The third quarter was a very fortunate one for Middle Coast Investing. Our collective portfolio rose ahead of the S&P 500 and is ahead of our benchmarks year to date. Buyouts of four portfolio companies, including two significant positions, provided a big boost. Several other stocks jumped in price, for reasons that matched our investment theses. Only two stocks dropped meaningfully. And we managed these results without overextending ourselves. Before discussing those results, though, I’d like to cover a topic that has been on the market’s mind for 2+ years, but that I’ve only touched upon a little: Artificial Intelligence in Investing.

When a person with the slightest shred of interest in investing learns what I do, they ask: ‘how are you investing in artificial intelligence?’ More than crypto, software, or any other trends or fads I’ve known in my 14 years of investing, artificial intelligence has captured public attention. It of course has also been a major economic force, and raises a lot of questions about our market and the future of our economy.

I believe two things about AI are most relevant to Middle Coast Investing clients and future clients.

1. How we are trying to use it for our business
2. How we invest in it

Photo by Steve Johnson on Unsplash

Using AI in the business

Middle Coast Investing’s business can be broken into three components: the investing process, administration, and marketing. We’ll set aside marketing – that is its own problem that doesn’t affect current clients. The most pressing area for me to improve efficiency is in my administration process. Automation of any form would be valuable in tracking performance, transactions, balances, and billing.

AI is a broad term, though over the past few years generative AI, ala ChatGPT, Google’s Gemini, and so on, has been the most common reference. GenAI is usually an application built on a large language model that produces text in a natural, conversational way.

I’m less confident of AI’s ability to handle numbers and math properly, as we’ll see below. My initial line of inquiry has been to use Gemini as a guide to how to get smarter with Excel. It has given me a few small ideas and one potentially bigger one – the query tool that might save me big step. It has been an unsatisfying experience so far, but I need to invest more time. There are more ways to save time and automate processes. But it’s not clear to me if the solution is ‘find a way to plug AI tricks in’ or ‘just go learn how to use Excel properly.’

AI in Investment Research

Investment research and decision making is the core to Middle Coast Investing. It behooves us to consider any tool that can make that process better. It’s easy to puff oneself up and say finding stocks and making investment decisions is a human art form, one that can’t be replicated by computers. That’s obviously untrue – quantitative hedge funds have been around for decades – but also irrelevant.

I don’t intend to outsource decision making. But there are tedious parts of the research process that we are doing manually, and would like to automate.

For example, compiling data about a stock. 3^rd party sources often use non-standard definitions for financial figures. I prefer to pull information from a company’s SEC filings. If an AI program could do that for me, it would be a great help.

I started with Business A, a recently spun-off company. I asked Gemini to give me the numbers for a variety of categories: shares outstanding, market cap, book value, trailing twelve months revenue, etc.

Gemini first pointed out that for the spun off company it’s difficult, since it is a new company with less than a year of 10-Ks and 10Qs. Fair enough. I asked Gemini to give me a list of competitors; its list did not match what A said in its own filings. I then asked for data on the four publicly traded peers A listed in its filings.

The results: Gemini was correct each time on shares outstanding; close to right each time on price; right 4 times out of 5 on book value (and way off on one); close to right 3 times out of 5 on trailing revenue; and fairly poor on profitability metrics and net debt numbers. In other words, you have to accept a large degree of error to find that useful.

There are valuable uses to Gen AI: summarization, transcribing, translation. I’m not sure, though, that summarization is that useful in the research process. If a company is easily summarized, the value added from Gen AI is low; if a company is hard to summarize, it deserves the closer reading to grasp details. One doesn’t need to get investment ideas from arcane corners of company filings to benefit from a broader and deeper understanding of a given company.

Successful investing, the way we attempt it, is about identifying opportunities other people are not seeing properly. That is on the one hand hard to do with a tool that aggregates and approximates text. On the other hand, if that tool does a good job, it saves time for the decision making that is the core to successful investing.

I’ve just started experimenting, but haven’t unlocked anything yet. Suggestions are welcome.

AI as an investment theme

How are we investing in artificial intelligence? Cautiously.

