This shareholder letter is part of the Bretton Fund 2022 Annual Report (pdf).

February 14, 2023

Dear Fellow Shareholders:

The stock market had its worst year since 2008, and listed companies we owned were not immune to the general bearishness. The fund, however, did perform quite a bit better than the broader market. In down markets, the hardest hit areas are often those with the most egregious excesses, places we tend to avoid. Faster-growing companies tend to get hit harder as well, and our high-growth investments—Alphabet, Microsoft, S&P Global—were not spared. But all in all, we’re pleased with how the portfolio fared and confident that we continue to own quality businesses at compelling prices.

Returns as of December 31, 2022 (A)

4th Quarter1 YearAnnualized
3 Years
5 Years
10 Years
Since Inception
Bretton Fund11.59%-12.56%6.60%9.97%10.56%11.06%
S&P 500 Index (B)7.56%-18.11%7.66%9.42%12.56%12.61%

Calendar Year Total Returns (A)

YearBretton FundS&P 500 Index
9/30/10 – 12/31/106.13%10.76%
Cumulative Since Inception261.59%328.40%

(A) All returns include change in share prices and, in each case, include reinvestment of any dividends and capital gain distributions. The inception date of the Bretton Fund was September 30, 2010.

(B) The S&P 500® Index is a stock market index based on the market capitalizations of 500 leading companies publicly traded in the US stock market, as determined by Standard & Poor’s, and captures approximately 80% coverage of available market capitalization.

Performance data quoted represents past performance. Past performance does not guarantee future results. The investment return and principal value of an investment will fluctuate so that an investor’s shares, when redeemed, may be worth more or less than their original cost. Current performance may be lower or higher than the performance data quoted. You may obtain performance data current to the most recent month-end here or by calling 800.231.2901.

All returns include change in share prices, reinvestment of any dividends, and capital gains distributions. Indices shown are broad-based, unmanaged indices commonly used to measure performance of US stocks. These indices do not incur expenses and are not available for investment. The fund’s expense ratio is 1.35% (effective January 1, 2020). The fund’s principal underwriter is Arbor Court Capital, LLC.

4th Quarter
In the fourth quarter, our retailers put up solid sales numbers, and the market responded favorably given how much concern there’s been over inflation and weak consumer spending. TJX Companies had the biggest impact with a 1.4% contribution to performance, followed by AutoZone with 1.1%. JP Morgan also had a strong quarter and added 1.1%.

Tech had a rough quarter—it’s had a rough year—and Alphabet and Microsoft dragged down performance by 0.7% and 0.3%, respectively. Homebuilder Dream Finders Homes took off 0.4%.

Contributors to Performance for 2022
For the full year, Google parent company Alphabet took a whopping 5.1% off the fund’s returns. Despite this hit, it has been the biggest positive contributor to the fund since we first invested in 2015, and we expect it to return to being a significant positive contributor in the years to come. Our other large tech holding, Microsoft, took off 1.8%, and Dream Finders took 2.2% as demand for housing stalled due to high interest rates.

On the other side of the ledger, Progressive added 1.5% as it was able to push through higher prices and return to its pre-pandemic margins. AutoZone added 1.2% and TJX 0.6%.

The fund issued a long-term capital gain dividend of $0.948741 per share to shareholders on December 16, which amounted to 2% of the fund’s NAV. Though it was our largest tax distribution to date, it was still much lower than the average mutual fund’s.


Security% of Net Assets
Alphabet, Inc.8.07%
AutoZone, Inc.7.87%
The Progressive Corporation6.92%
Ross Stores, Inc.6.43%
UnitedHealth Group Incorporated6.09%
The TJX Companies, Inc.5.97%
Mastercard, Inc.5.21%
NVR, Inc.5.08%
Visa, Inc.5.07%
Union Pacific Corp.5.07%
American Express Co.4.98%
S&P Global, Inc.4.98%
JPMorgan Chase & Co.4.92%
Microsoft Corporation4.90%
Bank of America Corp.4.28%
Berkshire Hathaway, Inc.3.70%
Moody's Corporation3.69%
PerkinElmer, Inc.2.46%
Dream Finders Homes, Inc.1.72%
Armanino Foods of Distinction, Inc.1.41%

* Cash represents cash equivalents less liabilities in excess of other assets.

Our best performing investment last year, Progressive, encapsulates the typical Bretton Fund investment: 1) it’s in a so-so industry (car insurance), 2) yet it has a structural advantage over its competition (superior data and lower costs), and 3) most important, it trades for materially less than what we estimate it can earn over time. Progressive’s economics were upended by post-lockdown driving behavior. People are driving a bit less compared to 2019, but they’re driving during less crowded times, which leads to faster speeds and worse accidents. More severe accidents coupled with higher prices for cars and repair costs mean higher payouts from car insurers. One of Progressive’s competitive advantages is their ability to identify shifts in behavior faster than the other guys, and they moved to raise prices to adjust for this change. The price hike didn’t allow them to add as many customers as they usually do, but they returned to their pre-pandemic margin target, which the market rewarded. Though the core auto insurance business improved, earnings per share were down 79% on losses Hurricane Ian caused in their property division. The stock returned 27%.

