Actionable insights — Canadian Stocks at Records: Time to Get Selective

12:40 1. "There is no catalyst" — buy quality where the sector is feared, and refuse to wait for a trigger

1. Start from the sector, not the stock: find a group the market has abandoned, where valuations reflect a below-trend price for the underlying driver (commodity, cycle, growth rate) rather than a normalized one.
2. Inside it, screen for businesses whose quality you would want to own at any point in the cycle — franchise, management, profitability profile — not the cheapest optionality.
3. Explicitly drop the catalyst requirement. "At the risk of sounding glib, often we'll say and there is no catalyst — but if the stocks are cheap and the businesses are good, that's when we want to be active."
4. Accept that you cannot forecast the trigger. He had no crystal ball on what would lift oil in April 2025; the Middle East events ten months later were unknowable, and unnecessary to the decision.
5. Buy while the tape is still bad and size up through the drawdown, not after the re-rating.

• Sectors trading on a below-trend assumption for their key input while the businesses themselves are unimpaired; your own instinct to "wait for a catalyst" is the signal you're early enough.

5:50 2. Decompose the return — how much was earnings, how much was "borrowing from the future"

1. Take the realised annual return over the last several years and compare it to what the fundamentals actually delivered. Here: 15–20% a year for lengthy periods, while "the backdrop… is not 15 or 20% good."
2. Attribute the gap to multiple expansion. That portion is not new wealth — it is future return consumed early: "we're borrowing from the future… we're pulling that forward."
3. Cross-check the valuation regime with a second-order tell: the equity's yield versus the bond yield. When the old "Canada is cheap and yields more" argument inverts, the re-rating is complete.
4. Lower your forward return expectation for the whole index accordingly — and demand that any name you keep at full weight be justified by fundamentals, not by the multiple.

• Multi-year returns running well above the fundamental trend; a market's dividend yield falling below the bond yield after years of the opposite.

9:41 3. Fund the buy list by trimming the consensus winner — the crowded name is your source of cash

1. If your mandate keeps you fully invested ("we are not the market" — an active manager rotates, it doesn't go to cash), every purchase must be funded by a sale. Name the funding source deliberately.
2. Pick it by price extremity, not by deteriorating fundamentals: rank holdings by forward multiple against their own history and flag the ones at or near all-time highs.
3. Trim modestly and repeatedly rather than exiting — "trim, albeit modestly" over successive quarters, so you keep the exposure while harvesting the re-rating.
4. Explicitly separate the two judgments in your notes: "it's not really a concern about what's in front of the banks fundamentally. It's just what we're paying for it."
5. Accept being early. He trimmed utilities "maybe a little bit too early… in many cases they've even moved up after we've sold them" — and still calls the discipline right.

• Any top-10 holding whose forward P/E is at a record while its earnings outlook is merely fine; sell-side and macro commentary converging on it as the consensus trade (Rosenberg's "banks are our AI bubble," 8:49).

51:38 4. The 50-cents-on-the-dollar test — buy the name you refused to own, at the multiple that finally works

1. Keep a watch list of excellent businesses you have declined to buy purely on price, with the multiple recorded. He never owned Thomson Reuters at ~30× EV/EBITDA.
2. When a narrative shock de-rates one of them, re-underwrite from scratch: work through the big segments and ask whether the terminal-risk story is actually visible in any of them.
3. Value the business on your own normalized numbers and express the entry as a discount to that value. Demand a wide one — here "more than a 50% discount," i.e. "close to 50 cents on the dollar."
4. Frame both outcomes before buying: right → "outsized returns probably for the next decade"; wrong (slower growth, real threats) → the discount still leaves "a slightly lesser return, but probably still a good return."
5. Sanity-check the other side of the trade: the buyer at 30× "had to have the world come in perfectly and still probably make barely an acceptable long-term return." If the discount buyer's bad case beats the premium buyer's good case, the asymmetry is real.
6. Expect imperfect timing — he started buying higher than today's price and says plainly "we'll never get that timing perfect."

• A previously untouchable compounder whose multiple has more than halved on narrative rather than reported numbers; "terminal risk" language entering sell-side notes.

37:21 5. The structural-versus-cyclical test — "there's nothing structurally broken"

1. When a name is hated, reduce the question to one binary: is the damage structural (the business model is permanently impaired) or cyclical (headwinds that pass)?
2. Answer it from the industry, not the stock — for rails, an "incredibly challenging environment for North American rail" affecting everyone is cyclical evidence; a competitor taking permanent share would be structural.
3. If cyclical, the negativity itself is the opportunity: "that's the time to be more interested in the stock and we'll wait for better days."
4. Underwrite the wait explicitly. CN was "backend loaded" and took more than a year to work — accept dead money as the cost of the entry price.
5. When it works, re-ask the question in reverse to decide on trimming: is there still multiple expansion left, or are you in "the later innings of that relative trade"?

