Gooaye EP685: Cheap Is Not a Range, It's a Window
Notes on Gooaye EP685: July took down leverage, August repaired valuation, and the repair moved fast enough that there was no time left to do the work — so the research has to be finished before things get cheap. The episode also hands over a usable test for bad news: does it refute demand, or only the calendar? Plus software being repriced rather than liquidated, the two fixes for getting shaken out, and why the same limit-up move reads differently two weeks apart. Educational notes, not investment advice.
The skilled warriors of old first made themselves impossible to defeat,
and then waited for the enemy to become possible to defeat.
Being impossible to defeat lies with yourself; the enemy’s defeat lies with the enemy.
So a skilled warrior can make himself impossible to defeat,
but cannot make the enemy’s defeat a certainty.
—— Sunzi, The Art of War, “Dispositions”
These are my personal notes on Gooaye EP685 (released 2026-08-05). This is not a transcript and not official content. Please support the original show for the full episode.
What the episode is about
In one line: July took down leverage, August repaired valuation, and the repair moved fast enough that there was no time left to do the work.
The first half is the market. Small and mid caps gave up in one month roughly what a whole bear year once took, and then took back a large slice of it in a handful of sessions. The host’s emphasis is not that he called it — it is the speed itself. The middle section takes apart the scare stories circulating that week, and the common conclusion is that a slipped schedule is not vanished demand. The listener questions return to temperament: how to stop being the person who capitulates at the low, why the friend who buys the top and sells the bottom is worth keeping around, and why the same limit-up move gets read two different ways two weeks apart.
The Sunzi passage fits. Everything actionable in this episode lands on the “lies with yourself” half — cut leverage, cap single-position weight, write the exit before you need it. When the rally comes back, and whether there is a second leg down, belongs to the other half. That one is not yours, and it cannot be computed.
The main points
One: the speed of the repair is itself the information, and a smile curve is the realistic expectation. When good assets genuinely get cheap, the repair runs faster than you can react — limit-up moves that lock you out entirely. That is usually how the low gets away from you. Conversely, nobody fills their whole position at the bottom. The realistic shape is: you buy some and are immediately underwater, one lot happens to land near the low, and you chase another one or two on the way back up, averaging out somewhere acceptable. Grading yourself against “all-in at the exact low” only ever returns a failing grade.
Two: he refuses to assume a second leg down. His observation is that the tempo has visibly accelerated since 2020 — straight V and A shapes have become the norm, and the classic double bottom shows up less often and with a much fainter signal. He recalls one selloff that printed only two or three token down bars before leaving; the people who shorted those bars got run over. So his conclusion is not “it must rally from here.” It is that the plan “sell now, re-enter once the second leg prints” is itself a bet — on a pattern that has become rare.
Three: three pieces of bad news, one shape — the calendar was refuted, not the demand. Pulling back the stack height on high-bandwidth memory checks out, but the nature of it is a cost-and-yield trade-off: the higher you stack, the worse the yield and the higher the cost. A chip’s ramp timing may slip a quarter, which is the ordinary elasticity of a ramp. And on datacentre power architecture, the high-voltage spec has not disappeared — the intermediate spec simply lives longer, while the customers who were going to migrate still migrate. None of the three touches whether the thing is wanted. All three touch when.
Four: with geopolitical headlines, price the actual shipping route before you price the fear. On the rumour that optical components out of one region might be restricted, his read is that the real impact is far smaller than assumed, because the leading vendors’ North American volume already ships from overseas plants — which is why those names did not collapse. Rational money was pricing the route, not the headline. The line he does think is worth tracking is retaliation: if a key upstream material gets choked, that shakes the whole chain. But with demand this tight, nobody has much incentive to let it harden.
Five: AI did not kill software, but software’s cost structure changed permanently. Software used to carry almost no marginal cost — build it once, sell it endlessly. Once it is deeply coupled to a model, every output burns compute and carries a real cost, so the margin compression is structural rather than temporary. He does not buy the “future software gets generated on demand” line either: enterprise buyers need stability and reproducibility, and cannot accept a fresh dose of randomness on every run. A personal gadget, fine; a business system, no. His favourite framing came from a listener — AI is a legendary set of kitchen equipment, plenty of people claim to have legendary dishes, and yet remarkably few legendary dishes have actually reached the table. So the conclusion is that thin wrappers and single-function tools with no data integration get cleared out. This is a repricing, not a liquidation.
Six: “the expensive stuff holding up” is a thermometer, not a recommendation. What he watches is not whose earnings looked good — that part held no surprises — but whether the market is willing to push an already-expensive multiple higher after the print. If the most expensive slot holds, the valuation ceiling for everything beneath it has room to lift; if it spikes for a day and gives it all back, the meaning inverts. He uses the same logic on his own change of stance: cautious a month earlier because he did not know what was being sold, then willing to add back once the driver turned out to be a leverage unwind and a sharp drop in margin balances.
Seven: two fixes for getting shaken out, and why the same limit-up reads differently. A listener sold at the low the day before the bounce and asked whether he is simply unsuited to this. The answer was inexperience, not incapacity, plus two routes. One is to respect the technicals and cut earlier — most of those positions had broken their moving averages and bled lower long before; you held on until a psychological threshold snapped, and that particular sale is the least rational one available. The other is to build allocation in — smaller single-position weight, rebalancing to absorb the swings, and as little leverage as possible. As for “why did your view change,” his answer is that a judgement is made against a specific information set, and when the set changes the judgement should change; markets are not about winning an argument outright, and what you are actually training is the ability to adjust. That answer is right, but it has a dangerous side effect: it converts very easily into a universal retroactive excuse. So I add one constraint — write the information set down at the moment of the decision, so there is something to grade later.