I’ve said before that I both don’t believe in and am naturally predisposed to not chase trends. At the same time, artificial intelligence is making some sort of impact on companies across the market. We like boring companies, but also will take opportunities where we find them.

Everus Construction Group (ECG) is a good example; I invested because it was a spin-off that got oversold after its own and a peer’s Q4 earnings report. It also, as a construction company that also builds out power lines, benefits from the data center boom fueled by AI. The company’s performance has been very strong so far this year, its guidance for the second half of the year seems improbably low. At the same time, the follow through on this demand is yet to be proven, and this may not be a forever winner.

Big tech companies play in and are obviously affected by AI. We own Amazon (AMZN), Apple (AAPL), and a small position of Microsoft (MSFT); of those, Microsoft has been seen as a winner due to its association with OpenAI. But really, the story seems to be that Microsoft’s Azure is gaining ground on Amazon Web services. AI is the source of demand driving the continued growth.

We invested in Alphabet (GOOG) earlier this year when the market viewed it as an AI loser, which seemed unlikely to me. Gemini has only grown in prominence since then. The other part of this thesis was that a government break up of Alphabet would be good for shareholders. The market disagrees; but after the recent anti-trust ruling did not force Google to sell its Chrome unit, the stock has gone up. Waymo, Alphabet’s autonomous driving unit, is also an AI-powered company that is deploying fleets in more and more cities. Not GenAI, exactly, but in the ballpark.

Which gets to some of our ‘that’s also sort of AI’ investments. Trip Advisor (TRIP) has a motherlode of user-generated content in its travel reviews. It could productize that better with summarization tools, or license the text to an LLM developer. Lyft (LYFT) recently announced a partnership with Waymo. Whether that is a good thing or a threat is to be determined.

Semiconductor companies are the ‘picks and shovels’ of AI. Taiwan Semiconductor (TSM) is the bedrock manufacturer of semiconductors, while Broadcom (AVGO) is increasingly viewed as the non-Nvidia winner in chip designers. Broadcom trades at roughly 85x trailing earnings, 65x trailing free cash flow, and 36x next year’s earnings. I have never owned such a bubbly large cap stock. We trim the position regularly, leery of booking too large a tax hit, and also aware that Broadcom is a great company, both profitable and fast-growing. It is also tremendously valued.

The AI Bubble

I thought about discussing AI’s broader societal impact, but I’m not sure that my views are all that relevant in a professional capacity. I have concerns: the rash way the corporate class wants to implement AI and cut jobs; the environmental cost; the effect on education and on children. These will seep into our investing thinking and how we run our business, but they’re background. Its broader market impact is more on topic.

I lived through the internet bubble as a high schooler. I had no investing interest at the time, but could see the adoption of the internet picking up. AOL instant messenger, eBay auctions, ESPN.go.com, and Napster were gradual steps towards the always plugged in world. It was a novelty, but it was clear it could become more.

AI has been different. The potential became apparent as soon as Chat GPT was launched in 2022. It has real value to some people: many of my non-native English-speaking acquaintances love it as a way to write in English. Likewise, it’s valuable for us to summarize and synthesize text we’re working on in Spanish. Stepping out of generative AI, I can point to a friend who works on movies and raves about AI in sound editing. Peter Jackson used AI to edit Get Back, the Beatles documentary, for example, to isolate sound much better.

All of which is to say, if we call AI the cutting edge of modern software and computing technology, it’s no surprise that some of what it does will make things easier to do, and some of it is hype. Many key participants have admitted AI is in a bubble phase. The speed with which we’ve tried to use AI could make it harder to profit from. Whether there is just a slowdown or a more acute bursting of the bubble, it seems likely some sort of reckoning will come.

Part of investing is knowing thyself. I start from skepticism and value investing principles. I work to stay open minded, and will with AI, but almost three years on from the major Chat GPT model launch, the final real impact seems an open question to me. We will continue to balance caution with looking for ideas, and prepare our portfolio for both good and bad market environments. A natural inclination.

