Lesson 2 | Statistical v. Logical Optimization

Hello Bubble Riders!

I’ll start this with a quick reminder that we at AOTB are giving away 1 year’s worth of the Crypto Rider subscription for mostly doing what you’re already doing.

Just click this link. You get additional points for following on FB and Twitter (and referring a friend).

And onto the lesson!

Most people understand the meaning behind the admonition “don’t put all your eggs in one basket.” It finds its origin in Don Quixote. The idea is that if you put all your “eggs” (investments) into one “basket” (strategy), and if that strategy fails, then you’ll lose everything.

Like many adages, it has obvious difficulties. Warren Buffett, widely regarded as the world’s best investor, is on record for opposing diversification so understood. His reply (in paraphrasing Mark Twain) is that it’s easier to put all your eggs in one basket and watch that basket very closely than to spread it among many baskets and care for each less well.

The difficulty with Buffett’s retort is that diversification actually isn’t like putting eggs in a basket.

Let’s start this one off with an image that Ray Dalio (the founder of Bridgewater Associates, which is the world’s largest hedge fund) calls “the holy grail” of investing.

You can see from that image that if you get assets (or strategies) that are uncorrelated with each other, then the probability of a loss over a given year declines significantly. The return to risk ratio likewise improves.

Look a bit more closely at that top red line, which represents coins, stocks, or strategies that have a 60% correlation. Even though those aren’t too strongly correlated, after about 3 such selections for your portfolio, your risk stays flat. That means that if you have a portfolio of 12 coins and they all have a .6 correlation, then your portfolio is no safer than one with 3 coins and .6 correlation. The added number of coins does not diversify your portfolio.

In metaphorical terms, even if you’re carrying 6 different baskets, they’ll all still break if you happen to trip. You gained nothing by putting eggs in different baskets, since you didn’t eliminate the basic source of risk. (Notably, the metaphor starts to break down here).

If your portfolio has coins with a .2 correlation (the black line), then you’ll lower your risk by adding up to about 7. Your return to risk ratio is still .5 and you’d have a 31% chance of losing money on a given year.

With truly uncorrelated returns, represented by the green line in the graph, you can add 20 such strategies (or stocks or coins) and make a meaningful difference with each one. The inflection point sits around 13 but continues to de-risk even after that.

A portfolio of 20 such truly uncorrelated investments thus constitutes the “holy grail” of investment portfolios.

In coaching sessions, our team helps clients think about how to structure their own portfolios this way. My 3-Stage Strategy automatically incorporates some of these principles. But I find that people are more likely to stick to a strategy if they understand it. 

So, let’s develop 2 related points about statistical and logical correlation.

Statistical v. Logical Correlation

What we’ve been doing thus far is examining the idea of portfolio diversification through statistical measures.

It can be a pain to find these correlations as it often involves downloading the entire history of a coin from some site as a .csv file, cleaning that data (I use closing prices) and then adding it together with the repository of other coins I’m tracking (and I do the same thing for each).

Honestly, most data analytics is pretty tedious.

A site like coin predictor can help, though it only gives you correlations with BTC. Be sure to hit the correlations tab, as their actual “predictions” are not great (you’d do better just holding ETH).

If you search for such pairs you can come up with a list that looks a little like the following.

Looking at that list, you’ll learn that, for portfolio purposes, adding ETH, BNB, and LTC to your Bitcoin portfolio doesn’t diminish your risks in any meaningful way.

As I noted before, ETH is really just a leveraged bet on Bitcoin. It moves more than Bitcoin, but ultimately, it moves about the same time as BTC (usually with a 1-week delay). Read more here.

Adding LUNA and MATIC, however, is likely to improve your risk-adjusted returns–at least given the historical performance of the coins.

And it is at this point that you’ll recognize the limitations to a portfolio built on historical correlations alone–one that only uses statistics. What reason do you have for thinking that the correlations as so presented will hold up in the future?

With LUNA, the reason is obvious. People flee main coins like BTC during a dip/crash and put their money into stable coins like UST. In turn, when UST is minted to meet the increasing demand, LUNA is burned. The total supply of LUNA thus reduces and the price per coin jumps.

I find no similar rationale for MATIC’s performance. The negative correlation (and anything over .3 is statistically significant) may simply vanish in the future. It may be an artefact of the differences in the historical runs of the two coins (MATIC is much younger than BTC). It may be that current hedge fund interest has boosted the coin recently why BTC has fallen. But funds are often fickle lovers, so I’d not count on that going forward.

To find uncorrelated returns, then, what you really want are logically unrelated strategies–not just ones that have, historically, been unrelated. The reason is that you want to know why they’re unrelated to be sure that they will stay that way in the future.

Concluding Thoughts

It’s often easier to write uncorrelated strategies than it is to find uncorrelated asset classes–which is partly why I’m always on the hunt for uncorrelated strategies (in addition to uncorrelated asset classes).

That interest also partly explains why I took cryptos seriously when I was initially just a stock trader.

In the next lesson, I’ll discuss how to achieve logical diversification through strategy diversification. For right now, it’s enough to point out that I’m developing a meta-machine learning algorithm and a whale watch algorithm for exactly this reason.

It’s also why I include yield farming in my crypto portfolio at all times (20%). Here’s a look at my actual investment portfolio (excluding commercial real estate and private equity). Also, CAGR = compound annual growth rate.

• Stocks – 40%
  • QQQ Trader (unleveraged) – 20% CAGR
  • Greenblatt (unleveraged) – 18.5% CAGR
  • UPRO Trader (leveraged) – 40% CAGR
• Cryptos – 60%
  • Yield Farming – 30-40% CAGR
  • 3 Stage Strategy – Target of 300% CAGR
  • Moonshot Portfolio – about 300% CAGR

I wasn’t aiming for such high returns on cryptos (I’ve done better tan 800% CAGR for the past couple of years) and I’m not sure than consistent 10x plays will be available for much longer. Still, I expect crypto returns to be life transforming for a bit.

Subscribers will of course be getting access to all the new stuff as they come out.

This week I wrote a number of pieces that are related to this post, so you might want to have a look if you’ve missed them.

That’s it for this week. Remember to join us on Discord if you haven’t already.

Happy Trading!!

Disclaimers

General financial disclaimer: This post is provided for entertainment purposes only. I am not giving you financial advice and I am not a financial advisor. You should expect no financial returns one way or another based on my statements. These points hold equally for any statements that could be attributed to The Art of The Bubble or any related business entities. If you decide to buy or invest in anything, then your returns and potential losses are your own. No statements about taxation are taxable advice and you are encouraged to consult your own tax professional. You are also encouraged to do your own due diligence before investing in anything.