Inside Warren Buffett’s Tech Bag: Fintech in Focus

In my previous article, https://marketzen.quora.com/What-s-In-Warren-Buffett-s-Tech-Bag, I identified six tech stocks held by Warren Buffett’s Berkshire Hathaway. In this article, I will take a closer look at two of them: Nu Holdings (parent company of NuBank) and StoneCo, fintech companies operating in Brazil, Colombia, and Mexico. Relying on soft information that Berkshire Hathaway invests very selectively in great businesses, we can infer that NuBank and StoneCo are great businesses. The question is whether these companies are fairly priced at this moment; to do this, I will compare price to book value, revenue, and profitability.

Background

For context, the U.S. has a population of approximately 330 million people and over 4,300 depository institutions (https://www7.fdic.gov/sdi/main.asp). By contrast, Brazil has a population of about 212 million, two-thirds the size of the U.S., but has only about a dozen large banks (https://banksdaily.com/world/Brazil). I could not find information on smaller regional banks, but this paints a picture of an industry controlled by few legacy companies, high monopoly power, limited reach in rural areas, and most likely inefficient/high-cost operations.

Being tech-centric, NuBank and StoneCo have an opportunity to compete at a higher scale, lower average cost, and higher profit margin than legacy banks. If the fintechs can achieve very high volume (measured by revenue and number of customers), they will dominate the banking industry — one day. As leaders in fintech, they also have the opportunity to erect barriers to entry (“moats”, as Warren Buffett calls them) to discourage new entrants to fintech; this includes providing cheaper services than new entrants, making it difficult and costly to switch, and even buying the competition out of existence.

It’s important to note that NuBank and StoneCo are not competitors. NuBank markets to consumers, providing online checking & savings accounts, credit cards, and loans (similar to Sofi in the U.S.). StoneCo provides payment solutions to merchants (similar to Square, now Block in the U.S.) .

Book Value

Book value is one of the key metrics for Warren Buffett and value investors. It is the net tangible assets (total assets minus liabilities) reported on the balance sheet. It answers the crucial question: in the worst-case scenario, if the company is in dire need to sell off its assets, what will be left for investors?

For historical context, bankruptcies and liquidations do happen, albeit infrequently. When it does, it really hurts. Some notable examples are Enron, Worldcom, and dozens of financial institutions in the 2008 financial crisis (Bear Stearns, Washington Mutual, etc.); their demise led to the sale of individual assets (buildings, equipment, etc.). This is when it really matters to investors that they can recoup as much of their investment as possible.

To drive home the point, this would be the same question that one would ask when considering whether to take on more debt to buy a house: “what will I have left if I am forced to sell my house and pay off my debts?” If the answer is “not much”, then the investment would be highly risky and ill-advised.

Using the latest available data, 2020, I compare NuBank and StoneCo in the table below. For comparison, I added PagSeguro, a Brazilian fintech competitor to StoneCo not owned by Berkshire.

StoneCo and PagSeguro’s numbers were reported in Brazilian Real. They were converted to USD at $.19 USD/BRL. NuBank’s numbers were reported in USD.

The middle column tells us that investors would be able to recover at maximum $.09, $4.93, $4.73 per share from NuBank, StoneCo, and PagSeguro respectively, if these businesses had a fire sale. The higher the number in this column, the better. But, we have to put this context described below.

When buying a business, investors should be prepared to pay a price premium higher than book value, but not too high. To perform this calculation, we divide the share price by book value per share, shown in the last column. The lower the resulting ratio, the better.

Benjamin Graham, Warren Buffett’s mentor, advised that the Price-to-Book-Value ratio greater than 1.5 is too expensive, and any positive number between 0 and 1.5 is fair. In other words, a stock investor should not pay more than 50% premium over a company’s book value. By this standard, StoneCo is the most fairly priced, PagSeguro second, and NuBank last.

Surprisingly, even by one of Warren Buffett’s favorite metrics, NuBank is extremely expensive. For it to be more attractive, the price has to come down a lot, its assets would have to be assessed at higher values, or the company has to pay down its liabilities by a lot (or some combination of these things). The best thing going for NuBank is that it has first-mover advantage in the consumer fintech in Brazil.

Key Takeaway: StoneCo is the most attractively priced on a price-to-book-value basis.

Revenue

Without revenue (sales), there would be no profits and a viable business, so this is another key metric. The table below gives us an idea of the each business’ growth rate over time, and its size relative to competitors’ (market share).

Table 2 shows that PagSeguro is the largest fintech by revenue/market share — all fintechs experienced an impressive 35%-55% compounded annual growth rate (CAGR).

StoneCo and PagSeguro’s numbers were reported in Brazilian Real. They were converted to USD at $.23 USD/BRL. All others were reported in USD. Revenues for 2021 are trailing twelve month (TTM) estimates (Feb 2021 to Feb 2022).

Also shown in the table are revenues for the legacy banks in 2019, the latest available year (https://www.statista.com/statistics/954704/leading-banks-brazil-financial-services-revenue/). It is striking that fintechs’ revenues pale in comparison to the legacy banks’; NuBank, StoneCo, and PagSeguro’s combined revenues were less than even the smallest legacy bank’s (Safra) in 2019. Furthermore, fintechs’ combined market share is 1/100th the size of the legacy banks’ market share. This underscores the tremendous opportunity for the fintechs to take market share from the legacy banks.

Next, we evaluate whether the fintechs are priced fairly relative to revenues, shown below, using 2021 data.

In addition to paying a premium over the net tangible assets, investors should also expect to pay a premium for the revenues generated by the business. We calculate the premium (or discount) by calculating the price-to-revenue ratio (last column in the table). We want the lowest possible ratio (resulting from higher sales for a given price). As a rule of thumb, the price-to-revenue ratio should not exceed 10.

Key Takeaway: PagSeguro is the most attractively priced, having the lowest price-to-revenue ratio. StoneCo is also attractive, with a ratio below 10. NuBank looks way too expensive.

Profitability

Finally, we assess price attractiveness based on profitability as measured by fully diluted net earnings — earnings spread across currently outstanding shares plus yet-to-be-issued shares (convertible bonds, employee stock options, etc.).