Going further
1. Finish the research before things get cheap
The operational takeaway here is not “buy the dip.” It is that valuation repair moves fast enough to leave you no research time.
My old sequence was wrong: wait for the drawdown, then open the filings, run the valuation, and look for a reason to buy. The problem is that the window is a few days wide, and in the middle of a panic you are being asked to do three things at once — finish the analysis, override the fear, and commit. None of the three goes well in that moment, and all three draw on the same pool of attention.
So the sequence became: write the list first, the price first, the reason first. Keep a standing list of “companies I believe in that are simply too expensive right now,” each with three things pinned down in advance: why I believe in it, at what price I start buying, and what event strikes it off the list. When the selloff comes, the only thing left is execution.
It also quietly solves the “I couldn’t pull the trigger” problem. Freezing in a panic is usually not timidity — it is that the judgement is only just beginning to form. Form it early and what remains is discipline.
2. For any piece of bad news, ask what it actually refutes
All three stories in this episode share a shape, and so did the market’s first reaction: sell first. That is worth hardening into a test — for any supply-chain headline, ask whether it refutes the demand or only the schedule.
If demand is refuted (customers stopped buying, another technology took the socket, the budget was cut), the thesis itself is broken and must be rewritten rather than rationalised. If it is only slippage (yield is not there yet, cost forces an intermediate spec, the ramp moves a quarter), then nothing about your reason for owning it was overturned — only your calendar was. And the calendar was never why you bought it.
Why not read it the other way and say repeated delays are a demand problem? You can, but that needs separate evidence: either budget visibly moving to a competing approach, or the intermediate spec turning into the destination instead of a bridge. “Delayed again” on its own does not get you to demand refutation, and skipping that step is exactly how people end up re-arguing the thesis on every spec headline.
In practice, write the falsifier at the same time you write the thesis: “here is what would make me say this bottleneck no longer holds.” Without that sentence, the test is a nice-sounding taxonomy rather than something executable. Worth noting too: the line genuinely worth tracking — an upstream material getting choked — rarely comes with a headline, because it is not dramatic enough. That one does touch demand.
3. Whatever can be written into structure should not be left to willpower
The two fixes for getting shaken out look like alternatives. They are the same move: write the volatility you can absorb into the structure of the position, instead of handing it to your willpower in the moment.
Cutting early moves the “I can’t take this” signal forward, replacing the breakdown moment with an objective price condition. Allocation makes any single holding’s decline too small to trigger the breakdown at all. Both relocate the decision away from peak emotion. And there is an easily-missed piece of arithmetic here: your leverage multiple directly determines how deep a drawdown you can carry. On the same 40% decline, the unlevered holder is uncomfortable and the 2x holder is out. “I was right” and “I lasted” are independent, and the second one can be computed in advance — take the largest paper drawdown you can genuinely tolerate, divide by your leverage, and that is your real margin for error.
The opening anecdote is the same idea in another register. A family member got into a street argument with a stranger, and he was dragged into a pointless dispute an hour before a film he had been waiting months to see. Afterwards he landed on a line: happy people live very boring lives. Nobody is dumping trouble on you, nothing has to be handled right now — you get to feel bored, and that is what genuine quiet feels like. He admits his pursuit of quiet has become close to a compulsion, and that it is not coldness; it is that having once been sensitive to everything, he knows what happens to the work when attention gets split.
Translated to investing: your attention is a finite position, and the market asks for it every single day. If every scare story in a given week makes you re-argue your holdings, what you burn is not just time — it is the judgement you have left at the moment a decision actually matters. So I sort inputs into two layers. Structural: anything that changes how the company makes money, or changes supply and demand — stop and recompute. Noise: anything that might move the price this week but not the picture in three years — record it, do not act on it. When the two are hard to tell apart, one crude question works: if this is true, does it change my reason for holding? If not, it is noise.
The interesting twist is that the episode’s “contrarian friend” runs the opposite way: do not switch the noise off, wire it to the right instrument. The friend’s emotions carry no predictive power on their own; the issue is that his entries land systematically at the extremes — heaviest when the trend is most obvious, capitulating when there is nothing left to hope for. He is not stupid. He is a well-calibrated thermometer. That does not contradict the point above: one is an input to a decision, the other is an instrument for reading temperature, and an instrument should never be mistaken for a signal source.
Which is what “being impossible to defeat lies with yourself” adds up to. You do not get to decide when the rally comes back. You do get to decide whether you are still on the field when it does.
Worth reading alongside
- Gooaye EP685 (released 2026-08-05) — the source of these notes. Listen to the original show in full and support the creator
- Sunzi, The Art of War, “Dispositions” — the origin of “first make yourself impossible to defeat, then wait for the enemy to become possible to defeat”
- Tao Te Ching, chapter 16 — “Attain the utmost emptiness; hold fast to stillness. The ten thousand things rise together; I watch them return” — a useful companion to the noise-management thread and the “boredom is quiet” passage
Disclaimer: These are personal notes and learning material, written for educational purposes. They do not constitute investment advice, an offer, or a solicitation. No specific security is recommended and no price target is given; no company, industry, or product mentioned in the episode is evaluated or endorsed here. Individual tickers are deliberately left unnamed, and any related description exists only to illustrate reasoning. The observations, figures, and experiences in the episode are the host’s own account; they are quoted here to illustrate a method of reasoning and have not been — and cannot be — independently verified. Investing carries risk, past performance does not indicate future results, and you should reach your own conclusions based on your financial situation and risk tolerance, consulting a qualified professional where appropriate. The author may hold positions in the types of assets discussed.
This article is an educational discussion of investment method. It is not advice to buy or sell any individual security, offers no target prices, and does not analyze any current holding. Investing carries risk; make your own decisions or consult a qualified professional.