Market structure · Pattern psychology
A flat ceiling and a rising floor. One side is patient at a fixed price; the other is impatient at rising prices. Everything the pattern does — the grind, the dry-up, the violent break, the retest, the occasional trap — falls out of that single asymmetry.
This is the whole idea, and almost every write-up skips it. The top line and the bottom line are not two versions of the same thing. They are two different kinds of human decision, drawn on the same picture.
The flat top — a price
The ceiling is made of resting limit sell orders. They are passive. They sit in the book at a specific number and wait. The seller is not chasing anything; they decided once, at 100, and went to lunch.
A decision at a fixed price is time-indifferent — and, crucially, finite. There are only so many shares parked there. Every rally that reaches the level eats some of them. Nothing replaces them unless a new seller shows up and makes the same decision at the same number.
The rising bottom — a behaviour
The floor is not a level at all. It's a trajectory. Nobody placed an order on that diagonal line. What the line records is that buyers keep raising the price they're willing to pay, dip after dip.
A rising reservation price is time-sensitive and, unlike inventory, renewing. Urgency doesn't get consumed when it's expressed — it compounds, because each successful defence is fresh evidence that waiting costs money.
A finite stock of patience is being spent against a renewing flow of impatience. That is the entire thesis of the pattern.
Which is why the shape leans bullish: the ceiling can be exhausted but the floor cannot, so the structure has a natural resolution direction. "Leans" is doing real work in that sentence, though — see the odds further down. Plenty of these break the other way, and the mechanism for that is just as legible once you know what to look for.
"Resistance" is not a force field. It is four distinct groups of people who, for four unrelated reasons, all happen to have chosen the same number. They behave completely differently when the level breaks, so it's worth separating them.
The first three groups shrink with every touch and never come back at that price. The fourth grows with every touch and comes back as a buyer. That is why a level that gets tested four times is weaker, not stronger, than one tested twice — the population defending it has been quietly rotating from "finite supply" into "trapped demand."
Step down from the chart to the order book and the grind becomes concrete. Three snapshots: the wall intact, the wall thinning, and the moment after it's gone.
Nothing forces a buyer to bid higher. The lows rise because three separate mechanisms push the buyer's reservation price up at the same time — one emotional, one evidential, one purely arithmetic.
Notice these all point the same way, and notice what's missing on the other side: there is no corresponding mechanism that makes the seller at 100 want to sell at 97. Their whole thesis is the number. If they blink and lower their offer, they've abandoned the idea that made them a seller in the first place — so mostly they don't, and the triangle narrows from below only.
Buyer, bar 53
"Third time it's held above the line and the dip only lasted four bars. I'm not going to get 82 again. If I stop at 91 and it's 92 now, that's a tiny risk for a run to 118. Size up."
Seller, bar 53
"Still hasn't closed above 100 in four attempts. Triple top. My offer is at 100 and my stop is 101 — that's a one-point risk. Why would I move?"
Both are reasoning correctly from their own frame. Both have tiny stops sitting inches apart. That is the definition of a coil: two crowds with opposite theses and nearly identical pain thresholds, compressed into a two-point band.
The mini-charts below are the same series as Exhibit 1, revealed progressively. The highlighted segment is the phase under discussion.
Bars 0–20 · rally into 100, rejected to 82
Price rallies, meets heavy supply at 100, and gets thrown back hard on the highest volume of the whole formation. Nothing is a "pattern" yet. What has actually happened is that the market has just discovered and marked a price where large sellers live, and the sharp drop broadcasts that discovery to everyone watching.
Buyers
Surprise, then doubt. The ones who bought the rally are underwater within days. Some cut. The dip to 82 feels like the trend failing.
Sellers
Vindicated. The rejection was violent, so the level gets a name: resistance. More sellers now plan to sell there — the level recruits its own defenders.
Bars 20–38 · back to 99.7, dip only to 87.6
Price returns to 100 and stalls again — sellers are still there. But the pullback stops at 87.6 instead of 82. This single fact is the pattern's birth certificate. One higher low is not yet a trend, but it is the first piece of evidence that the buyer's urgency has increased while the seller's price has not moved.
Buyers
Quiet conviction forming. "It came back to the high fast. The dip was shallower. Someone is accumulating." The dip-buyer's limit order moves up for next time.
Sellers
Comfortable. Two rejections at the same price reads as a double top — one of the most-taught bearish shapes. Fresh shorts arrive, stops just over 100.
Bars 38–58 · touches three and four, low at 92.2
Now the triangle is visible on any chart, which changes the game: the pattern stops being a description and starts being a coordination device. Thousands of traders draw the same two lines, place buy-stops just above the same ceiling, and put protective stops just below the same rising floor. The structure becomes self-reinforcing because everyone can see it.