It is important to note that NuBank operated at a loss of 4 cents per share in 2020 — quite rare for a Warren Buffett-backed company to be losing money.

The higher the earnings pers share (EPS), the better. PagSeguro comes out on top on EPS. The next question is: are the current share prices fair relative to EPS? To answer this question, we divide share price by EPS, giving us the price-to-earnings (PE) ratio in the last column. The lower the PE ratio, the more attractive the price. As a rule of thumb, a reasonable PE should be between 10 and 20.

Note that in 2020, PagSeguro and StoneCo looked attractive on PE basis, but full-year 2021 earnings have not yet been reported. At this point, StoneCo has reported that it lost $100 million in the first three quarters of 2021. PagSeguro and NuBank will probably report losses for 2021 as well.

Given that we have incomplete data for 2021, I will refrain from making conclusions about price attractiveness on a PE basis at this point.

Key Takeaway: It’s too early to call the clear winner on the basis of PE ratio, since we are still waiting for the full-year 2021 earnings reports.

Conclusion

Based on the first two criteria of determining price attractiveness (price-to-book-value and price-to-revenue), StoneCo and PagSeguro are tied, while NuBank lags behind. Based on the price-to-earnings multiple, though, the jury is still out, due to incomplete data; I will update this article once all the data are available for 2021.

I close with two unanswered questions:

  1. Why does Berkshire Hathaway prefer StoneCo over PagSeguro when PagSeguro is larger and has better price-to-revenue valuation than StoneCo?
  2. Why did Berkshire invest nearly $1 billion in NuBank when it is extremely overpriced and operating at a loss?Perhaps I will be able to answer these questions in future articles as more information becomes available.

If you find these articles helpful and would like to get regular updates, please subscribe below.

What’s In Warren Buffett’s Tech Bag?

During the dotcom boom, Warren Buffett famously resisted investing (through Berkshire Hathaway) in technology companies. His reason was simple: don’t invest in anything that you don’t understand. Fast forward to 2022, and Berkshire Hathaway is invested several large tech companies.

This article looks inside Mr. Buffett’s tech holdings (Tech Bag) to gain insight from his and Berkshire Hathaway’s view of the tech sector. Thorough financial analysis of these companies is beyond the scope of this article, but will instead be covered in future articles. Stay tuned!

The dataset used for this article is below. It was compiled from the SEC’s 13F-HR quarterly filings, which contain Berkshire Hathaway’s ownership of publicly-traded shares, from all four quarters of 2021.
https://docs.google.com/spreadsheets/d/1vR1rrf-onc27zqjicdC2Yut41Tyro-Olf6F_smYsBY8/edit?usp=sharing

The original data source is here: https://www.sec.gov/edgar/browse/?CIK=1067983&owner=exclude

The Tech Bag

It is typical for mega institutional investors such as Bridgewater, Renaissance Technologies, TRowe Price, FMR (Fidelity’s parent company), Blackrock, etc. to hold as many as 5,000 stocks in their portfolios. By contrast, Berkshire Hathaway holds relatively very few stocks (only 52). This difference in portfolio size reflects Warren Buffett’s distinctive approach to investing: be very selective and very focused.

The selection criterion for the Tech Bag is: the company generates revenue (output) by creating technology, as opposed to using technology as input to provide a product or service.

Verisign, American Express, Visa, and Mastercard are in Berkshire’s portfolio. Some may argue that they are tech companies, but they rely on technology as input to their business model to provide payment services. Therefore, they were not included in the Tech Bag.

On the other hand, Apple, Snowflake, Amazon, Nu Holdings, Activision Blizzard, and StoneCo were selected into the Tech Bag because their revenues come directly from the technologies that they create. In other words, technology is an output in their business model. And they are well-loved by Mr. Buffett, as he directed Berkshire Hathaway to either maintain or added more shares in 2021; he did not sell a single share of these stocks.

Highlights

Below are highlights from Warren Buffett and Berkshire Hathaway’s Tech Bag, in order of relative size in the portfolio. These highlights are intended to help you get started on your research process.

Apple: Berkshire has held 887,135,554 shares throughout 2021, worth a whopping $157 billion (comprising 47% of its entire portfolio of publicly traded stocks) as of Dec. 31, 2021. This makes it, by far, the largest holding; for perspective, the next largest holding is Bank of America, at 13%. It is well-known that Apple is highly profitable (currently earning $6 per share), and even pays a dividend to the tune of $14 billion annually from 2018-2021. Approximately $800 million has gone to Berkshire Hathaway every year, based on a ~ .52% dividend yield, allowing the firm to deploy the cash into smaller, newer, riskier ventures.

Snowflake: It was widely reported that Berkshire invested $735 million in pre-IPO shares of Snowflake, a cloud-computing company that went public in October 2020. This investment is now worth $2 billion, representing .63% of the entire Berkshire portfolio — a tiny drop in the bucket relative to Apple, but a very good return on investment for Berkshire. Though Snowflake has negative profit margins, its revenue growth has been growing impressively at 120% compounded annual rate (or tenfold from $97 million to $1 billion) since 2019.

Amazon: Berkshire owns $1.8 billion (.54% of its total portfolio). Amazon may have its roots in selling books online, but its innovation in cloud computing (Amazon Web Services/AWS) has made the company dominant in this area. In 2021, Amazon’s AWS revenue was $62 billion, 3x Google Cloud’s $19 billion, and 5x Microsoft Azure’s $11.7 billion. Though already a mature company, Amazon’s business has been growing rapidly at a compounded annual rate of 26% from 2018 to 2021.

As a side note, despite being friends with Bill Gates and having high respect for him, Warren Buffett does not own any Microsoft shares. This is an example of how Warren Buffett separates his emotions from his business decisions.

Nu Holdings: A new position as of 2021Q4, this is the parent company of Nubank, a fintech/digital banking company competing with traditional banks in Latin America (Mexico, Colombia, and Brazil). It went public in Dec. 2021, and Berkshire was a pre-IPO investor, whose position in this company is worth ~$1 billion. NU has been growing its revenue at a compounded 55% annual rate since 2018.

