Part 1 (Climax Run), Part 2 (Story is weaker than you think), Part 3 (Semi-Sell Rules)
Stage analysis says topping. But the 1998 analog and Livermore's hand-drawn cycle both say we're at point 6 — shakeout next, melt-up after. So we stop predicting and use the twelve rules that work either way.
«If the job has been correctly done when a stock is purchased, the time to sell it is almost never.»
Phil Fisher, investor (1907–2004)
We have arrived at the hardest question in investing.
In Part 1, the chart said late-stage: a three-stage parabola, semis at 22% of the index, record flows, narrowing leadership. In Part 2, the story cracked: circular revenue, cyclicality dressed as secular, depreciation borrowed from tomorrow.
Both halves point to caution. And yet I am going to spend the first half of this piece arguing against myself — because the honest truth is that this move could run far higher before it ends, and any study that ignores that is propaganda, not analysis.
Then I am going to give you the resolution. Not a forecast. A plan. The twelve sell rules that have guided me for years — the ones that do not require me to know whether this tops next month or in 2028, because they respond to what the market does, not what I predict.
Buying is easy. This is the part that separates investors from tourists. Or bag holders.
Stage Analysis: The Topping Case
In Part 1 we used CAN SLIM’s three-stage parabola. Now let us bring in the more complete map: Stan Weinstein’s four-stage model, which we covered in «Stan Weinstein’s Secrets for Profiting». It is the cleanest framework ever drawn for locating yourself in a trend, and its single tool is the 30-week moving average — which, on a weekly chart, is essentially the 200-day moving average in different clothing. That matters, because it is the same long average my sell rules lean on later in this piece: the 30-week and the 200-day are two names for the line that separates a healthy trend from a broken one.
Chart 1: Stan Weinstein’s four stages — the complete lifecycle of a trend
Weinstein four stages — 30-week MA
Stage 1 — the base. Sideways, ignored, range-bound after a decline. The 30-week average flattens; volume is dead; the public is bored or bitter. Smart money quietly accumulates from discouraged sellers.
Stage 2 — the advance. Price breaks out above the base on expanding volume and rides above a clearly rising 30-week average, making higher highs, outperforming the market. Volume swells on the up-weeks and dries up on pullbacks. This is the part you want to own.
Stage 3 — the top. The advance stalls. The 30-week average flattens and begins to roll over. Price action turns erratic and choppy — wider swings, failed breakouts, no net progress — and the volume signature inverts: now the down-weeks carry heavier volume than the up-weeks. That is the fingerprint of distribution — institutions selling into a still-euphoric public, the early money handing stock to the late money, while the headlines stay bullish.
Stage 4 — the decline. Price below a falling 30-week average, lower highs and lower lows, relative strength collapsing.
Here is the crucial bridge between the two frameworks, because it is exactly where we are. CAN SLIM’s “third-stage climax run” is the same event as Weinstein’s late Stage 2 tipping into Stage 3. O’Neil’s vertical blow-off is the overheated end of Weinstein’s advance — the moment price stretches farthest above the rising 30-week average right before that average flattens. Two great technicians, two vocabularies, one phenomenon. When you see them as the same map drawn twice, the picture sharpens rather than blurs.
So where are the chips?
Semiconductors are in a late, stretched Stage 2 — the climax phase in O’Neil’s language — with the first fingerprints of Stage 3 beginning to appear: the vertical price far above the 30-week line, the record inflows, the euphoric volume. The one Stage-3 confirmation we have not yet seen is the decisive one — the 30-week average flattening and the down-week volume overtaking the up-week volume. Until that happens, this is still technically Stage 2. That is exactly why this is the most dangerous part of the cycle to chase, and also why it is not yet, definitively, over.
That is the topping case, and it is a strong one. If I stopped here, the conclusion would be simple: trim, protect, prepare for Stage 3.
But I am not going to stop here, because there is a second chart, and it tells a very different story.