Volume falls away. This is the most misread moment in the whole formation — falling volume feels like apathy, like the move is dying. It is the opposite. The participants who are left are only the last patient sellers and increasingly aggressive buyers; everyone else has gone flat because a three-point range offers nothing.
Buyers
Impatience with an edge of greed. Stops are now only 2–3 points away, so positions get larger. Some start buying at the ceiling rather than waiting for a dip that keeps not coming.
Sellers
Splitting in two. The breakeven crowd has mostly filled and left. The remaining shorts are getting nervous — price isn't going down between tests any more, and their stops are inches away.
Bars 58–67 · range collapses; spring through the floor at 61
The range is now two points wide on the lowest volume of the formation. Both stop clusters — longs below the rising line, shorts above the flat line — are packed into a band a car length wide. Whatever happens next happens fast, because the first move triggers a chain.
And very often the last thing that happens before the real move is a false one in the wrong direction. Price cuts below the rising trendline at bar 61, every textbook long is stopped out, the pattern is declared failed — and then it snaps straight back inside within three bars. Wyckoff called the analogue a spring. Mechanically it's a liquidity raid: the cheapest supply left in the market is sitting in the stop orders of impatient longs, and the last buyer takes it before the move.
Buyers
Two populations now. The weak ones just got stopped and are furious, watching from the sidelines with a grudge — they will chase later. The strong ones bought the shakeout at 93 with a very small stop.
Sellers
Brief euphoria on the break of the trendline, then confusion when it doesn't follow through. The short who added on the "breakdown" is now offside almost immediately — a fresh, panicky, tightly-stopped seller sitting right under the ceiling.
Bars 67–73 · 100 → 108.8 on peak volume
People describe breakouts as "buying pressure." It's more specific than that. Here is the actual cascade, in order, over minutes:
Five of those seven are forced or mechanical, not discretionary. That is why breakouts feel disproportionate to the news that supposedly caused them — most of the participants aren't reacting to information, they're reacting to their own risk rules.
Buyers
Vindication into greed. Holders add. The chart is suddenly in every feed. Position sizes taken at the coil are now large and working.
Sellers
Denial, then pain, then capitulation — usually within one session. The short who was "right" at 100 four times is now buying at 106 to make it stop.
Bars 73–95 · back to 100.9, then the measured move
Price falls back to the broken ceiling and, more often than not, holds. In Bulkowski's sample a throwback to the breakout level happens around 64% of the time after an upward break — it is the normal case, not the exception.
"Resistance becomes support" is usually stated as if it were mystical. It is four concrete facts stacked on one price:
The volume signature is the tell: the retest should be quiet. Low volume on the pullback means no urgent supply — just profit-taking and nerves. A retest on heavy volume means real sellers have re-appeared, and that is a different picture entirely.
Functionally, the retest does a job: it transfers inventory from the impatient hands acquired during the breakout panic to patient hands at a better cost basis. Every trend needs to do this periodically or it runs out of people willing to hold it.
You asked specifically how people switch sides. There are four routes, and they differ enormously in how much they move price. The distinction that matters: does the conversion remove supply, or does it remove supply and add demand in the same order?
Which is exactly why failed patterns move so violently the other way. A trader who bought the breakout at 106 and gets stopped at 99 is forced to sell — and a large share of them flip short on the spot, because a failed breakout is itself a recognised setup. Buyer becomes seller by the identical sign-flip mechanism. The direction the market travels is simply whichever crowd's stops are reachable first.
Roughly a third of ascending triangles break downward, and some that break upward fail immediately. Both outcomes have a clean order-flow story — no mystery, no "the pattern lied."
This is counter-intuitive and worth sitting with. If you need to sell a large block, the single best moment is the one where the maximum number of eager, price-insensitive buyers arrive simultaneously. A breakout manufactures exactly that: short stops, breakout entries, momentum systems and chasers all hitting the offer in the same few minutes. A seller with real size doesn't defend 100 forever — they defend it until they can sell more, faster, at 101.5 than they ever could at 100.
Ascending triangles are routinely described as 70–80% reliable. The largest published dataset of hand-identified patterns says something more sober.
Figures from Thomas Bulkowski's study of 1,400-plus ascending triangles meeting strict identification criteria, in bull-market conditions. Treat all pattern statistics loosely: they depend entirely on how the pattern is defined, which market and era was sampled, and a hand-selection process that is hard to replicate mechanically. The number worth remembering is the first one — 63% is an edge, not a certainty, and anyone quoting you 85% is selling something.
The practical consequence: the pattern is not a prediction. It's a map of where the stops are and who is trapped. That map is useful whichever way price goes, because it tells you where the violent move will start and what will fuel it. Traders who do well with this structure are usually the ones who care more about location and invalidation than about direction.
What to actually watch, in the order it becomes visible.
Nothing here is trading advice — it's a description of market mechanics. Patterns describe crowd behaviour; they don't guarantee it, and any real decision needs position sizing and risk rules this page doesn't cover.