Activision Blizzard: Also a new position as of 2021Q4, this is a gaming company that is an acquisition target by Microsoft, pending regulatory approval. It has been a public company since the early 90s, when it was trading in the $1 dollar range. Berkshire’s entry price was about $66. This is another reflection of Warren Buffett’s philosophy (to paraphrase): it is not too late to buy a great company when the price is fair relative to its future potential.

StoneCo: This is another fintech company, based in Brazil. It is the most beaten-down stock in the Berkshire Tech Bag. Throughout 2021, Berkshire held on to every single share, from Q1 (worth $655 million) to Q4 (worth $180 million) — a 72% plunge in value. From 2018 to 2021, the company’s revenue grew from $1.5 billion to $3.7 billion, a 35% compounded annual rate.

Again, I will save detailed analysis of these six tech companies for future articles.

What Can We Learn From This?

One can infer Warren Buffett’s view on technology companies based on Berkshire’s buying and selling of their shares. As mentioned at the beginning of the article, Berkshire Hathaway has not sold a single share of the Tech Bag stocks discussed here, reflecting high confidence in them, unlike the recently-purged airlines and IBM.

For further perspective, Nu Holdings and Activision Blizzard sit in the middle of the portfolio when ranked by value; they are ranked 23rd and 24th out of 52, despite being new investments. This is another sign of Warren Buffett’s confidence in them.

The common threads that run through Warren Buffett’s Tech Bag are market dominance and high growth. For the larger positions, such as Apple, profitability, value, growth, market dominance, and dividend income play a role in his portfolio. For the smaller positions, high revenue growth and market-dominance potential are key.

Mr. Buffett’s approach is to reinvest dividend income generated from Apple and others (Bank of America, American Express, etc.) in smaller, newer tech positions, such as Nu Holdings and Activision Blizzard. This is a strategy that I am happily employing and advocating.

Conclusion

In his own words, Warren Buffett once said “It’s far better to buy a wonderful company at a fair price than a fair company at a wonderful price.” This means that it’s okay to be late on an investment and pay a little higher prices, as long as it is a great company. As shown by Apple and Amazon, great companies keep thriving and adapting to new challenges.

Now that Warren Buffett and company have helped us to identify great tech companies, the next step is to evaluate how fair their current prices are. Are we too late? Or are there still buying opportunities? These questions will be answered in upcoming articles, specifically for the six stocks in Warren Buffett’s Tech Bag.

If you find this article helpful, and want to be notified of future articles, please subscribe below!

Does The Super Bowl Winner Predict the Stock Market?

No! Every rational person knows the answer intuitively, but this question is often discussed in the media for fun, of course. One prediction often made is that, if the NFC team (L.A. Rams this year) wins, then it bodes well for the stock market by yearend; the opposite is true if an AFC team wins (Cincinnati Bengals this year). In this article, I will present data that proves your intuition correct — football and the stock market have nothing to do with each other, so one cannot predict the other.

The Data

The dataset that I prepared for this article is included below:
https://docs.google.com/spreadsheets/d/1kHW85G5x91cpChG6K_DXi0RTIg9FY7POgf8k_M_1T9s/edit?usp=sharing

In the first tab of the spreadsheet, you’ll find the results of the Super Bowl since 1967. The second tab are the daily closing levels of the S&P 500 index. The third tab summarizes the S&P 500 performance by year, lined up with the Super Bowl results, starting in 1971 (the earliest year for which I have data on the S&P 500).

There have been 55 Super Bowl games from 1967 to 2021. The NFC has won 29 games, while the AFC has won 26. The NFC’s edge is so slim, it is virtually a coin toss. Since my stock market data covers 1971 onward, the rest of this article will use 1971-2021 data; during this period, the NFC’s edge is even slimmer — 26 to 25. In the meantime, the S&P 500 has had 38 up years and 13 down years.

The Correlation

Each year is flagged for whether the NFC or the AFC team won, and whether the S&P 500 went up or down from Jan 1 to Dec 31 in the same year. Next, I calculate the correlation coefficient between NFC/AFC wins and whether the S&P 500 went up or down. A low correlation coefficient/score would be less than .25; a medium score would be about .40 to .60 range; a high score would be in the .70 to .80 range. Even a high correlation is not cause for celebration, as it can be due to coincidence or some other unexplained factor.

The correlation matrix is below:

As you can see, NFC wins have a .056 correlation with the S&P 500’s going up or down –indeed a very low correlation. For good measure, I also ran the correlation on AFC wins, which yields the same result (below).

Conclusion

Although the results were not surprising, this exercise made it possible to gauge how close our intuition was to the actual calculation. If your a priori assumption was 0 correlation, then you would be very close.

I root for an NFC team with a terrible name and record, the Commanders, but I’m always happy to root for another NFC team that competes in the Super Bowl. Go Rams!

Compound Interest:  How To Make It Work For You

It is commonly accepted that the earlier you start investing, the longer time horizon your investments will have to grow exponentially (compound). The power of compounding saves you the burden of having to put more principal into your savings and investments. Yet, why do so many have trouble getting started or have not gained traction?  In this article, I offer some explanations and share my personal experience.  I will drive home the point, using a compound interest calculator.

Human Psychology

Much of why many people fall behind on saving and investing has to do with human psychology; I fully recognize that the comments in this section do not apply to those who are struggling to meet basic needs, and investing may be the last thing on their minds.  That is understandable and I will try to be helpful without being insensitive. 

Reason #1Humans tend to avoid or delay pain in favor of instant gratification.  Saving and investing are inconvenient, if not painful for some.  For others, over-consumption gets in the way of saving and investing. This is why the savings rate in the U.S. is very low, and for years, behavioral economists and consumer advocates pushed for legislation to force employers to automatically enroll their employees in 401K and similar plans.  The U.S. Census Bureau estimates that in 2021, 59% of employees were eligible to have a 401K plan, but only 32% participated.

Reason #2Fear and Greed.  Investors, especially inexperienced ones, tend to overreact during times of distress and uncertainty (war, recessions, pandemics, etc.). This leads to panic selling.  On the other hand, during good times, they tend be “irrationally exuberant”, using Nobel-laureate economist Robert Shiller’s phrase (later made famous by former Federal Reserve chairman Alan Greenspan). This is why bubbles form, encouraging excessive risk taking, followed by severe losses in the ensuing crash. 