The 1998 Anomaly: The Reacceleration Case
Chart 2: Today’s Nasdaq overlaid on the 1994–1998 run — a potential roadmap into 2028
Nasdaq 2022-2026 vs 1994-1998 overlay
This maps the current Nasdaq onto the mid-1990s. Different decade — same shape, so far. And if the rhyme holds, the script is not «top now.» It is something far more interesting.
Cast your mind back to 1998. The Asian financial crisis and the collapse of Long-Term Capital Management took the market down sharply. It felt, at the time, like the end of the cycle — exactly the kind of late-Stage-2 exhaustion the bears were calling. And then the Nasdaq did not roll over into Stage 3. It reaccelerated — melting up for another eighteen months into the 2000 peak, in one of the most violent advances in market history. The exhaustion was not the end. It was the pause before the blow-off.
The overlay suggests the same shape could be playing out now: a sharp, frightening drop into the back half of 2026 — the kind that feels terminal — followed by a final, parabolic blow-off into 2028. With semis leading again, because the most cyclical, most stretched sector is precisely the one that leads the melt-up out of the shake.
Now the honesty, because you have come to expect it: this is a chart laid over another chart. Analogs fit beautifully right up until the moment they don’t. The market does not owe 2026 the courtesy of repeating 1998. Every overlay you have ever seen looked compelling in the window where it worked. 1998 is a possibility, not a prophecy.
But it is a possibility serious enough that calling an outright top here would be reckless. Stage analysis says topping. The 1998 analog says maybe the opposite. Both are credible. Both are drawn by honest people looking at the same tape.
And there is a second, older roadmap that says the same thing the 1998 overlay does — drawn by hand, a century ago.
The Most Important Number in the Market Is 6
Jesse Livermore mapped the speculative cycle by hand, point by point, from the first spark to the final danger signal. Volume and price, ten numbered stages. It is crude, it is a century old — and it is one of the most useful things ever drawn, because it does not just tell you that a top is coming. It tells you roughly where in the sequence you are standing.
Chart 3: Livermore’s hand-drawn speculative chart — the «all-important action» of a stock
Livermore speculative chart
Chart 4: Livermore’s written speculative cycle — volume and price, point by point
Livermore speculative cycle — handwritten
Walk through it with me, because the punchline is a single number.
Points 1 to 5 are the build. Volume surges, price climbs, pulls back, climbs again — the accumulation nobody notices until it is well underway. We have lived exactly this since 2023.
Point 6 is now. Volume +++, price +++. The intense, high-volume push where everything works and conviction peaks. Semiconductors melting up on record inflows is the signature of this precise point — not the end, but the hot, loud advance that runs just before the shake.
Point 7 is what comes next. Volume drops, price drops. Profit-taking. In Livermore’s own handwriting: a possible change of leadership, the original bulls exiting. The shakeout. And — read this against the 1998 overlay — that is the exact drop into late 2026 the analog projects. Two independent century-apart roadmaps, pointing at the same pullback.
Points 8 and 9 are the melt-up. Volume +++, price +++ again, to new highs. The final, violent leg that only the patient survive to see — the blow-off the 1998 analog calls into 2028.
And point 10 is the one to remember. Price still rising, but in the last few hours: a sudden break. The danger signal. That is where you sell.
Now hold both charts in your head at once. The 1998 overlay and Livermore’s cycle were drawn decades apart, by different men, from different data — and they tell the same story. We are not at the danger signal. We are at point 6. The buy, if it comes, is the point-7 shakeout. The melt-up is points 8 and 9. The danger signal does not ring until 10.
This is why we lean, on balance, constructively — why we think the genuine risk is being shaken out at point 7 and missing points 8 and 9, not riding point 6 into a 2026 top. It is entirely plausible that this cycle carries the S&P 500 toward 10,000 into the back half of this decade. We hold that view loosely — point 6 can fail, analogs break — but two hand-drawn maps a century apart agreeing is not nothing.
And yet — notice what just happened. I have now given you three frameworks. Stage analysis says trim. The 1998 overlay says hold for the melt-up. Livermore’s cycle says we are at point 6 with the best still ahead. They do not fully agree, and the honest truth is that no amount of chart study will make them agree, because the future has not been drawn yet.
