Actionable insights — Michael Burry Keeps Buying This Stock

3:18 1. Screen for "priced for capital destruction," not merely for "cheap"

1. Take the forward PE and check what earnings number it is built on. A low multiple on rising EPS is common; a low multiple on shrinking EPS is the rarer, more informative signal.
2. Benchmark that multiple against commodity businesses — miners, refiners, autos, airlines — not against the company's own history or its sector. History anchors you to the mania; commodity multiples are the floor the market reserves for businesses with no pricing power.
3. If a branded, capital-light or high-margin business trades below the commodity band, the market is no longer discounting slower growth — it is discounting value destruction. That is the threshold worth stopping at.
4. Enumerate what the price already contains before you argue with it: name each known problem (deceleration, marketing failure, product misses) and confirm it is baked in. Cheapness is only interesting when the bad news is not news.
5. Cross-check with a second yardstick — free cash flow yield — so you are not fooled by a single accounting-sensitive multiple.

• A multiple that is low only because the E is a one-off peak; a "cheap" name whose cash flow yield contradicts the PE; a business that genuinely is a commodity, where the commodity multiple is correct rather than an anomaly.

8:53 2. Reduce every turnaround to one question: temporary or permanent impairment?

1. List what actually broke, and sort each item into fixable by management (marketing, product design, merchandising, capital allocation) or structural (the customer left, the technology changed, the moat is gone).
2. Separate the brand (or franchise) from the operators. A great asset with bad stewards is a different security than a decaying asset with good ones — and only the first one is a turnaround.
3. Test whether the customer still wants the product. Competitors imitating you is evidence the product is desirable; customers not caring is evidence it is not.
4. Look for prior examples in the same industry where the identical damage reversed — and check the actual price path, not the anecdote.
5. Size the asymmetry: if partial recovery ("get back on track to any degree") already produces a big move, you do not need the full recovery to be right.
6. Accept that this question is not answerable with certainty. It is the question you must have an opinion on — not one the data settles for you.

• Calling a structural loss "temporary" because the brand is famous; a management change with no change in behaviour; competitors whose product is now genuinely better rather than a copy; and the tell in this very episode — Burry is deep in the red and still adding, so "he is sticking with his pick" is conviction, not confirmation.

6:42 3. Treat a depressed multiple as an asset — check the buyback math

1. Ask what the company can do with its own low price, not just what the price says about it. At a high cash-flow yield, every dollar of repurchase retires more stock, so the same buyback becomes more accretive as the shares fall.
2. Measure the current pace: annual share-count reduction as a percentage, from the actual share count trend — not the announced authorization, which is often unspent.
3. Net it against dilution. Heavy stock-based compensation cancels a buyback; low dilution means the reduction is real and the pace can be raised.
4. Check the balance sheet for the fuel — cash on hand, and whether the debt capacity exists to lever a repurchase. Separate genuine debt from lease obligations, which are an operating expense in disguise.
5. Add the achievable buyback rate to whatever EPS growth you assume. A flat business shrinking its float 10% a year still compounds earnings per share at 10%.
6. Confirm this is the board's actual intent — the mechanism only pays if management chooses to run it.

• Buybacks that only offset stock comp (a share count that never actually falls); leverage added to repurchase into a business that is still deteriorating — the fastest way to convert a cheap stock into a broken one; and a company hoarding cash while the shares are at their cheapest.

11:01 4. Test a hardware business for the razor-and-blades (printer-and-ink) engine

1. Split reported revenue into the one-time sale (the printer), the consumable that must be repurchased per use (the ink), and the service/maintenance stream. If management does not disclose the split, that is itself informative.
2. Check whether the consumable is genuinely locked to the installed base — proprietary, single-use, compatibility-gated — or whether a third party can supply it. Only the first is a moat.
3. Find the leading indicator: new units installed per period, not cumulative installs. Every new install begins an annuity; the cumulative number tells you about the past, the new-install rate tells you about the next several years of consumables.
4. Estimate consumable revenue per unit per year and multiply by the base, so you can see how much of "growth" is already contracted by machines already in the field.
5. Then value the recurring portion differently from the equipment portion — the annuity deserves a higher multiple than the box.
6. When the equipment cycle slows, ask whether the consumable stream slows with it. If usage per installed unit keeps rising, a slow equipment quarter is not the same as a slowdown.