My Experience

I took advantage of time being on my side.  I remember Warren Buffett saying that when he was 10 years old, he figured that if he lived long enough, his investments would compound enough to make him a very wealthy man.  I took that to heart.  Even though right out of college, I had a negative net worth, overwhelmed by debt, living expenses, supporting my parents, and saving for my future, I made sure to contribute 5% of every paycheck to my retirement account.  I was fortunate enough to have a stable, average- paying job that matched my contributions.  Every time I got a raise, I increased my contribution by at least 1%. 

Key Takeaway:  I contributed at least 5% of every paycheck to my retirement savings early in my early 20s.  Over the next 25 years,  the investments grew exponentially, giving me financial independence in my 40s.

If your employer does not provide a retirement plan (401K, 403b, 457, etc.), there are other options, such as a traditional IRA, Roth IRA, brokerage account, etc.   Advantages and disadvantages of these accounts are beyond the scope of this article.  You can own any or all of them, in addition to having an employer-sponsored retirement account.  Over time, I’ve owned all of them.

I was consistent.  The key to success in most disciplines is consistency.  The words of my high school tennis coach, Mr. Martinez, still ring in my ears today.  What wins tennis matches is consistently making good shots, not the Andre Agassi-type, ESPN-highlight-worthy shots (we couldn’t make those consistently like Andre could). 

This principle worked well for me in investing, and for many others.  I let the professional traders and excessive risk takers do the day trading, while I consistently contributed to my accounts; small amounts of money were automatically transferred from my bank account to my investment accounts regularly.  This is what Warren Buffett meant when he said that you should pay yourself first.  

I visualized success.  I didn’t just imagine it. I planned it.  I ran simulations on spreadsheets.  This allowed me to make adjustments to my contributions when necessary.  I always added more when I could, especially when the market was down.  Below, I am sharing my compound growth calculator.  Feel free to save it to your Google Drive, and run the simulations yourself.

https://docs.google.com/spreadsheets/d/1Xr3Uf5qi_xBnHJFN7PrG2ZnEhHh0o-Mnl_L5L_qiLSU/edit?usp=sharing

The first tab, ‘Historical Examples’, contains historical annual rates of return on four Fidelity mutual funds from 1992 to 2020.  These funds are examples of one broad index fund and three sector funds focused on technology, biotechnology, and healthcare stocks.  I have no association with Fidelity nor am I sponsored by them.  The examples assume that, through 4 boom and bust cycles (recession of 1991-2, tech bust of 1999-2000, financial crisis of 2008, and Covid-19 in 2020), consistently contributing $6,000 per year in these representative funds would yield a portfolio value between $1.7 and $2.3 million.

Of course, this is hindsight. History doesn’t repeat itself exactly, which is why I created the second tab, ‘Simulations’.  This looks to the future (2022 to 2050), so many assumptions and guesses have to made.  This is helpful for visualizing possible outcomes, especially if you are now starting to invest or unsure if we’re in a bubble that is about to pop.  There are four scenarios with different assumptions and outcomes.  For all scenarios, I’m using the FDGRX fund, which invests in growing companies (appropriate for young investors with long time horizons).

Scenario 1:  this is the baseline scenario where nothing changes (still contributing $6,000 per year and getting the same returns in the same pattern as historically).  The final outcome is, in 2050, your portfolio will be worth $2 million.  This is the most optimistic scenario that is unlikely to repeat itself.

Scenario 2:  this is less optimistic, assuming a moderate downturn for the next six years, but having the same recovery pattern as historically.  Still contributing $6,000 per year, your portfolio will grow to $1.5 million by 2050.

Scenario 3: even less optimistic than Scenario 2, assuming a moderate downturn in the next six years, but having a weaker recovery throughout, and still contributing $6,000 per year.  In 2050, you end up with a portfolio worth $355,000.

Scenario 4:  worst-case scenario, suffering severe losses in the next six years,  followed by a weak recovery throughout.  But, realizing this, you increase your contributions from $6,000 to $10,000 per year until 2050, ending up with a portfolio worth $500,000.  This outcome is better than Scenario 3, where you did not make adjustments to your contributions.

Key Takeaway: Spreadsheet simulations help us to visualize a range of outcomes, and empower us to make adjustments in order to achieve a desired outcome.

Feel free to copy this spreadsheet to your own Google account and tweak it to your liking.   As the years pass, you can update each year with actual returns.  This is what I do every January.

Conclusion

As the saying goes, a journey of a thousand miles starts with one step.  If you haven’t started, I encourage you to take the first step.  Visualize success.  Be fearless.  Build confidence.  Find financial zen.

Which Bitcoin Whales Bought the Crash?

In my previous article, I reported that Bitcoin whales were net buyers to the tune of $4.7 billion during the Jan 14-22 crash.  Please refer to this article for more detail:  https://marketzen.blog/are-btc-whales-still-hodling/

In this article, I delve deeper to answer the question:  which subgroup of whales did the bulk of the buying/accumulation?  The dataset that I prepared for this article can be viewed here: https://docs.google.com/spreadsheets/d/13UDKvj1XLwHDHJcKizmHEpwMtmb9awIq3LN7i3Olhnw/edit?usp=sharing     

In the spreadsheet, the first tab lists the net inflows and outflows into, or out of, each whale wallet for each day between Jan 14 and 22. Column K (net_flow_sum) shows the aggregate net flow for the period for each wallet. This is the key variable that I use for the analysis. A positive net flow sum makes a wallet a net buyer, while a negative net flow sum makes it a net seller.

In this study, there are 5258 whale wallets, accounting for 47% of all Bitcoin ownership as of Jan 22.  During the crash, while the market panicked, most of the whales did nothing with their coins;  only 5% (289) of the whales had any activity (non-zero net flow). 