Heading into a potential recession—or at the very least a weak consumer environment— American Express is going to see increased defaults by its cardholders. Despite this, we think the business is well positioned. It’s got the best high-end card franchise, wealthier cardholders with lower default rates, and a mostly recurring fee business model that grows right along with the growth in card purchases. The stock is up 320% with dividends reinvested since late 2010, when we first bought it, and we still think the shares are cheap. It returned -9% last year, and earnings per share declined 2%.

JPMorgan and Bank of America didn’t fare as well last year, with their stocks returning -13% and -24%, respectively, as investors grew concerned about the weaker economy. Business-wise, 2022 was mostly good for the banks. Interest income was up sharply from higher rates, but because they had artificially high earnings in 2021 (from the reversal of pandemic-era reserves they didn’t end up using), earnings per share decreased last year, down 21% for JPMorgan and 11% for Bank of America. The current environment—however long it lasts—seems especially beneficial for banks: high interest rates, low unemployment, and low default rates.

S&P Global and Moody’s are primarily in the business of rating newly issued bonds, and naturally businesses are a little reluctant to issue bonds in high interest-rate environments. Moody’s earnings per share went down by an estimated 29%, while S&P Global, which is less reliant on ratings revenue, saw earnings per share decline an estimated 5%. S&P Global’s stock returned -28% last year, while Moody’s, a new acquisition, is down 6% from where we purchased it.

Often in down markets, businesses seen as stable and recession-proof see their shares outperform significantly. We suspect a bit of that was happening with AutoZone’s shares last year, which increased 18%, but the business also had a great year. AutoZone was able to pass its higher product costs to consumers through higher pricing, demand continued to be strong as people resumed driving, and the company further expanded their business serving auto mechanics. All this increased earnings per share by 23%.

Ross and TJX are seeing some challenges from inflation. They depend on selling name-brand clothing at prices significantly below department stores, and when wholesale prices rise rapidly, retailers can be hesitant to pass those through to consumers, who can (understandably) balk at paying 8% more for the same item they bought just a year ago. So far, TJX has handled this better than Ross and increased its earnings per share by an estimated 16%, while we estimate Ross saw a decline of 12%. Inflation will likely continue to be disruptive, but the core model is still very much intact: people really like buying nice, branded clothes at low prices, and no one does it better than these two. Ross’s stock returned 3% and TJX 7%.

Higher interest rates have been awful for homebuilders. NVR’s earnings per share declined an estimated 26% and Dream Finders 44%, and their stock prices were down 22% and 55%, respectively. We do suspect that the worst is behind them. Mortgage rates have been trending down the past few months as inflation has abated, and there is still a structural shortage of housing in the US. New home buyers will likely have to opt for smaller houses with lesser finishes, but there is still demand for new houses. The primary competition for a new house is a used house, and used houses have something of a lock-in effect as owners with inexpensive mortgages are reluctant to sell and take on more expensive mortgages elsewhere. It is worth noting that NVR and Dream Finders do not hold land, so unlike conventional builders, they are not working their way through a backlog of expensive lots.

Armanino Foods continues to rebound nicely from the pandemic. With more and more restaurants and cafeterias open, revenue and earnings were up, and the stock returned 13.48%.

Technology companies bore the brunt of this market downturn, and our tech investments were not spared. Alphabet’s and Microsoft’s stocks returned -37% and -28%, respectively. Alphabet increased its earnings per share by an estimated 11%, which for many companies would be great, but it was shy of its usual pace of closer to 20%. People googled more things and watched more YouTube videos than ever, but the prices advertisers were willing to pay for those eyeballs were weak as marketing budgets were cut. We describe Google at greater length below.

Microsoft, less exposed to that capricious advertising revenue, managed to increase revenue 19% and earnings per share 17% as the boom in shifting IT infrastructure to the cloud continued.

Visa and Mastercard continued their pandemic rebounds. People around the world started traveling again, spending money abroad, and continued to shift more of their purchases to cards instead of cash and checks. Visa increased earnings per share 27%, Mastercard increased an estimated 15%, and their stocks returned -3.39% and -2.66%, respectively.

Over the past decade plus, Union Pacific and pretty much every railway in the US and Canada completely overhauled how they run their rail networks. “Precision railroading” was—for the most part—a superior way of running trains. Trains run on set schedules on simpler routes using fewer workers and locomotives. It worked great. Customers got their shipments earlier and more reliably, and railroads were more profitable. But this model also has thinner margins for error. Post-pandemic, traffic surged—and so did employee turnover as the labor market tightened. The company did increase earnings per share 13%, but it would have been even more if they had worked out their operational kinks. Precision railroading is still relatively new to the company, and we expect they’ll work it out. The stock returned -16%.