Q3 2025 Performance

Disclaimers and notes:

• See performance disclosures for more details on performance.
• These results are net of fees and include reinvested dividends or the cash received
• Core portfolios exclude a portfolio with huge Amazon and Apple positions.
• All calculations are done by me and subject to error.

Portfolio stats

• Our portfolio level price to earnings for the trailing 12 months (TTM) was 22.5. This compares to 22.4 P/E at end of Q2. (Note: these calculations are not to be relied upon and are based on best efforts.)
• Cash and equivalents (the ETFs MINT, JPST, SGOV, and BIL, the money market fund SNOXX, and short-term U.S. T-bills) was 14.6% of our quarter end portfolio, with an estimated average yield of 2.9%. This compares to 14.6% of our portfolio and 3.1% yield at the end of Q2. We also had a bond portfolio that accounts for 5.1% of the total portfolio size compared to 6.3% at end of Q2, and a hedge position of 0.9% of the portfolio as compared to 0.9% at end of Q2.
• We sold 11% more equity positions than we bought in Q3.

Our top 12 equity positions as of October 1, 2025:

• Amazon.com – 11.1% of our portfolio
• Apple – 7.8%
• Grupo Aeroportuario del Centro Norte (OMAB) – 4.9%
• Progressive Corporation (PGR) – 4.7%
• Astronics (ATRO) – 4.5%
• Capital One Financial (COF) – 3.9%
• Steelcase (SCS) – 3.9%
• Broadcom – 2.9%
• AerCap (AER) – 2.7%
• Lyft – 2.6%
• Everus Construction Group – 2.6%
• TripAdvisor – 2.2%

Q3 Winners / Losers

Winners	% Gain*	Losers	% Loss*
Dallas News (DALN)	2.3%	Portillo’s (PTLO)	-0.8%
Apple	1.6%	Progressive	-0.4%
Steelcase	1.6%	Lululemon (LULU)	-0.3%
Astronics	1.3%	First Citizens National Bank (FCNCA)	-0.2%
Lyft	1.0%
Everus Construction Group	0.7%

* % gain or loss in this table is as part of our total U.S. portfolio value (including non-core). PGR stock dropped 7.4% in the second quarter, resulting in 0.4% in total portfolio losses, for example.

Stock Commentary

Top Losers

Progressive

Progressive is the least concerning of our performance detractors this quarter. It has been our second biggest loser two quarters in a row – reflective of how big a position it is – and is just above flat on the year. Analyst earnings estimates continue to rise.

That usually is a contradiction – earnings drive stock prices. But the medium-term concerns are reasonable. Progressive’s earnings are so high because it is posting very low combined ratios. That tends to be unsustainable because claims regress to closer to what Progressive priced, or Progressive lowers rates. If Progressive doesn’t lower rates, competitors come in. Indeed, Progressive’s policy in force growth rate in August slowed down to 8.5% on an annualized, month over month basis, and 13% year over year, compared to 18% at the beginning of the year.

At the same time, Progressive trades for less than 13x trailing earnings, is growing policies at an 8.5%-13% pace, and has done fine through cycles. We have an order slightly below the current price to fill out the position for portfolios where we are underweight, and I am not terribly worried about PGR over the years ahead.

Lululemon

Lululemon is on the loser list for the 3^rd quarter in a row. It’s fair to say I got the analysis wrong, even as things I couldn’t quite predict fouled up the picture. The company suffers from tariffs and especially the de minimis application of tariffs on smaller shipments. It is fending off competition from Alo and Vuori and other big clothing retailers trying to muscle in on athleisure. And the part I think I got most wrong is that international growth is not strong enough to pick up the slack.

The shares are now quite cheap. If/when the company gets back to growing earnings, it could be a very attractive proposition. It doesn’t seem like that turning point is imminent, however. I’ve reduced our stake, both for tax loss purposes and given the bleak near-term outlook.  We’ll see if it flips, and also see how contagious this is for Deckers Outdoor (DECK), another of our holdings.

Portillo’s

We also got Portillo’s wrong, and quickly. The company had a disappointing Q2 report, lowering guidance. But I bought more shares, as the company was sticking to its strategy and its issues – especially getting a foothold with its newer restaurants in Texas – seemed plausible and solvable.

