Home › Learn › Lesson 36

What price action is made of — trendline, levels, candles, fib and volume

By Cuongboss · Updated 5 Oct 2026 · Published 5 Sep 2026

Most introductions to price action give you the list and stop there, as though five tools meant five opinions and your job were to wait until all five nodded. This lesson takes the list seriously enough to ask a duller question about it: what does each of these five actually read? The answer reorganises the whole subject. Four of them read one column of numbers and one reads another, two of them can never be true at the same time, and the course’s own rule for spotting a dead trend turns out to carry a price you can calculate exactly rather than estimate.

The five components on one screen. Note the dividing rule above the volume panel: it is not a layout choice, it is the lesson. Everything above the rule is a reading of the same numbers; everything below it comes from somewhere else. Schematic only — every measured claim below sits on a chart drawn from its own figures.

What exactly is price action made of?

Five things, and our course states them as a closed list rather than a starting point. The slide is titled What price action is made of, and beneath it are five items and nothing else: TRENDLINE, SUPPORT / RESISTANCE, CANDLESTICKS PATTERNS, FIBONACCI RETRACEMENT LEVEL, VOLUME. There is no sixth item, and no invitation to add one.

A closed list is more useful than an open one, because it tells you when to stop shopping. Each of the five already has a lesson of its own here, and this lesson deliberately does not repeat them: how to draw a trendline and its channel, how to find support and resistance, what individual candlestick signals mean, how a Fibonacci retracement grid is drawn and priced, and how to read volume. What has never been asked on this site is the question that the list itself raises.

Ask it about any one of the five: what numbers does this thing consume? A trendline is a straight line through two or more swing lows — swing lows being highs and lows of bars. A support level is a price where bars turned before. A candlestick pattern is the open, high, low and close of one to three bars. A Fibonacci retracement is a swing low, a swing high, and arithmetic fractions of the distance between them. Every one of those is a function of the same four columns: open, high, low, close. Volume is not. Volume is a second column of numbers that the exchange records alongside the price, and nothing about it can be derived from the price series at all.

That is the structure of the subject, and it has a consequence people rarely state: when four tools agree, you may be hearing one fact four times. Whether that is a small effect or a large one is not something to reason about from the armchair. It is measurable, so the next section measures it.

Why do four of the five agree more often than they should?

Because they are arithmetic functions of the same input, and that alone makes them co-occur — no market psychology required. To put a number on it, we generated a price series that contains no trends, no memory and no crowd behaviour at all: a Gaussian random walk of 400,000 bars at σ = 0.4% per bar, seed 20260905, of which 399,900 bars were usable once the lookback windows were filled. Alongside it we generated a volume series drawn independently by construction, which serves as a control: if our measure is sound, volume must come out independent, and if it does not, the measure is broken rather than the finding interesting.

Each component was given one plain, checkable bullish reading:

The measure is the overlap multiplier, an identity this site established in Lesson 23: the frequency with which two conditions are true together, divided by the product of their separate frequencies. A multiplier of 1.000 means genuine independence — the two are telling you separate things. Above 1.000 means they co-fire; the surplus is double-counting.

Two things are worth sitting with. The first is that the volume row comes back at 1.000 to three decimal places across all four pairings, which is the evidence that the multiplier is measuring what it claims rather than manufacturing patterns. The second is that the entanglement among the price-derived four appears in a series with no structure in it whatsoever. Nobody was trading this data; it has no trends to synchronise the tools. The overlap is algebraic, so it should survive on a real chart. That is a claim worth checking rather than asserting, so we checked it.

Does the same test hold on real candles?

Mostly yes, with one correction to how this lesson first put it. We ran the identical five definitions on Binance spot four-hour candles for BTC, ETH and SOL, from January 2020 (SOL from its August 2020 listing) to 26 September 2026 — 14,665, 14,665 and 13,327 usable bars, closed candles only. For volume we added a second control: the same real volume shuffled into random order 300 times, which keeps its size and spikes but breaks any link to price. Shuffled volume against a broken high lands between 0.95 and 1.05 in 95% of shuffles on all three coins.

Three things changed and one did not. The price tools still double-count heavily on real candles, but less than in the random walk — 1.54–1.62× rather than 1.856× — so the model overstated the overlap a little; an earlier version of this page called the simulated figure a floor, and on these three coins it is not. Real volume is not independent: it shows up above average on broken highs 7–11% more often than chance, outside the shuffled band on every coin, and slightly avoids the pullback zone — loud breakouts, quiet pullbacks. But that 7–11% excess is a fraction of the 54–62% the trendline and the level share. And the exclusion between a broken high and the Fibonacci zone is exactly zero on real data too, because it was never a statistical result in the first place. On daily candles (about 2,100–2,400 per coin) the price-tool overlaps point the same way, but there are too few bars to separate volume’s small link from noise.

Which two of the five can never agree at all?

Support/resistance and Fibonacci, as they are ordinarily used. Their overlap multiplier is not merely low; it is 0.000, and the two conditions were true together on 0.0000% of 399,900 bars. This is not a quirk of the sample. It is a matter of definition: one condition says price is above the swing high, the other says price is between the swing low and the swing high. Both cannot describe the same bar.

