Portfolio update - 2023 December 31

My top (>2%) positions in no particular order: BRKB, GOOGL, META, AAPL, AMZN, NVDA
In:
Out: RILY (price drop)

Fixed income: 1%
Cash: 11%

Sectors (kinda): Insurance (BRK): 12%, Malone/media: 1%, Banks/financials: 6%, Stock funds: 3%, Industrial: 0%, Consumer: 4%, Medical/pharma: 3%, Tech: 44%, Various owner-operators (not included in other categories): 2%

New positions: DHR, BYDDY,
Positions increased: ZS, DOCU,
Positions reduced: PAGS, GBTC, EXPE, ETHE, CNNE
Positions eliminated: WMICX, SSTWS, SPLK
Flip-flop: RILY

My returns for time periods ending December 31, 2023: 

Stock accounts: 2023: 45%, 2y: 0.6%, 3y: 5.6%, 5y: 13.7%, 10y: 9.4% annualized. 

All investment accounts including 401(k)s, ESPPs, etc (some unreliable numbers): 2023: 34.7%, 2y: 0.17%, 3y: 4.9%, 5y: 13%, 10y: 8.4% annualized.

For comparison SP500 returns (without dividends and dividend reinvestments): 2023: 24%, 2y: 0%, 3y: 8%, 5y: 14%, 10y: 10% annualized.

Sold SPLK as merger spread narrowed. Reduced GBTC, EXPE, ETHE on run up. Reduced PAGS, CNNE as low confidence positions.

Bought initial positions in DHR and BYDDY. Added to ZS and DOCU at attractive prices.

Traded around RILY position on drop and bounce.

Do most stocks underperform T-bills?

Bessembinder study on whether stocks outperform one month T-bills is pretty widely known. A lot of people summarize its results in attention grabbing one liners. Even the official ASU page link is called "do-stocks-outperform-treasury-bills".

In fact, the one liners are quite misleading. The authors on ASU page carefully choose phrases so that the claims would be technically correct. But these careful word choices hide the actual picture, which is not as sensational. Let's dig in.

The study measures wealth creation from company going public to either the end of the study (2016, 2019, 2022 - I take spreadsheet from 2019, since it was used in the paper) or to company ceasing to exist as a public company. There are no adjustments for inflation or for economy growth. This methodology strongly biases the positive wealth creation to companies that are around in 2019 as can be seen in the top of wealth creation table. Companies that have ceased to exist usually don't have positive value creation, except for the few that get acquired. Since there are only about 4300 public companies in US in 2019, it is not surprising that "All of the wealth creation can be attributed to the thousand top-performing stocks". That would be about 23% of companies existing in 2019. Even assuming that half of the wealth creation came from companies that no longer exist in 2019 (this is unlikely based on the study's construction), 11.5% of companies existing in 2019 created a large percentage of wealth. This is much higher percentage than the quoted 1-2%.

Furthermore, the study measures absolute value creation and not annualized percentage return, which leads study to claim that 1-2% of companies create over 50% of value. This is true, but only for the absolute size of the value created. It does not mean that investing into the other 98% of companies leads to losses or inadequate returns. The study itself shows that in most 3-year periods over 50% of companies create positive wealth, i.e. their returns outperform treasury bills. The study does not show how many of these 50% of companies have high annualized percentage returns. For example, if a tiny company TinyCo has absolute wealth creation of $1M in 3 years, while a huge company HugeCo has absolute wealth creation of $1B in the same time, the study considers that HugeCo created 99.9% of wealth, even if TinyCo returned 10x in terms of appreciation, while HugeCo returned only 20%. As I have mentioned, this effect is exaggerated by the fact that in 2019 HugeCos are closing to $ trillion valuations, while first $10B company appeared only in 1955, so wealth created by most companies before 1955 would be about hundred times smaller than wealth created around 2019 by HugeCos.

Bessembinder methodology assumes buying at IPO and never selling. For investors, who do not hold forever, it is quite possible to have high returns in companies that later underperform or go bust (see Blockbuster).

Some of the numbers in the Bessembinder spreadsheet look suspicious. Apparently American Airlines Group had positive wealth creation from 1939 to 2016 (though negative 1939 to 2019). In 2013 AMR corporation was in bankruptcy and merged with US Airways Group. US Airways is the company that has history going back to 1939, so it is the company considered in the wealth creation 1939-2016. But it also went through bankruptcies in 2002 and 2004. It is possible that the positive wealth creation 1939-2016 comes from dividends or other corporate actions but it is suspicious. It might be worthwhile to dig into the "positive wealth creation" of American Airlines Group from 1939 to 2016 and into the quality of Bessembinder's data.