A month later, the company lowered its guidance and changed its strategy, slowing its growth plans. I don’t think that’s a bad move, necessarily, but it shows the company is a bit lost at the moment. Tariffs and inflation are part of the problem; McDonald’s is pressuring other fast-food companies with a value play; and the ICE crackdown on the Hispanic community probably hurts Portillo’s on the margin. But unsteady management is always concerning.

Indeed, a week after we sold half our position for tax losses, Portillo’s announced it was looking for a new CEO. The concept is very profitable, but does it play outside of Peoria and Chicago? I remain intrigued, but we’ll watch with a smaller position for the time being.

New Stocks

Corpay (CPAY)

Corpay, formerly known as Fleetcor, is a payments company. I’ve always struggled with understanding payments companies as there are so many different companies taking a bite of the pie and the value proposition. Corpay’s business is easier to grasp, though. Its fastest growing business is corporate payments, where it helps companies manage their accounts payable and do cross-border transactions. It also has a classic fuel card business, allowing companies to oversee their spending on gas or electric charging; and a lodging business, which targets hotel payments when people need to stay for emergency reasons, flight delays, or other business reasons.

The company juggles around its business lines, buying companies and then divesting units regularly. But it has grown its sales 8.5% per year for the last five years, and its earnings at a 7% clip. It trades at less than 15x 2025 cash earnings. I like how we can hedge against the price of oil – if gas prices go up, Corpay’s revenues go up – without investing in an oil & gas company. This is not the easiest company to understand, but the growth track record and runway going forward and the valuation have earned it a place in our portfolio. I should say this was an idea I first read about on Thomas Lott’s service, Cash Flow Compounders.

GOGO Inc (GOGO)

We’ve owned private jet in-flight connectivity provider Gogo before. I have my questions about the business, its competition with Starlink, and management. But the management team is new, the company is finally close to launching its 5G service and has launched its low earth orbit satellite service, and if we have turned a corner, the stock could do really well.

You can read my full argument here.

Hillenbrand (HI) & Brighthouse Financial (BHF)

You’ll notice in the winners section that we had luck with a couple reported deals that finally came to fruition. Hillenbrand and Brighthouse are similar type bets.

There are reports Hillenbrand is putting itself up for sale and that bids should come in soon. It is an industrial company that makes equipment for companies developing plastics. It is in a cyclical low point, and for now pays a good dividend and is profitable when you exclude one-off issues. My rule with these is to only invest if I’m comfortable owning the stock if a deal does not play out. We opened a small position feeling there is real value, and that a deal process might lead to it being realized soon.

There is also extensive reporting that Brighthouse is in negotiations to be sold. One of the insurance company’s biggest shareholders has come out in favor of the deal.  That could be a bad sign about management’s intentions to sell, but the company – which is a spin-off of Metlife and sells life insurance, with a complicated hedging strategy – has been a bit of a mess as a public company. Our position is smaller here, as I feel worse about owning this in a no-deal scenario.

TFS Financial (TFSL)

One more Thomas Lott idea. This is a small Ohio bank that doesn’t seem cheap, but that’s because most of its shareholders are shareholders in the mutual holding company, who waive their rights to the dividend annually. TFS is interesting, then, because it pays out an 8%+ dividend, making it a viable bond replacement, and because it benefits from interest rate cuts. We have just seen interest rates cut for the first time in 2025, and more cuts may be in store. We have stocks, like Charles Schwab, that benefit from higher interest rates, so it’s good to have the other side of the equation. And for now, this is just taking the place of a bond in one of our portfolios.

Air Lease (AL)

Air Lease is a smaller competitor to AerCap, which we’ve owned for a long time. It has agreed to a buyout. We bought shares after the deal was announced, so no big spike. Instead, we bought for two scenarios. Scenario 1 is nothing dramatic happens, we earn 3.2% in tax-advantaged accounts over the next 6-9 months, more or less equivalent to a treasury bill. Scenario 2 is that another buyer comes in and bids more for the company, as the agreed-to takeout price is a steal (a cheaper valuation than AerCap’s, for example). I don’t think scenario 2 is that likely, but it’s enough that I feel it worth buying a small position. Scenario 3 would be the deal falling apart, but Air Lease stock would still be cheap, and there’s no specific reason (i.e. anti-trust, financing) that would get in the way of the deal closing.

