weeks the introduction of some of the best performing assets in this backtest. This resulted in the inclusion of a reduced number of tokens for december and accounts for bigger rebalancings as well lower performance than planned (see below).

With a view to maximize diversification and minimize - when possible - the ratio trade size / token liquidity, we have pushed here the maximum number of index components up to 17 and the minimum pool liquidity to $4M.

Although the most critical phase would have been the initial constitution of the index, the next step could be investigating how an additional ā€œMVI-likeā€ liquidity-based weighting would reduce friction and affect performance.

- Average Daily Return (%) :arrow_right:it is interesting to confirm that while the whole market performed well from 1st to 15th February, iRobot outperformed. The opposite is also true : in shaky market conditions observed between 15th February & 1st March, iRobot produced one of the less bearish outcomes :

- Standard Deviation of Negative Returns (%) :arrow_right:globally confirms that iRobotā€™s downside volatility over these 3 months was either the same or better than BED or DPI :

- Annualized Sortino Ratio :arrow_right:taking into consideration the final combination of both indicators above, the Robot Index globally outperformed the market over the considered period :

Focusing exclusively on volatility-adjusted returns since the beginning of 2021, this trend is even more obvious : particularly compared to DPI, iRobot offered roughly the same daily returns with a better downside volatility - a nice sector-agnostic, auto-managing addition to a BED portfolio ! :wink:

For transparency these backtest data have been shared with @Lavi so far.

With regards to evaluating market size in the medium term, an interesting announcement was made last week by Wealthfront, one of the top 5 robo-advisors with over $25B AUM and more than 440000 customers.

They are looking to attract digitally native generations with an ETF-like approach to crypto offering : assuming only 1% of their actual AUM goes into this offering would result in $250M under management for their crypto robo-advisor.
Even more importantly the article clearly underlines the direction that TradFi robo-advisors are taking by offering crypto-native indexes !

4 Likes