Actionable insights — Three Top Stocks To Buy Today

1:19 1. Keep a buy list of dislocations only — and retire a name the moment it re-rates

1. Separate two categories that usually get blurred: names you hold and names you are urging capital into today. A position can stay in the book long after it stops being a recommendation.
2. Apply one test to the second list: is the price still dislocated from the fundamentals? If the gap has closed, take the name off — regardless of how well it has worked or how attached you are to the thesis.
3. Use your own output as the audit trail. If you notice you have "stopped talking about" a name, check whether that is because it re-rated (correct) or because it went wrong and you're avoiding it (a problem).
4. Re-derive the entry point retrospectively for each retired name and write down the price and the date — that gives you a library of what a real dislocation looked like, in the same names, for calibrating the next one.

• A "best ideas" list that never loses members; a thesis you keep repeating at a price 2× the one where you wrote it; and the reverse tell — silence about a holding that has fallen, which is the same failure with the sign flipped.

6:33 2. Classify why the quality business is discounted before deciding whether it's cheap

1. Start from the constraint, not the screen: a high-quality, fast-growing company is only ever available cheap when "there's something wrong." Accept that the buying window will feel bad.
2. Name the discount's category. Macro/exogenous — tariffs, recession fear, a war, a rate move — or company-specific: a management change, a new disruption risk, a broken segment.
3. For each, ask the same question: does the dislocation make sense? Does the stated reason actually impair the cash flows this business will earn over the next five to ten years, or only the next few quarters' sentiment?
4. Only buy the ones where you can articulate specifically why the market's reasoning fails. "It's down a lot" is not a category; it's a price.
5. Then act at scale while the reason is still live — averaging in through the ugly period, not waiting for confirmation.

• A discount with no nameable cause (usually means you haven't found it yet); a cause that is genuinely permanent (then the lower price is correct); and your own comfort level — if the purchase feels uncontroversial, you are probably not being paid for anything.

9:36 3. Size a disruption threat in units of volume, not units of narrative

1. Take the disruptor's most-quoted operating number — the one used in headlines — and convert it into the incumbent's units: rides, orders, subscribers, transactions per period.
2. Express the comparison as elapsed time, not as a percentage. "How long does the incumbent take to do the challenger's entire week?" is far harder to hand-wave than "1% share."
3. Then check the derivative: is the absolute gap widening or narrowing? A challenger growing fast off a tiny base can still be falling further behind in units.
4. Ask what the disruptor would have to build — physically — to close the gap, and at what capital cost. Compare that to what the incumbent had to build (often nothing: an asset-light aggregator borrows the roads, fuel and vehicles it runs on).
5. Only after that arithmetic, decide how much of the incumbent's multiple the threat should reasonably remove.

• Threat coverage quoted in growth rates rather than absolute volumes; a challenger whose expansion requires buying physical assets city by city; and the moment the absolute gap actually starts closing — that, not the percentage growth, is the signal to revisit.

10:15 4. The "can they be late and still win?" test

1. Assume the disruptive technology works and becomes mainstream. Do not build the thesis on it failing — that is the fragile version.
2. Ask instead who ends up owning the customer relationship when it works. If the incumbent can adopt the new technology as an input — buying, licensing or integrating it — the technology is a supply question, not an existential one.
3. Identify the asset the incumbent has that the technology cannot manufacture. Usually it is demand-side and local: enough riders and suppliers in each individual city to make wait times acceptable.
4. Test whether that asset must be re-earned per market. A challenger that has to rebuild density in every city separately faces a serialized cost the incumbent already paid globally.
5. Then price the timing risk: if the incumbent can adopt the technology years late and still lead, late adoption is not a thesis-breaker, and the market's urgency discount is mispriced.

• Whether the new technology is available for purchase or license (input) versus exclusively owned by a competitor (threat); whether the incumbent's advantage is local and per-market; and whether management is already piloting integration rather than denying the shift.

11:51 5. Split the bear case into parts and weight them — most of the fall is usually one part

1. Write out every distinct bear argument on the name as a separate line item, rather than treating "the bear case" as one object.
2. For each, estimate the actual cash impact and its durability: a recurring drag on earnings power, or a bounded, one-off cost the company can absorb and adapt around?
3. Rank them by how much of the share-price decline each can plausibly explain. Usually one dominates and the others are noise the coverage over-weights.
4. Discard the bounded ones explicitly, in writing — "settlements plus product changes plus some legal fees" is an adaptation, not an impairment.
5. Spend all remaining research effort on the dominant argument. That is the only one your differentiated view has to be about.

• Coverage volume that doesn't match cash impact (headline-friendly risks are usually the bounded ones); a bear case that keeps changing shape (a sign the real objection is the price); and a single argument that survives every counter — that one is your actual risk.

