Weekly Issue #6
Between 2010 and 2016, the company run by Prem Watsa, often called Canada’s Warren Buffett, lost US$4.4 billion hedging against a stock market crash that never came.
It did something different with its bonds. At the end of 2021, it kept about half its portfolio in cash and Treasury bills rather than lock money into long bonds paying under 2%.
Rates rose, and once the yields were worth having, it bought US$8.2 billion of Treasuries in 2023. Its interest and dividend income went from US$641 million in 2021 to US$2.57 billion in 2025.
Same man, same company. Both began as views about the future. One needed the forecast to be right. The other only needed him to refuse to overpay.
The news tells you where to look. The price tells you whether to buy.
Six issues in, here is the habit
A traditional value investor starts with one question: is it cheap? I ask that one too — last. First I ask what just changed in the world, and who gains from it.
When COVID shut the world’s doors in 2020, I had no idea what the market would do. Nobody did. But a hundred million people stuck at home still needed everything delivered, so I went and looked at Amazon, Walmart and Costco. The day Mark Carney was sworn in as Prime Minister, I ran the Canadian banks through my valuation toolkit and bought the two I judged the best value: more TD, and Scotiabank.
Neither is a forecast. Each is a place to look. Then the price decides — and sometimes it says no.
The same habit tells you when to leave. I owned Canadian cannabis stocks into the 2018 run-up and sold the lot near the top — not because I knew it was the top, but because no business those companies could ever build would earn those prices.
The timing was luck. The reading wasn’t.
The first five: Canadian Tire said wait, Ingredion and Stifel looked cheap, and Dick’s and Lennar were contrarian bets, one after a 40%-plus fall and the other on a homebuilding industry marked down by rates. Every call, with its price when it went out →
This week: who gains when rates go up?
Bond yields have been climbing: the 10-year US Treasury yield went from 4.79% on September 1 to 5.01% on September 16. The same day, the Federal Reserve raised its key rate a quarter point, to 3.75%–4%.
Who gains? Insurance companies. They collect your premium today, pay the claim later, and invest the money in between — the float Buffett built Berkshire on. When rates rise, that money earns more.
The one I kept coming back to is Canadian, and it is the company at the top of this letter.
Fairfax Financial (TSX: FFH) — C$2,244.95 (about US$1,600) at the Sep 18 close
It insures and reinsures property and casualty risk around the world, and invests the float — US$40.8 billion of it at the end of 2025. At the Sep 18 close my app has it at 1.25× book value and 7.4× earnings, with a 16.9% return on its equity (see FFH →).
Each dollar of book value costs a dollar and a quarter and earns almost 17 cents a year.
| The check | Reading | Verdict |
|---|---|---|
| Price to book | 1.25× | ✓ Under Graham’s 1.5 |
| Price to earnings | 7.4× | ✓ Under Graham’s 15 |
| Return on equity | 16.9% | ✓ Justifies paying above book |
| Combined ratio, H1 2026 | 93.6% | ✓ Underwriting profit |
| Share count, H1 2026 | Down 4.2% | ✓ Buybacks at about US$1,631 |
| Graham screen | 5 of 8 | ⚠ Borderline: a 2016 loss |
My app’s price-based readings at the Sep 18, 2026 close. Combined ratio: claims and expenses as a share of premiums; under 100% is an underwriting profit.
My app knows this is an insurer and tells you to set Graham’s verdict aside, rightly: two of the five passes, the current ratio and the debt test, mean little on an insurance balance sheet, and the only failure is a loss in 2016, the year it took its index hedges off.
What could break it
Insurance prices are falling. Property catastrophe reinsurance, which Fairfax sells to other insurers, is down about 16% in price this year, and commercial insurance prices have fallen eight quarters in a row. Watsa’s own annual letter says the hard market, a run of rising prices since 2019, “is now beginning to soften.”
Rising rates cut both ways. They also mark down the bonds it owns: US$486 million in the first half, about 40% of it offset as the same rise shrank the value of the claims it owes. The September 16 hike does both, in that order: its bonds, averaging 3.2 years to maturity, are marked down first and roll into higher yields over the next few years.