• Whole-industry weakness (all peers suffering the same thing) as the cyclical tell; share loss to a new model as the structural one.

21:57 6. The management check — do it, then discount it ("management is the last to know")

1. Get close to the company: he had "a good check-in with management a couple of months ago" before reaffirming CGI, and is "pretty close to these management teams" on Stantec.
2. Then apply the standing discount: "we don't take everything at face value. Management teams can get too close to their own story."
3. Hold the explicit prior that reassurance is weakest exactly when it matters most — "management teams will be the last to know when the bad news finally arrives because they're so close to it and can't see the forest for the trees."
4. Note what they claim and treat it as one input, not the thesis. Managements now say AI is a tailwind through cost savings and new opportunities; he declines to go that far — "to construct a really bullish narrative on AI for all these businesses is taking it way too far in the other direction. Truth will be somewhere in between."
5. Diversify the bet on the answer: hold a portfolio of exposed names and expect "some that are surprisingly resilient… some that might get disrupted pretty significantly."

• Unanimous management confidence across an entire threatened cohort — a sign of proximity bias, not of safety; size positions so no single management call decides the outcome.

26:59 7. Narrative vs numbers — require the disruption to show up in reported results

1. For any "X will be destroyed by AI" story, separate the forward-looking concern from the reported evidence. "We're ultimately fundamental investors and we like to see it in the numbers."
2. Test each threatened name against its own results: is revenue, organic growth or margin actually rolling over, or are "the numbers still very much intact"?
3. If nothing has broken yet, price the fear as compensation rather than confirmation — "the stock's off a lot, too, so you're getting compensated to own it now relative to where it was in the past."
4. Refuse both extremes. He is "really not in that camp" on disruption, but equally won't join the fully bullish AI-benefit narrative, and won't claim a five-to-ten-year crystal ball.
5. Sanity-check whether the threat is even physically plausible for that business — "it's hard to claw AI up a building"; engineering and infrastructure work is not a software workflow.

• The first quarter in which the feared effect actually appears in organic growth or churn — that, not the headline, is the moment to reverse the call.

46:26 8. Read the expectations embedded in the old price — a small miss breaks an expensive compounder

1. Before diagnosing a collapse, look at what the stock cost at its peak. "It wasn't a cheap stock at its peak, and with that comes lofty expectations."
2. Quantify the actual shortfall. For Stantec it was organic growth "a few percentage points less" than hoped, largely in the US business — "still solid," just not what was priced.
3. Conclude that the de-rating measures the gap to expectations, not to health. That's the setup you want; a genuine operational break is not.
4. Apply the same lens to names you own that are still expensive: with "super high expectation stocks, it's not enough to just deliver growth and maybe even meet expectations — you got to keep upping the ante."
5. Which yields the practical rule: own the expensive compounder only in size you'd tolerate through an expectations reset, and be willing to say "I wouldn't argue it's cheap here."

• The size of the actual miss versus the size of the de-rating; when a "few percentage points" costs a third of the market cap, the price is doing the work.

49:17 9. The roll-up gate — price paid beats deal pace, always

1. For an acquisition-driven compounder, stop scoring management on deal cadence. "You expect all of these stories to just neatly and predictably do acquisitions at the pace you would like, and the size and the metrics — the reality is the world's much more difficult than that."
2. Expect lumpiness: deals "come in fits and starts," and as a company scales, the size and profile of the available targets changes.
3. Score them on price paid instead, benchmarked against the competitive bid: "there's as much competition as ever… private equity and other players have bid up acquisition multiples."
4. Treat a pause in dealmaking as evidence of discipline, not decay — "we don't want to see Stantec or anybody just do acquisitions for the sake of acquisitions. They've got to make the numbers work."
5. Underwrite the name so it works on organic growth alone; acquisitions become upside rather than a requirement.

• Acquisition multiples in the target sector versus the acquirer's own multiple; deal announcements that look like pace-chasing rather than value.

44:05 10. The communication discount — separate a messaging failure from a business failure

1. When a stock is down far more than the numbers justify, examine how the story has been told, not only what happened. "Management communication and just the way that this story has been messaged and perceived has really played into stock price weakness."
2. Look for the shareholder-base rotation: a former "go-go story" that missed growth expectations sheds its growth holders and finds new ones only at a much lower price — an overshoot, not a valuation.
3. Test whether the cause is fixable: a genuine change in investor-relations emphasis, clearer guidance, and possibly activist pressure ("definitely a possibility") are all repairable levers.
4. Set the repair timeline honestly before buying: confidence returns only after "consistent numbers and guidance that when articulated is generally met" — "it won't happen overnight. It won't happen in a quarter."
5. Size for that timeline and keep adding through it, rather than expecting a single catalyst to reprice it.

• Guidance repeatedly set and missed; a visible change in IR posture or a new investor-day cadence as the first sign the discount can start closing.