Active Whales

Of the 289 active whales that bought or sold,  213 (75%) of them were net buyers of BTC, adding 131,410 Bitcoins to their wallets.  Even though we cannot identify the owners of these wallets, we can profile them by wallet age and balance to get a sense of their common characteristics.   Wallet age is the number of days between the first date that Bitcoin was deposited into the wallet and the last date that the wallet had any activity (transfer in or out).  Balance is the number of coins in the wallet as of Jan 22.

Having identified the active whale population, we ask the following questions:

1. What is the typical BTC active whale wallet balance?
2. How do they compare to Ethereum whales? See my previous article on ETH whales https://marketzen.blog/which-ethereum-whales-bought-the-crash/
3. Crucially, which whales bought the crash?

Given that Bitcoin’s market capitalization is twice that of ETH’s, one would expect that the typical BTC wallet is richer than its ETH counterpart.  This is evident in the chart below.

The typical active Bitcoin whale wallet is richer than its ETH counterpart by a lot ($8.8 million gap for net-buyer whales, and $25 million for net-seller whales).  The $25 million net-seller gap is quite startling, suggesting that ETH sellers either had less confidence in their investment, or started with less than their BTC counterparts, or both.  This is a future study where I look across the distribution of BTC and ETH wealth to see which investors are more confident in their investment (recognizing,  of course, that some investors hold both assets).  Stay tuned!

Next, we look at the relationship between wallet age and net flows to answer the main question.  The table below helps us to see that the newest 61 wallets, 0 to 2 weeks old, did the vast majority (81.17%) of the net accumulation.   These results are quite stunning, but similar to what we saw in Ethereum (85% net buying by the same age group).

The next table shows a lot less net selling, although pretty evenly distributed across 3 wallet age groups. The total net selling (12,125 BTC) is dwarfed by the total net buying (145,273 BTC) by a factor of twelve to one.

Please refer to the spreadsheet shared above for calculations.

Key Takeaway: The large gap between total net buying and total net selling reassures us that the large whales weren’t simply shuffling BTC from old wallets to new wallets. If that were the case, then those totals would be roughly equal to one another in absolute terms.

Conclusion

In Bitcoin and Ethereum, the bulk of net accumulation/buying was done by whale wallets that were created less than 2 weeks before the crash, suggesting that these wallets likely belong to new institutional investors. Did these investors know of the impending crash? Did they somehow cause the crash by manipulation? Or is this a continuation of the fundamental shift into Bitcoin, Ethereum, and other crypto assets? These tantalizing questions could be future topics.  Stay tuned and feel free to subscribe below!

Are BTC Whales Still Hodling?

From January 14 to 22, 2022, Bitcoin (BTC) led the crypto market on a downward spiral. BTC lost 19%, while altcoins lost as much as 30% in nine days. In a previous article, https://marketzen.blog/are-ethereum-whales-still-hodling/, I reported that Ethereum whales collectively accumulated $1.7 billion worth of ETH during the crash. In the meantime, what did the BTC whales do? Did they collectively sell, hold/hodl, or buy?

Data Sample

The list of whales was sourced from bitinfocharts.com. After downloading and cleaning the data, I put the BTC whale list here: https://docs.google.com/spreadsheets/d/1Ryrik7tXNouT_hqv3P5N2vbEO-QgzRn0l0d3OmaM_Ig/edit?usp=sharing

I created a BTC sample similar to the ETH sample in the previous article. Both samples cover non-exchange, non-mining pool, non-custodial, non-depository wallets; both samples include about 5,000 wallets, representing about 45% of coin ownership in their respective blockchains. Having similar samples allows us to make apples-to-apples comparisons between BTC and ETH.

I queried the sample against the bitcoin blockchain, made some calculations, and included the results in this spreadsheet: https://docs.google.com/spreadsheets/d/1D3SaM2mULU7lWaai0k6I8ZACkUBD5TxlSYJCTHzyTnY/edit?usp=sharing

The tabs in the spreadsheet above show inflows, outflows, net flows (inflows minus outflows) of BTC from and to active whale wallets (those that had at least one transaction) during Jan 14-22. Summary tables and charts are also provided, and discussed below.

January 21 In Focus

The price chart shows the steady slide from Jan 14, leading up to a sharp drop of 10% on Jan 21, culminating in a 19% drop for the nine-day period. This shock was felt hardest among the new investors who bought BTC at much higher levels, up to to $67,000 per coin. Behind the scenes, what were the whales doing?

Behind the scenes, only 289 of the 5200 whales sampled were active during this crash; this is an indication that they were mostly unfazed. Their activities are summarized in the chart below:

Not only were they unfazed, but the active BTC whales accumulated a net 134,104 coins, valued at $4.7 billion as of Jan 22. The sharp price drop happened on Jan 21. What happened on and around Jan 21 is very interesting.

There was sporadic buying by the whales in the week before Jan 21. Suddenly, one day before the sharp price drop, they started buying heavily. The buying spiked even more on the day of the sharp price drop, and continued thereafter. This behavior may be counter-intuitive to newer investors who panic-sell.

For comparison, I included a similar chart for ETH whales below:

The ETH chart tells a similar story. A full discussion on ETH whales’ activities during the same time period is here: https://marketzen.blog/are-ethereum-whales-still-hodling/

Conclusion

Combined, BTC and ETH account for 76% of the crypto market’s total value. BTC and ETH active whales poured additional $6.4 billion into the combined assets during a time that spooked the masses. This show of confidence in the two largest cryptocurrencies should bode well for a market recovery in general. In future articles, I will delve deeper into activities of BTC and ETH whales. Stay tuned and feel free to subscribe to email alerts below.

Which Ethereum Whales Bought the Crash?

In my previous article, https://marketzen.blog/are-ethereum-whales-still-hodling/, I reported that ETH whales collectively were net buyers during the Jan 14 to 22 crypto market crash. In this article, I dig deeper into which whales were the largest net buyers and sellers during this crash.

The dataset that I prepared for this article can be viewed at the link below:

https://docs.google.com/spreadsheets/d/19RjyvwAiQa_0YManKId3IYCZ3cUBP8epRf5DZEWr4Cg/edit?usp=sharing

In the spreadsheet, the first tab lists the net inflows and outflows into each active whale wallet for each day between Jan 14 and 22. Column K (net_flow_sum) shows the aggregate net flow for the period for each wallet. This is the key variable that I use for the analysis. A positive net flow sum makes a wallet a net buyer, while a negative net flow sum makes it a net seller.