13:10 6. Read the operator's stated motive before writing your own bull case

1. When a controlled company makes a decision the market hates, go to the primary source — the founder's essay, shareholder letter or long-form interview — and extract the stated reason before supplying a rationalization.
2. Check the stated motive against the operator's history. A strategic obsession usually has a biographical origin, and past conflicts predict how far they will go.
3. Distinguish the popular bull case from the operative one. The consensus defence ("AI will improve the product and therefore revenue") may be true and still not be why the money is being spent.
4. Ask what structural position the spending buys: independence from a supplier, ownership of a distribution layer, removal of a toll. Those are durable even when the headline ambition isn't reached.
5. Then judge the spend against that objective, not against next year's earnings per share.

• Founder-controlled companies where the stated strategy is consistent across years; a repeat of a structural trap the operator has already lived through once; and the difference between spending that buys independence and spending that buys a headline.

17:24 7. Find the salvage floor — what the assets are worth if the ambition fails

1. For any company spending heavily toward an uncertain outcome, ask the failure question explicitly: if the moonshot does not land, what remains and what is it worth?
2. Prefer spending that creates general-purpose, re-deployable assets (compute, land, power, fleets, distribution) over spending that only has value if the specific goal is reached (bespoke R&D, marketing, one-use tooling).
3. Identify the fallback business the assets could be redirected into, and be honest about its quality — "not the best business in the world, but it's still pretty good" is a perfectly good floor.
4. Check whether that fallback earns an acceptable return on the money already spent. If it does, the capex is closer to an investment with an option attached than to a gamble.
5. Only then judge the risk/reward: high ambition plus a real floor is a different asset from high ambition alone, and it justifies holding through drawdowns the pure version wouldn't.

• Whether the capex produces assets someone else would rent or buy; a credible second use already operating at scale in the industry; and the opposite signal — spending whose only exit is the original promise.

18:46 8. Audit the narrative claim-by-claim against disclosed metrics — and decompose any falling average

1. Write the bear narrative down as a list of checkable assertions, not as a mood. "Engagement is deteriorating" becomes three separate claims about retention, subscribers and total usage.
2. Match each assertion to a number the company actually discloses, and record whether it confirms, contradicts, or is silent. Silence is a finding too.
3. Where a metric genuinely has deteriorated, decompose it before conceding. A falling per-user average is often a mix effect — new cohorts from different markets — not a decline in any existing cohort.
4. Ask whether the mix shift is itself bad news. Growth into lower-intensity markets lowers the average while raising the total; that is expansion, not decay.
5. Challenge the unit of measurement where the comparison is unfair — an hour of one kind of usage is not interchangeable with an hour of another.
6. Finish on the financials the narrative never mentions: growth, margins, cash flow, buybacks, churn.

• A per-user metric falling while the total rises (almost always mix); management explanations that are specific and geographic rather than vague; and a bear case that survives only if you accept an apples-to-oranges unit of measurement.

21:39 9. Hold a research shop to its own prior claims — and treat absolute probabilities as a tell

1. Before adopting a respected source's conclusion, pull what the same source said about the same subject 6–18 months ago. Consistency across time is the cheapest credibility test there is.
2. Flag absolute language. Analysts who write "practically zero" are estimating; ones who write "zero" — and say they mean it literally — have left the domain of forecasting.
3. Test the causal claim's proportionality: can the stated cause (a handful of departures) really produce the stated effect (permanent impossibility) at an organization of that scale and resource?
4. Check whether the achievement being ruled out was itself a surprise. A company that came from behind once has demonstrated the capability, which sets a floor under the probability.
5. Separate the parts of the report you disagree with from the parts you don't — a flawed headline claim doesn't invalidate the analysis underneath, and the agreed portion may be the part that matters to the stock.

• A source reversing itself within a year without addressing the reversal; probability language that admits no error bar; and a report whose secondary argument (the one nobody quotes) is the one that actually determines the investment outcome.

25:16 10. The deleted-expense screen — when a defence requires removing a real cost, walk

1. When a promoter argues a company is profitable, check which accounting basis they are using and which line items have been removed to get there.
2. Apply the ownership test to each removal: does it reduce what an existing shareholder ends up owning or receiving? Share-based compensation does — new shares are issued, and every existing holder's slice shrinks — so "non-cash" does not mean "not a cost."
3. Distinguish unit economics from company economics. "Profitable per home / per order / per customer" says nothing about whether the enterprise earns money after the overhead required to produce those units at scale.
4. Weigh the behaviour as evidence. Someone who answers critics but goes silent when a credible operator makes the same point is signalling which challenge they can't answer.
5. Then check the outcome, not the argument: pull the multi-year chart of the defended stock. A confident thesis with nothing to show for five years is data.

• "Adjusted," "operating basis" or "per-unit" profitability doing the work in a bull case; dilution rising every year while adjusted earnings look clean; and a promoter who engages weak critics loudly and credible ones not at all.