That single zero rearranges how the list should be used. The five are not a checklist to be ticked off, because ticking all five is not merely rare, it is impossible. They are a toolbox, and you have to choose a reading before you know which tools apply:

The asymmetry in that last column is the part worth carrying away, and it runs against intuition. The breakout reading — the one that feels unambiguous, because a broken high is a visible event — overlaps itself by 2.32×. Its three green lights are worth substantially less than three. The pullback reading, which feels vaguer, comes in at 0.95×: near-perfect independence, three green lights that are close to three separate pieces of evidence. Confidence and information are running in opposite directions here. On real four-hour candles the gap narrows but keeps its direction: the breakout reading overlaps itself 1.88–1.93×, the pullback reading 0.86–1.03×.

None of this says the pullback reading wins more often. It says that per tool consulted, the pullback reading gathers more distinct information. What you do with that is a question about how much you risk, not about which setup to prefer.

How do you read the end of an uptrend from price action?

Our course splits this into a warning and a confirmation, and keeps them strictly apart. The warning is described on the slide as an uptrend that “sets two highs with the second no longer far above the first”, with buying that has “weakened a great deal”. The confirmation is blunter: “as soon as price breaks the first support of that uptrend, the reversal is officially confirmed.”

The phrase that people misread is the first support of that uptrend. It does not mean the bottom of the whole trend. It means the most recent higher low — the point from which the final push started. Get that wrong and every number below changes.

Here is the warning and the confirmation on a chart most readers will remember: Bitcoin’s run to its August 2025 high, the same daily candles as the screenshot in Section 1, now measured with the chart’s price-range tool.

Now the part that is not in the usual write-ups. Wait for the confirmation and you give up the distance from the final high down to the broken level: 124,474 − 111,920 = 12,554. Compare that with the length of the final leg: 111,920 → 124,474, which is 12,554. They are the same number, and not by coincidence. The level you are waiting for is the level where the final leg began, so the give-back is that leg’s full length — 100% of it, always, by definition rather than on average.

Follow that through and you get a counter-intuitive result. Suppose the trend had died one leg earlier, after the 98,200 → 123,218 push. The first support would then have been 98,200, and confirmation would have cost the whole 25,018. Because the trend instead produced a weaker second leg, confirmation cost 12,554 — 49.8% less.

Patience and cost are usually presented as a trade-off: the longer you wait for certainty, the more it costs you. Here they move together. The exhaustion signal that makes you suspicious — a short, tired final leg — is the same fact that makes waiting for proof cheap. When a trend ends with a genuinely feeble last push, waiting for confirmation is close to free, and there is little reason to jump early.

The third row is the part a definition cannot promise. “Breaks the level” is a price; “closes below the level” is a candle, and a candle has to finish somewhere. On 25 August it finished 1,808 below the line — another 14.4% of the leg on top of the 100%, or 0.58 of an average daily candle (3,130 over the previous 20 days). That extra is simply the cost of asking for a close rather than a touch — price had already traded below the line on 22 and 24 August without closing there. (What to make of breaks that do not hold is Lesson 38’s subject.) It has a ceiling you can know in advance: when the breaking candle opens above the line, as this one did at 113,493.59, the overshoot cannot exceed that candle’s own range. Bigger candles on your timeframe, bigger closing premium.

What is an “area of struggle”, and what is it worth?

An area of struggle is our course’s name for what happens when price fails at a level above and then, in its words, “starts to struggle and to trade tight against the lower support of the pattern”. Two things happen at once: the highs stop reaching the level that rejected them, and price keeps leaning on the same floor. The range between them closes.

What follows, per the slide, is that “the moment price breaks that support, a sharp move appears and a very fast sell-off follows. Sellers pour in volume and overwhelm the buying completely.” Note where the fifth component finally does its work: the volume spike is what separates a break that means something from a drift across a line. It is the only one of the five capable of adding a fact the price bars had not already stated.

Here is one on the live chart, a month after the daily reversal above: BTC/USDT on the four-hour candles, 19 to 22 September 2025.

Now price it, using only what was on the screen when the 00:00 candle closed. The zone was 1,021.81 tall (116,121.81 down to 115,100), so the measured move from the floor points to 115,100 − 1,021.81 = 114,078.19. The short is entered at the breaking candle’s close, 114,649.89 — and that close is already 450.11 below the floor: 44% of the measured move is gone before you are in. Then the stop:

Both rows are poor trades on paper, and that is the honest finding: by the time a four-hour candle has closed through the floor, the measured move — the standard textbook target — has largely been spent. The stop is not the problem here. Measured in the candles you had been watching, the tighter stop is 1.69 candles, which looks comfortable; measured in the candle that actually broke the zone, it is 0.73 — less than one bar of the kind that turns up on the break. Converting a stop into candles, the unit Lesson 35 introduced, only works if you count in the right candles.