There are 538 active whale wallets. I wanted to analyze the most active of them (wallets with the biggest net buys and sells, using 3,000 ETH as a threshold). There are 88 buyers of at least 3,000 ETH, and 38 net sellers of at least 3,000 ETH. In total, there are 126 such wallets; the analysis is based on this subset.

Grouping the whale wallets as described above allows us to ask the following questions:

1. How old is the typical wallet in one group vs. the other?
2. What is the typical ETH balance in one group versus the other?
3. Given the above, did new whale investors panic-sell or buy the crash?

A wallet’s age is determined by calculating the number of days between the first day that ETH was deposited into the wallet and the last transaction date. A measure of what is ‘typical’ can vary. In this analysis, I use Median, not Average in the calculation. Median is usually a better measure of ‘typical’ because it is not affected by outliers in the data, whereas Average does get affected by outliers.

The chart below shows that the median age of the most active net-buyer wallets is 12 days versus 499 days for net sellers (a surprisingly large discrepancy). This suggests that the most active net buyers were actually the new whale wallets. By contrast, the older wallets (early investors) were the net sellers.

The tables below provide a more detailed picture of buying and selling activity. The first table shows that the newest 49 wallets, 0 to 2 weeks old, did the vast majority (85%) of total buying

The next table shows that the wallets older than 2 years did the bulk of the selling, accounting for 64% of total selling.

Please refer the spreadsheet shared above for calculations

The key takeaway: new whales overwhelmingly bought the crash.

Conclusion

New small investors are known to panic-sell during market crashes. By contrast, the data clearly show that the new whales have the opposite mentality: buy the dip, and load up during the crash, even as everyone else (including the older whales) was selling. I suspect that the new whales might mostly be institutions, which would bode well for a market recovery.

Stay tuned for a similar analysis on Bitcoin whales!

Are Ethereum Whales Still Hodling?

In the nine days from January 14, 2022 to January 22, 2022, ETH followed Bitcoin on a downward spiral. ETH, the cryptocurrency of the Ethereum blockchain, lost 27% in that time. Were the ETH whales collectively selling, holding, or buying ETH?

Whales are the top holders of a cryptocurrency. On the Ethereum network, the top whale wallet holds 1.99 million ETH, representing 1.6% of all ETH in circulation. A person may hold multiple wallets, but linking wallets to people is mostly impossible (except for those in the field of crypto forensics or working at centralized exchanges). For on-chain analytics, I will focus on the wallets. I’ve included a Google spreadsheet below, which contains data from the Ethereum blockchain, for your reference.

https://docs.google.com/spreadsheets/d/1NOae2_aG2DTTKM6NRjXqmeSy3-SH8dFA7qCyX681KjM/edit?usp=sharing

Data Sample

In the spreadsheet, the first tab shows the top whale wallet addresses; these were sourced from etherscan.io. The rest of the tabs were generated by me for analysis. I started with the top 5,000 wallets — the most that I can use before running into computing limitations. From there, I excluded non-exchange, non-contract wallets, which are intermediary pools of ETH used to handle transactions for millions of buyers and sellers of tokens, NFTs, defi, etc. Including them would be meaningless — akin to including a bank or stock exchange’s assets when analyzing individuals’ wealth.

Finally, I end up with 4,787 top private individual wallets. These account for 43.7% of all ETH in circulation, giving us quite a broad coverage of ETH ownership. The timeframe covers nine days from Jan 14 to Jan 22, 2022.

What’s in the Data?

The second tab in the spreadsheet shows inflows into each whale wallet every day from Jan 14 to Jan 22; the third tab shows outflows in the same way. The fourth tab shows the inflows, outflows, and net flows side by side for each wallet, where net flow is equal to inflow minus outflow; a positive net flow is a net inflow (i.e. inflow is greater than outflow), while a negative flow is a net outflow. In the sample period, 568 out of the 4,787 wallets had any inflow or outflow. In other words, 88% of the wallets had no activity; this is a sign that the vast majority of whales did not panic during the market selloff, nor did they buy the dip; in other words, they were ‘hodling’, as the lingo goes.

The analysis below are based on the summary data found in the summary tab of the spreadsheet. The first chart shows the price of ETH steadily declining from Jan 14 to 20, then a sharp drop of 14% on Jan 21, followed by another 6% decline on Jan 22, for a 27% total loss.

Meanwhile the total net inflow of ETH into whale wallets was steadily increasing, with the small exception of Jan 18, when there was a net outflow of 22,522 ETH. When the price dropped sharply on Jan 21, the active whales pounced at the opportunity to buy.

In all, whale wallets collectively accumulated a net of 713,198 ETH (i.e. $1.7 billion of new money was spent on ETH) in the face of a 27% price decline during the nine-day period. This also implies that the net sellers were non-whales or smaller whales that aren’t in the sample.

So What?

On one hand, whale accumulation in the face of widespread panic shows the whales’ confidence in ETH. Whales tend to be early investors, whose dollar-value holdings are in the millions to billions; they tend to be long-term investors that don’t panic, but rather buy on the dip. This should give us, the ‘small fish’, some comfort that there is price support for ETH from at least 47% of the ownership.

On the other hand, as whales accumulate more of an asset, they set themselves up to later increase their influence on the price of the asset (à la Dogecoin). However, given that ETH is widely held, and the largest wallet accounts for less than 2% of all ETH, it is unlikely that the price-changing power is concentrated in the hands of one or a few players, at least for now. Future articles will cover this topic. Stay tuned.

Conclusion

As a caveat, let’s not pin all of our hopes of price stability and gains on the whales. Though the Jan 14-22 selloff can mostly be attributed to the small players, one should not underestimate the power of many small players, especially if their moves are coordinated, intentionally or otherwise. Look no further than Dogecoin, Shiba Inu, and many other meme coins.

All this to say that the whales’ show of confidence in ETH is comforting, but there are no guarantees. The way to deal with uncertainty is to diversify. Diversification across & within asset classes, across time, has shown to be an effective approach to building and preserving wealth.