So which roadmap do you bet on?
That is the wrong question. And the whole point of this finale is why it is the wrong question.
The Tyranny of the Unanswerable Question
Spend any time around investors right now and you will hear the same conversation, in a loop. Is this the top? Is it 1999 all over again, or 1998 all over again? Are we at Livermore’s point 6, or his point 10? Do I sell, or do I hold for the melt-up?
It is the most natural question in the world. It is also a trap — because it is unanswerable, and building your actions on an unanswerable question is how you end up frozen, or worse, flip-flopping at exactly the wrong moments. Three roadmaps lean constructive; stage analysis leans cautious; none of them can be proven until after the fact.
Watch what the unanswerable question does to people. They sell the whole position on the first scary down-week, convinced Stage 3 has arrived — and then watch the melt-up resume without them, and chase it back in, higher, just in time for the real top. Or they hold everything, white-knuckled, through every warning sign, because selling would mean admitting the run is over — and they give the entire gain back in the decline. Both behaviours come from the same root: trying to answer a question that has no answer, and letting the answer drive an all-or-nothing decision.
The professionals who survive the end of great trends do something different, and it is almost boring. They stop trying to answer «is this the top.» They accept that it is unknowable. And they replace the prediction with a process — a set of mechanical responses calibrated so that they do the right thing in each scenario without ever having to know which scenario they are in.
This is the deepest idea in this entire trilogy, so let me state it as plainly as I can. A forecast has to be right. A rule only has to be followed. A forecast is a single bet on an unknowable future. A rule is a conditional: if the tape does X, I do Y. You do not need to know whether X will happen. You only need to have decided, in advance and in cold blood, what you will do if it does. The forecaster is paralysed by uncertainty. The rule-follower is liberated by it, because the rules have already absorbed the uncertainty on his behalf.
That is the mental shift. Now the machinery.
Stop Predicting. Start Following Rules.
Here is the liberating truth that took me years and real money to internalize.
You do not need to know whether this tops in 2026 or melts up into 2028.
The investor who survives the end of a great trend is not the one who calls the peak — nobody calls the peak. It is the one who decided, in advance and in cold blood, exactly what they would do in each scenario, and then did it, mechanically, while everyone around them was paralysed by the same unanswerable question.
Stage analysis topping? The rules get you out. 1998 reacceleration — or Livermore’s point-6-into-point-9 melt-up? The rules keep you in — with a trailing stop — through the point-7 shakeout and the final leg, and then get you out at the danger signal. The same rules handle every one of these roadmaps. That is why a plan beats a forecast: a forecast has to be right; a plan only has to be followed.
So here they are. The twelve sell rules I have published and refined over the years — applied directly to a semiconductor position right now. (For the full reasoning behind each, see «When Should You Sell Your Stocks? 12 Rules Show the Way».)
First, the Two Starting Points
Before the rules, the frame that organises them. There are only two reasons you ever sell, and they feel completely different.
The first is selling out of strength — taking profit while the position is still rising. This is the pleasant one, and the one almost nobody does enough of, because it means leaving a winning party early. The second is selling out of weakness — cutting a loss, or refusing to give back a gain. This is the painful one, the back-against-the-wall sell, and it is where fortunes are actually preserved or destroyed.
Here is the crucial insight, and it is why most investors’ sell discipline fails. Most people build their sell rules entirely on fundamental reasons — earnings, valuation, the story. But fundamentals are slow. By the time the deteriorating fundamentals are obvious in the income statement, the market — which prices six to eighteen months ahead — has already taken the stock down 40%. The fundamental seller is always late, because the tape moves first.
That is why my sell rules lean heavily on the technical — price and volume, not earnings. Not because fundamentals do not matter, but because in a topping market the chart is the early-warning system and the income statement is the post-mortem. Remember Alibaba: revenue grew over 1,000% from its IPO, the valuation got cheaper the whole way down, and the stock still turned a gain into a loss. «Markets are never wrong,» as Livermore said. «Opinions often are.» The price is the verdict. The rules below listen to it.
Now the twelve.