What actually paid in this example was the part of the slide that no measured move captures: “a very fast sell-off follows. Sellers pour in volume.” The next candle traded down to 111,800 — 3,300 below the floor, more than three zones’ worth — on 6.44× volume, and neither stop was touched in the following 24 hours: the highest price after entry was the entry itself. One example is not a rate. It shows what the course’s description looks like on a real screen, and that in this case it was the volume, not the R:R column, that carried the outcome. What a tightening range is worth as a rule of its own — the course’s fourth rule of price action — is priced in Lesson 37.

Which component is actually worth adding next?

Volume, and by a margin that is measurable rather than a matter of taste. Start from a bar that already satisfies two conditions — the trendline reading and a broken old level — which together describe 13.928% of bars. Then add one more filter and see how much of that population it actually removes, compared with how much its own rarity says it should remove.

An ideal, independent filter keeps precisely its own frequency of whatever you point it at. In the random walk, volume does that to within 0.05 of a percentage point. On real four-hour candles it keeps 41.2–41.6% against its own 38.3–38.8% — a small leak, because real volume likes breakouts — while the candle keeps 38.5–39.6% against 32.2–32.4%. In the model the candle keeps 42.53% where its rarity promised 34.07%, meaning about an eighth of the filtering it promises is work already done by the level condition — the same 1.263× from Section 2, seen from the practical end.

The lesson for anyone assembling a routine is unglamorous: the fourth price tool you reach for is the one that adds least, because by then you have read the same numbers three times. This is the same shape of finding that Lesson 23 reached about indicators, arrived at from a different direction — there by algebraic identity between a moving average and RSI, here by shared input across the price-action toolkit.

When is everything above wrong?

The measurements come from a model, and the model’s limits are specific and worth stating plainly rather than in a footnote.

Volume was independent in the simulation because we made it so — on real candles it is not, though by less than you might fear. On BTC, ETH and SOL four-hour candles, above-average volume co-occurs with a broken high 1.07–1.11× (shuffled volume: 0.95–1.05) and slightly avoids the pullback zone at 0.94×. Real volume carries some of the information the price bars carry. Its excess is still a fraction of what the price tools share among themselves (trendline × level: 1.54–1.62×), so the ranking in this lesson stands; what changes is that volume is a mostly separate witness, not a perfectly separate one.

The multipliers depend on the definitions we chose. A 20-bar trendline test and an 80-bar level test are reasonable but arbitrary; lengthen or shorten them and the numbers move. What does not move is the sign: tools computed from one series cannot be independent of each other, whatever windows you pick. The 0.000 is stronger still, since it follows from the definitions rather than from any window length.

None of this measures whether anything wins. There is no hit rate on this page, because none was measured: the real-candle tests count how often conditions coincide, not what happened next, and the two Bitcoin charts are single cases chosen to show what the course describes. Every percentage here is a frequency, a measured distance or a break-even threshold — the win rate a decision would need in order to be worth making.

A confirmed reversal is not a forecast. The 25 August 2025 close below 111,920 did exactly what the course says: it confirmed that the uptrend which ran from 98,200 had ended. Six weeks later Bitcoin traded through 124,474 on 5 October and closed above it on 6 October, on the way to a new high of 126,199.63 — a new uptrend had been built. Only after that did price fall to 80,600 by 21 November. A short taken on the August close with its stop above the final high would have been stopped out on 5 October, before the decline. The rule told the truth about the trend that existed; it promised nothing about the next one. That is why it belongs in a reading of structure and not in a trade plan on its own.

And the struggle-zone arithmetic assumes you are filled at the close. A market order into a fast break is exactly where slippage is worst. The R figures are geometry; the fill is not.

What are the most common price-action mistakes?

What else do people ask about price action?

Educational content only — not financial advice, and not a trade recommendation. The five-component list and the two readings — the trend reversal and the area of struggle — are from Part 9 of our own slide course, quoted where quoted and otherwise described in our own words. The overlap figures are ours, from two runs of the same five definitions: a Gaussian random walk of 400,000 bars at σ = 0.4% per bar, seed 20260905, 399,900 bars usable, with a volume series drawn independently by construction as a control; and Binance spot closed candles for BTC, ETH and SOL, four-hour and daily, January 2020 (SOL from its August 2020 listing) to 26 September 2026, with real volume shuffled 300 times as the control. The overlap multiplier is P(A ∧ B) ÷ (P(A) × P(B)), the identity used in Lesson 23. The two chart examples are real Binance BTC/USDT candles shown on our live chart, times UTC: daily, 22 June to 25 August 2025 (low 98,200; high 123,218; higher low 111,920; high 124,474; first close below 111,920 at 110,111.98; average daily range 3,130 over the prior 20 days), and four-hour, 19 to 22 September 2025 (zone 116,121.81 to 115,100; entry at the 114,649.89 close; average range 534 over the prior 20 candles, breaking candle 1,229.25). R:R is reward ÷ risk and break-even win rate is 1 ÷ (R + 1) throughout. Every percentage on this page is a measured frequency, a measured distance or a break-even threshold. No historical win rate is quoted anywhere, because none was measured. Published 5 September 2026; updated 5 October 2026, when the illustrative charts were replaced with real ones and the overlap test was repeated on real candles.