How to Calculate Risk and Reward

In investing, it is widely accepted that without risk, there is no reward. By the same token, it doesn’t mean that one should take on excessive risk for a chance for little gain. The key to successful risk management, then, is the ability to calculate risk and reward; the right tool for this is the Capital Asset Pricing Model (CAPM). This article lays the groundwork for future articles where I will reference CAPM.

What is CAPM?

CAPM measures the risk and reward of an asset, relatively to the market. It answers the key question: is the risk that I’m taking worth it?

CAPM is expressed in the following equation:

Don’t let the symbols throw you off. This is just a fancy version of the high school algebra equation in the form y = q + bx, which says that y grows in a straight line as x grows, starting at point a and growing at a constant rate b.

Similarly, CAPM states that the expected return on your asset grows as you take on more risk (Beta, analogous to b), starting at the baseline, risk-free rate of return Rf, analogous to a. The risk-free/baseline rate of return is, in theory, the 10-year Treasury bond yield, but the S&P 500 is more practical as a baseline.

Beta (Bi) measures risk. This is a key term to hone in on, defined below:

Beta is first calculated as the covariance between the asset and the market (how much their prices move together), then divided by the variance (volatility) of the market (S&P 500).

In a nutshell, Beta tells us how sensitive the asset was to fluctuations in the market; it was unavoidable risk. Therefore, for any asset in our portfolio, we want Beta to be close to 0.

Another key component of CAPM is Alpha. I will spare you the equations, and you can get them at the wikipedia link provided below. Alpha measures the asset’s market-beating, risk-adjusted return; it tells us by how much the asset outperformed or underperformed the market after taking into account the risk (Beta) that we took. It is possible to beat the market nominally, but if Beta is too high (i.e. you took on took much risk by holding the asset), then Alpha becomes negative, meaning the asset underperformed the market on a risk-adjusted basis.

The bottom line: Alpha tells us if our risk was worth taking. For any asset in our portfolio, we want Alpha to be positive and as high as possible.

More detailed explanations on CAPM are here: https://en.wikipedia.org/wiki/Capital_asset_pricing_model

Shortcomings of CAPM

I referred to Beta and Alpha in the past tense because their calculations are based on prices from the past; we cannot predict the future with CAPM, but what model can? As such, we must treat Alpha and Beta as being one tool in our toolbox, rather than the end-all-be-all tool.

A more valid critique of CAPM is that it assumes Beta doesn’t change over time. I would agree that, as I’ve seen in my research, almost all assets have Beta that change over time. This makes sense because as a company matures, its risk profile changes. If it’s successful, its Beta falls as its Alpha rises over time. Furthermore, I don’t rely on Beta calculated by websites like Yahoo Finance, Google Finance, or any online brokerage website, because they give you one Beta calculated over 3 to 5 years. Instead, I calculate my own year-by-year Alpha and Beta to better reveal changes over time.

Give It A Try

I’ve set up an easy-to-use tool on a Google spreadsheet at the link below. Once you click on the link, the instructions can be found on top of the spreadsheet. Feel free to use it to do your own calculations.

https://docs.google.com/spreadsheets/d/15QPBwxS7Ij19X869jsMKPz1DLzCiBDs8veEwMuRE9zc/edit?usp=sharing

Conclusion

With the caveat that CAPM is imperfect and should complement other tools, you can use it to calculate risk and reward. By choosing investments that maximize Alpha and minimize Beta to zero, you can tip the risk-reward scale in your favor.

Do Cryptocurrencies Provide Diversification?

Many investors own cryptocurrencies to add diversification to their portfolios.   Diversification is achieved when asset prices in your portfolio are uncorrelated, i.e. they tend to move independently of each other.  Low correlation means low portfolio risk.

Given that cryptocurrencies far outperform stocks/equities (shown in the charts below), it would be reasonable to assume that crypto prices are uncorrelated with the stock market.  But is this assumption true?

Google Finance

To answer the above question, we must measure the strength of the relationship between cryptocurrencies and the other asset classes.  This means calculating the correlation coefficient between them (two assets at a time). 

Correlation coefficient values range from -1 to 1, where 1 implies a perfectly strong positive relationship (i.e. prices of two assets move perfectly together up or down). -1 also implies a perfectly strong relationship, but they move in opposite directions (one up, the other down, and vice versa).  0 means there is no relationship/no correlation (i.e. when one asset moves one way, there is no telling which way the other one will move).  For more information on the calculation and interpretation, go to https://en.wikipedia.org/wiki/Pearson_correlation_coefficient

The bottom line with correlation coefficients is that we want them to be as close to 0 as possible in order to consider two assets to be uncorrelated; somewhere from around -.35 to .35 is an acceptable range for me.  Conversely, owning assets that are highly correlated, say with a coefficient of .80 or above in absolute terms, would be redundant, hence not diversified. So there is an inverse relationship between diversification and correlation. Keep in mind that everyone should have their own tolerance threshold, depending on their risk appetite.

What Do The Data Tell Us?

To keep the analysis simpler, I use three proxies for the asset classes: Bitcoin (BTC) to represent the cryptocurrency asset class, the S&P 500 index for stocks/equities, and the Spyder Gold Trust ETF (GLD) for gold prices.  Below are correlation coefficients for the six-year window 2016-2021.   

Each square in the matrix shows the correlation coefficient for two intersecting assets.  Surprisingly, over the six years 2016-2021, there were quite strong positive relationships between all three asset classes.  Focusing on the orange squares, BTC/cryptocurrencies moved almost perfectly in the same direction as the stock market. At first glance, this is not desirable, as it implies that there is no diversification provided by cryptocurrencies.

To get a better sense of the data, I parse the six-year window into individual years, shown in the table and bar chart below:

In the yearly table and chart, I disregard 2020 as an outlier because there were massive Covid-response fiscal and monetary stimuli on a scale unlikely to be seen again.  In that year, we saw a large dip followed by a spectacular rise in all asset classes (the “everything” bubble), hence the spike in correlation between cryptocurrencies and the stock market.