Berkshire Hathaway generates 80% of its operating income from its variety of industrial assets—everything from gargantuan enterprises such as MidAmerican Energy and BNSF Railway to tiny businesses such as See’s Candies and Nebraska Furniture Mart. However, it still has an insurance company’s balance sheet, with half a trillion dollars of cash and investments against roughly the same amount of book equity. Some of these investments, such as Berkshire’s large bet on Occidental Petroleum, performed well, but even if they had not, the economics of insurance—policies are paid in advance—mean that higher interest rates help the business. We estimate Berkshire’s operating earnings increased a bit last year, while book value per share declined 10%. The stock increased by 3%.

UnitedHealthcare continues its stunning run as the 800-pound gorilla of the healthcare space. Back in 2002, UNH generated $25 billion in revenue; last year it generated $250 billion. Just as our portfolio companies Visa and Mastercard have established themselves as the low-cost tools to access the transition from cash to credit and the general growth in the economy, UnitedHealth piggybacks on an aging population, medical cost inflation, and seniors’ desire for low, fixed co-payments when accessing care. Earnings per share increased 17% over the course of 2022, and its shares returned 7%.

PerkinElmer was at the center of the Covid epidemic, with a sudden surge in demand for PCR testing tools and consumables. The windfall ended in 2022, as wide-scale PCR testing wound down. PerkinElmer took the windfall and transformed its business, divesting its food testing business to become 100% focused on healthcare and bulking up its drug discovery tools and diagnostics businesses. The company now generates 80% of its revenue from consumables and software. We estimate that it earned about $7.90 per share in 2022, a 30% drop from its 2021 results. The shares returned -30%.

Investments Initiated in 2022
Moody’s Corporation

Investments Exited in 2022, Internal Rate of Return
Canadian Pacific Railway, 23.8%

Alphabet is our largest position, and 2022 brought the biggest challenges it has faced in its history.

Google was founded in 1998 with a unique insight. Instead of trying to manually curate the internet and use simplistic word searches, as existing search engines tried to do, why not let behavior handle the ranking? Specifically, Google would assess a site based on how many other sites linked to it. If many web pages thought that a specific page was interesting, that was a powerful vote of confidence in the site. In addition, while brute force attempts to organize the internet only became more difficult as pages proliferated, Google became more accurate and efficient. In short order, it conquered the search market.

Search is special from an advertising perspective in that someone who is searching is, by definition, looking for an answer an advertiser can provide. A viewer sitting at home watching a football game might or might not be interested in beer and might or might not remember the commercial at some later moment when he faces a purchasing decision; someone searching on Google for beer is both interested and interested right now.

For the past quarter century, Google’s basic approach to search has been superior, and no one has managed to beat Google with Google’s tools. In November, a company called OpenAI launched ChatGPT, the first of what are sure to be many “large language models” (LLMs) that attempt to handle questions differently. LLMs crawl their training data—basically the entire internet—and try to use machine learning to identify relationships. They aren’t looking at links, they are looking at the content itself.

Reporters immediately published claims that ChatGPT is able to pass medical licensing boards, ace graduate school exams, etc. And it is certainly true that if given a prompt that might yield a concrete answer, ChatGPT will find it and answer with confidence, while Google will link to a source.

Here are two comparisons of the search output of Google and ChatGPT:
When we asked for “causes of World War 1”

  • Google returned a call-out box with text from the Indiana Department of Education and a link to the 31-page source document. For people who might find this a bit limiting—pages 13–31 focus on “Hoosier Stories” of Indiana residents’ experiences—the next four links are, Wikipedia, Norwich University, and Britannica.
  • ChatGPT returned a paragraph citing “nationalism tensions, imperialism, and alliances between different countries,” and attributing the proximate cause to the assassination of Archduke Ferdinand. Incidentally, this is extremely similar to the Indiana DoE’s summary.

When we asked for “best sushi Little Rock Arkansas”

  • Google returned Kemuri on Kavanaugh Boulevard, Wasabi on Main Street, and Mt. Fuji on Rodney Parham Road, and on the right of the screen returned a map of Little Rock with its selections marked. Selecting a restaurant icon brings up the restaurant’s contact information, website, and reviews of the restaurant.
  • ChatGPT returned that it was not aware of any specific restaurants in Little Rock that could be considered the best, and suggested checking with Yelp, TripAdvisor, or asking locals or the tourism department.

It turns out that it is easier to answer academic questions than to provide actionable consumer advice. If you use ChatGPT long enough, you will also start to notice that some answers are very correct sounding, but also very wrong. In one series of chats we had, it claimed that George Washington was married twice, once to a woman named Martha Dandridge and then to a woman named Martha Custis. (It was the same Martha; she had been previously married.) If you know there’s a good chance the answer you’re getting will be wildly wrong, it can defeat the purpose of using that tool in the first place.

LLMs are in their infancy; they will improve. Microsoft—conveniently, another of our portfolio companies—is investing in OpenAI and will integrate ChatGPT into its Bing search engine. And Alphabet has been quick to point out that it has spent billions on artificial intelligence and indeed has its own LLM program. Search is about to improve. And we expect consumers, advertisers, Google, and Microsoft will all benefit for years to come.

As always, thank you for investing.

Stephen Dodson            Raphael de Balmann
Portfolio Manager         Portfolio Manager