Fabrica Italiana Lapis ed Affini (FILA.MI)

In our European portfolio, we opened a small position in FILA. This is not a shoe company, but instead a pencil and brushes company. The idea comes from Overlooked Alpha. Essentially, FILA – owner of Dixon Ticonderoga, the pencils – owns a meaningful stake in Doms Industries, an Indian paper and arts company, that leaves the rest of FILA trading too cheaply.

Winners: A Bunch of Buyouts

Wrapping up, let’s talk about our four buyouts.

Dallas News was in our portfolio as a negative enterprise value company trying to turn around its newspaper operations. Newspapers is a tough business, and I’m not sure if the company would have finished the journey. But a sale was always logical, and it agreed to sell itself to the Hearst Corporation. Hearst runs the Houston Chronicle, which struck me as a good paper the one time I read it a couple years ago, and the offer was attractive.

Alden Global then made a larger offer. This ultimately forced Hearst to pay $16.5 instead of $14 for Dallas shares. Alden Global has a (deserved reputation) for gutting newspapers. While its final offer was $20/share, I was happy to take Hearst’s $16.5. At some point, a fiduciary responsibility includes thinking of what the news environment will be in the future for the political body, economy, and investors. We sold 2/3 of our position on the day the news came out – shares that were in tax-advantaged accounts or that would surely not reach a long-term gain – and then sold the rest across September at various parts in the deal process.

We rolled some of those DALN sales into a speculation in WideOpenWest (WOW). This is a cable company. I would not want to have owned it, but there was a standing offer to go private from a majority shareholder over a year prior, at $4.8/share, and some reporting that negotiations were continuing. We bought shares at $3.93 and then $3.74. A few weeks later, the company announced an agreement with that majority shareholder for $5.2/share. The final merger filings suggest really poor negotiating on the part of the management, but that wasn’t our problem.

Vimeo’s (VMEO) merger story also took a long time to play out. We bought on reports about a deal in March 2024. The deal didn’t come through, but I liked Vimeo’s ability to get back to growth and profitability. That was playing out in my view, but it was going to take time. Fortunately for us, Bending Spoons, the rumored buyer a year ago, came back and offered $7.85/share. While we did sell some of our shares too early at $4, we saw a nice boost in several portfolios from this.

The last buyout was the least expected. Steelcase, a major position for us I discussed last quarter, agreed to merge with HNI Corp (HNI). HNI corp previously bought Kimball International in 2023; we sold our shares then and missed out on owning HNI. This time around, merger docs show Steelcase did not shop itself around, and it probably could have squeezed more from a full deal process. But the merger makes sense, and SCS shareholders will own 36% of the combined company. We don’t plan on selling any of our shares in the near future, though we’ll happily take the $7.2/share cash HNI is offering as part of the exchange.

Stocks We Sold

Besides the buyouts mentioned, we close positions in ATKR and LICT.

Interested in more from Middle Coast Investing? Or in talking to us? Get in touch.

Please read our full performance disclosures.

Disclosure: I am long all bolded companies mentioned in this letter. I may change positions at any time. I have no immediate plans to make major changes. This is not investment advice. Investing is risky. Any investing decisions are your own responsibility and should be taken after speaking with an advisor or at your own risk. This is not a solicitation to buy or sell anything. Past performance is of course no promise of future results.

Disclaimer: I calculate performance and all portfolio figures manually, so it may be prone to error. The accounts I manage may deposit or withdraw money over the course of a quarter. I account for that in my calculations by adding/subtracting that money to/from the starting amount at the beginning of the period. This means withdrawals intensify performance and deposits dampen it. For half-year, 9-month, and full-year performance, I multiply quarterly performance by one another to control for deposits/withdrawals.