Putting aside 2020, we see that the correlation between BTC (cryptocurrencies) and the stock market was highest (above .80) in 2016 and 2017 during the previous crypto bull market; this is surprising given that the crypto market was still relatively undeveloped.  After the crypto market crash in December 2017, the BTC-stock market correlation came down  to .14, .55, and .28 in 2018, 2019, and 2021 respectively; these levels are acceptable to me, based on my tolerance threshold of about -.35 to .35.

Another noteworthy observation is the correlation between BTC and gold.  We see an increasing level of correlation between gold and BTC from 2016 to 2019 — that is before and after the previous crypto bull market.  Again ignoring 2020, the uptrend in BTC-gold correlation is followed by a steep drop to negative .43 in 2021. 

This sudden reversal of trends is an indication that, in the early years, investors didn’t buy into the notion that cryptocurrencies, particularly BTC, could be a substitute for gold; therefore their prices were positively correlated (or they could have been uncorrelated).  But the -.43 correlation coefficient in 2021 is a sign that investors may finally have bought into the notion that BTC is a gold substitute; therefore, they were selling gold to buy BTC/cryptocurrencies. Of course, one data point doesn’t make a trend, so it is something that I will track going forward.

Final Verdict

Yes, cryptocurrencies have provided portfolio diversification due to their lower correlation with stocks in the past four years. And yes, investors are finally treating cryptocurrencies as a gold substitute.

January Barometer:  Myth Busted?

Around this time of year, we often hear the maxim “As January goes, so goes the year”.   That is, if the market ends lower in January 31 vs. January 1, then it will likely also end lower on December 31 vs. January 1, and vice versa.   Is this maxim reliable, or just a myth? I set out to find the answer, using 40 years of price data from the Google Finance API. 

I create data panels based on market index (Standard & Poor 500, Nasdaq, Dow Jones Industrial Average) and two twenty-year windows (1980-2000, and 2001-2021).  In all, there are six data panels; the table below is one of them.

Google Finance API

In the table, price_jan_first is the closing index level for the first trading day of  the year (e.g. Jan. 2, 3, etc.).  price_jan_last is the closing index level for the last trading day of January.  price_dec_last is the closing index level for the last trading day in December.  To the right of price_dec_last are flags with values 0 or 1 (0 = false, 1 = true)  to indicate the direction of the index.   If the index moved up from the beginning to the end of January, then jan_up = 1, otherwise jan_up = 0.  When the index went down in January, then jan_down = 1, etc.  When the index ended higher for the year (last of December vs. first of January), then dec_up = 1.

The key variable to focus on is the last column:  same_dir.  We want to count the number of occurrences when same_dir = 1;  this is when the index moves in the same direction in January as it does from January to December (both up or both down).   In the table above, there are 11 such occurrences in the S&P 500 from 2001 to 2021.  For brevity, I will not show the other five panels, but you can view them at the link below:

https://docs.google.com/spreadsheets/d/1zjv92qiD7XlOpn7Qu8WF7_BfPSJK0ooWFwjFXSR5EO4/edit?usp=sharing

Below is a summary of my findings from the six data panels.  The bar chart shows the number of years that January Barometer was reliable (i.e. returns moved in the same direction as yearly returns) for each twenty-year window and market index.

Number of Years When January Barometer Was Reliable

The table below shows the reliability rate:

Is The Myth Busted?

Yes!  As one can see from the data above, the reliability rate in the last two decades is virtually a coin toss. This was not the case three to four decades ago, when the reliability was much higher, even nearly perfect for the Dow.  There could be many reasons for this divergence, which deserve more thorough investigation, but I would submit that the main reasons might include:

1) technological advances (e.g. faster computers that enable high frequency trading by professional traders; the internet and smart phones, which give retail traders more access to the markets), and

2) globalization of capital (investors can move funds across borders through mutual funds and ETFs); economists describe this as fungible capital.

In conclusion, the maxim “As January goes, so goes the year” should be modernized to something like: “What happens in January will probably stay in January”.

Lessons From a Hedge Fund Manager

Lessons From A Hedge Fund Manager

Steven A. Cohen is a multi-billionaire hedge fund manager at Point72 Asset Management. Though probably better known for his recent purchase of the New York Mets, Mr. Cohen has been a juggernaut in the hedge fund world for decades. As a student of the financial markets, I have followed Mr. Cohen’s portfolios since the late 1990s, thanks to public information provided by the Securities and Exchange Commission (SEC). Point72’s portfolio has ballooned from $5.6b in 2009 Q1 (filed under SAC Capital Advisors LP) to $22.7b at the end of 2021 Q3. Let’s examine Point 72’s latest portfolio as of 2021 Q3 for a brief lesson in portfolio management.

As a hedge fund, Point72’s portfolio consists of long positions (shares and call options), as well as short positions (put options). Being the prudent portfolio manager that he is, Mr. Cohen’s largest positions are in lower-risk investments as shown below:

13F-HR Filing, 2021 Q3

Gleaning from the above table, we learn that Mr. Cohen’s largest position is well-diversified in Standard & Poor’s 500 Index through the SPY ETF, followed by positions in individual stocks of profitable big tech companies. In total, these investments make up $3.5b of the overall $22.7b portfolio value.

For contrast, we can look to the new additions to the portfolio since Q3 of 2021. These show up in the SC 13G filings, which are required to be filed by investors (excluding company insiders) who hold at least 5% of company shares, shown below:

SC 13G Filings, 2021 Q4 – 2022 Q1. Prices and values are as of 1/13/21.

As one can see, all of the new additions are biotech and pharmaceuticals companies, many of which are pre-revenue and profit-negative, making them highly risky. Point72 makes small, incremental moves into these risky assets, evidenced by their small values relative to the overall portfolio value.

Conclusion

Over the years, Steven Cohen’s portfolios have grown by many fold, despite having made many bad stock picks. This success is because his overall approach is not very different to what is commonly taught in college finance courses and by financial planners: diversify, be patient, allocate less to riskier assets, and more to safer assets. These principles apply whether your portfolio is $1 thousand, $1 million, or $1 billion. In future blogs, I will dig deeper into Mr. Cohen and other influential investors’ portfolios to gain insight on market trends.