• Consumables that generic suppliers can legally clone once patents lapse; a declining new-install rate hidden behind a growing cumulative base; hospitals holding equipment longer in a capex freeze; and — the actual risk in this name — paying a 31 PE / 2.56% FCF yield for an annuity whose growth has already halved from ~20% to ~13%.

16:04 5. Before accepting that a fleet owner kills an asset-light network, price the peak

1. Plot the demand curve across the day/week, not the daily average. Ask how large the peak-to-trough ratio is — that ratio, not total volume, is what decides which cost structure wins.
2. For the capital-owning challenger, force the dilemma: size the fleet for normal demand and it cannot serve the peak; size it for the peak and the excess assets sit idle and unprofitable the rest of the day. There is no third option for an owner.
3. Ask who supplies the incumbent's peak capacity and what it costs them. Capacity contributed by third parties who already own the asset (a car in the garage, a spare room, a delivery vehicle) is free elastic supply the challenger has to buy.
4. Check whether the incumbent is barred from adopting the new technology. If it can simply plug the challenger's technology into its network, the technology is an input, not a competitor.
5. Count the actual deployed units on both sides before extrapolating a launch event into an industry outcome.
6. Conclude on the hybrid: if a mixed network serves both the baseline and the spike, it is a better product at the finish line — not merely a transitional stage.

• A city where demand is flat enough that owned fleets stay utilized; autonomous cost per mile falling far enough that idle capacity stops mattering; the incumbent being locked out of AV supply by exclusivity; and the honest failure mode — the peak argument protects the network's existence, not necessarily its take rate.

17:43 6. Disclose the conflict, then follow the details even when they reverse you

1. State your position in the company before you evaluate news about it — ownership, cost basis, or profits made. It converts a hidden bias into a visible one for you and your audience.
2. Take the headline claim at face value once, and say what it would mean if true. Do not defend a holding reflexively.
3. Then read the underlying facts to the end, especially the ones reported after the accusation. Most news stories bury the disconfirming detail below the lede.
4. Ask specifically what the accused party is alleged to have done versus what merely happened around them — and whether the report even establishes the causal link.
5. Update out loud, in either direction. A view revised on new detail is the process working, not a retraction.
6. Keep the product criticism separate from the investment view; a bad news cycle for a feature is not by itself a thesis change.

• Reversing toward the answer that happens to suit your book (that is the bias, not its cure); a "user error" explanation that quietly excuses a real product defect; and headline risk that does compound into regulation — at which point it stops being a fail-of-the-week and becomes a thesis input.

14:48 7. Use the beat/miss record as a base rate — and low sentiment as the asymmetry

1. Pull the company's last 12–16 quarters of reported EPS against consensus and count the misses. A long unbroken beat streak usually reflects guidance policy, not luck.
2. Estimate the typical size of the beat, not just its direction — a habitual 2–3% beat is a management style you can forecast.
3. Separate the print from the reaction. Ask what sentiment is currently priced in: after a long derating, the bar to surprise positively is low.
4. Look for the catalyst that changes the narrative rather than the number — guidance, a segment reaccelerating, a buyback — since that is what re-rates the multiple.
5. Note the asymmetry explicitly: when expectations are depressed, an in-line-to-slightly-better print can move the stock hard, while a miss is partially discounted already.
6. Never treat the streak as a guarantee; treat it as a prior you are willing to state in advance and be graded on.

• A beat streak maintained by guiding down each quarter (the beat is engineered and the stock still falls); a beat on EPS with weak revenue or guidance; and the trap of buying for the print — the process here is a stated prediction to be graded, not a position sizing rule.