This year leaned on a one-off. First-half earnings fell to US$2.09 billion from US$2.38 billion a year earlier, even with a US$838 million gain from selling part of its stake in the parent of Seaspan, the container-ship company.
You are a passenger. Prem Watsa controls 43% of the votes, almost all of it through shares that carry 50 votes each, and two of his children sit on the board.
It is betting on its own stock. Derivatives tied to its share price lost US$433 million in the first half, as the stock fell.
What would prove me wrong, set before Fairfax reports its third quarter: a nine-month combined ratio above 100%, third-quarter consolidated interest and dividends below last year’s US$655.4 million, or book value per share below US$1,245.19 at September 30.
Skin in the game
Prem Watsa is paid a salary of C$600,000 plus standard benefits — no bonus, no stock awards, no pension. Beyond that, he gains only as a shareholder.
I own Lennar at US$82. I do not own Fairfax, and I will not buy any until after this letter has gone out.
Which account?
Fairfax is a Canadian company, so no US withholding tax touches its dividend in any account. Yielding under 1% at the Sep 18 close, the dividend is not the point anyway: this is a book-value compounder, and a TFSA keeps all of that growth tax-free.
The workfile
Everything above is the short version. The full write-up has every valuation tool my app runs on Fairfax, its forty-year record, hedge years included, and the insurers I considered and threw out. It is in this week’s workfile →
Lennar’s test, graded
First, a correction. Issue #5 said Lennar passed 7 of Graham’s 8 tests and had 5 of my app’s 8 warning signs firing. Both counts leaned on a half-finished 2026 year that the app had read as Lennar’s latest full year. Without it, Lennar passes all 8 (its earnings grew 97% over the ten-year window) and has 4 warnings firing: the dividend-cut warning should not have fired. The app now skips any year a company hasn’t filed. Issue #5 also gave its book value as US$89.72 a share, which counted minority interests; on shareholders’ equity it is US$89.12, and the 0.89× price-to-book at the September 11 close stands. The archived letter carries both corrections.
Before Lennar reported, I wrote down four tests. The two that would change my mind: fewer homes delivered, or gross margin compressing further. It delivered 20,840 homes in its third quarter, 3.4% fewer than a year earlier’s 21,584: that test failed. Its gross margin on homes was 15.8%, against 15.6% last quarter: that one passed.
I said I would buy more only if both held and the shares fell to about US$71.64, 10% under the September 11 close.
One failed. By my own rule, I don’t add at any price.
The September 16 hike matters less to Lennar than the 10-year yield does, because mortgage rates track the 10-year: Lennar says the 30-year mortgage rate was about 6.8% at the end of August, “and even higher since.” All four tests, graded →
New in the app
A value built for banks and insurers. Their research pages now add a justified price-to-book value: book value per share times the average return on equity, divided by the return you require (9% unless you change it). On Fairfax at the Sep 18 close it comes to C$3,478 a share. It assumes Fairfax keeps earning its seven-year average of 17.4% on equity, investment gains included.
The rules, in writing. Every price dated, whether I own the stock and at what price, no buying ahead of the letter, and corrections in the next one, all on the standards page.
Where it stands out. Of 14 paid research platforms checked on September 14, none mentioned a TFSA or RRSP view of a dividend on its pricing or feature pages. See the details →
Your turn 👇
What changed in the world this week — and who gains from it? Hit reply.
Send me your pick →Curtis Travis is a retired AACI appraiser (B.Comm, P.App) and founder of Travis Valuation. He owns shares of Lennar, Amazon, Walmart, Costco, TD, Scotiabank, Dick’s and Berkshire Hathaway, all named above. He does not own Fairfax. Figures via Financial Modeling Prep and the Travis Valuation app, as of the Sep 18, 2026 close; Fairfax’s own figures are from its Chairman’s letters (2016, 2022 and 2025), its November 2, 2023 and November 6, 2025 releases, its 2026 quarterly releases and its 2026 management circular; insurance pricing from Guy Carpenter (June 29, 2026) and Marsh (July 23, 2026); Treasury yields from the US Treasury and the rate decision from the Federal Reserve’s September 16, 2026 statement; Lennar’s figures from its June 11 and September 16, 2026 releases. Prices move. Informational only — not a recommendation, and not tax advice. Do your own diligence.