Home › Learn › Lesson 37

Five rules of price action — distance from the average, and what tight ranges mean

By Cuongboss · Updated 6 Oct 2026 · Published 5 Sep 2026

Rules are cheap to state and expensive to obey, and the reason people disobey them is usually that nobody told them the mechanism. “Don’t trade far from the average” sounds like caution; read as a short signal it becomes a way to lose money slowly. This lesson takes the five rules exactly as our course states them, puts a chart under each — four of the course’s own TradingView screenshots and three Bitcoin examples captured from our live chart — and attaches a number to each: how far is far, what a zone is worth in R, what a tight candle buys you, where the body of a long candle is drawn and why. One measurement in the middle reorganises the first rule, and it is the part most write-ups get backwards.

The five rules as five sketches. Rules 1 and 4 are the same average seen from two distances; rules 2 and 5 are both about where a stop can be placed; rule 3 is about time. Schematic only — the measured claims below sit on a simulation, on six years of real candles, or on a real chart with its own numbers — four from the course’s slides, three captured from the live chart.

What are the five rules, and what do they have in common?

They are the first five of eight strategy statements our course puts under the heading Price action trading strategies, and each is a single sentence with a chart. The remaining three — false breakouts, trading with the trend, and the broken trendline — are the next lesson. Here are the five, in the course’s words:

Read as a group they share one thing that the individual sentences hide: four of the five are about where a stop can go cheaply. Far from the average, it cannot; at a zone, it can; at a tight candle on the average, it can; under the body of a long candle, it can be placed precisely, though not always cheaply. Only rule 3 is about something else, which is time. That shared thread is why the rules belong together, and why the first one is the hinge — it is the one people read as a signal when it is a warning about geometry.

One thing the course does not say, and popular guides often do: that price action means trading without indicators. Rule 1 names four moving averages by length, and the course’s trend method is to find “which moving average price is respecting and hugging”, then buy the retest of it or sell the rejection at it. Price action here reads the price first and consults the average about it — a priority order for candles and levels, not a ban. The trend read itself follows a different order on this site: the RSI wave first (Lesson 19), the MA bundle second, price structure last — three separate reads that meet only in the final mix. Lesson 18 covers what the averages are; this lesson only uses them.

Rule 1 — how far is “far from the average”, and who closes the gap?

Far enough that a stop under the average is several candles wide, and the gap closes mostly because the average moves, not because price does. Both halves of that answer are measured below; the second half is the reason the rule says avoid and not fade.

The course’s chart for this rule first, because it shows what the rule looks like before any numbers are attached.

How far is far. We generated a Gaussian random walk of 400,000 bars at σ = 0.4% per bar, seed 20260905, and computed the gap between the close and a simple moving average of four lengths, as a percentage of the average. “Far” was defined as two standard deviations of that gap — the same threshold for every length, so the lengths can be compared. The average candle range on this series is 0.639% of price, which is what the candle column uses.

The last column is arithmetic rather than simulation: an exponential average with smoothing α = 2/(n+1) closes a fixed gap by the factor (1 − α) every bar, so with price frozen the gap halves in ln 2 ÷ −ln(1 − α) bars. For a 200-period average on a 4-hour chart that is 69 bars — eleven and a half days of price going nowhere before the gap is even half gone, without a single down bar.

Who closes the gap. The half-life assumes price stands still, which it does not. So the second measurement follows every episode in which the gap first exceeded +2σ until the first bar on which it was back inside ±0.5σ, and splits the closing between the two things that could have done it: the average rising, or price falling.

Four lengths, one answer: the average does roughly four-fifths of the work, and it takes about half its own length to do it. A gap to the 20-period average does not promise a fall; it promises eleven bars of the average catching up. That is the mechanism behind the course’s wording. The trader who chases the spike buys with a stop several candles below; the trader who fades it sells against a mechanism that removes the gap without price needing to drop at all. Both are trading the gap, and the gap is not a force. Avoiding it is the only reading of the rule the numbers support.

Does it hold on a real chart?

A random walk has no trends, and real markets do, so we ran the same definitions on real data: Binance spot BTC/USDT, ETH/USDT and SOL/USDT, four-hour candles from January 2020 (August 2020 for SOL) to 26 September 2026, closed candles only. One change: the gap is measured in average candles — the mean high-to-low range of the previous 100 bars — because a percentage threshold computed over six years would let 2021’s wild bars decide what “far” means in 2026’s quiet ones. As a control we shuffled the same candles into random order twenty times, which keeps every candle’s shape and destroys their order.

The shuffled column lands where the random walk did — 77–88% — so the simulation was a fair model of a market with no memory. The real column is higher at every length. With the candles in their true order the median share is about 100%, which means that by the time the gap was gone, price had typically not fallen at all: it closed higher than at the spike in 46–56% of episodes, against about 41% when the same candles were shuffled. That is trend persistence, showing up exactly where a random walk cannot put it. The spread from one episode to the next is wide — the middle half of the 20-period episodes runs from 57% to 168% — so this describes the typical gap, not the next one. But the typical gap is the one a rule has to be built for, and a short entered because price is far above the average is betting on a fall that, at the median, did not happen.

Here is what one of those 465 episodes looked like, on the chart this site runs.

Rule 4, below, picks up the same chart three days later.

Rule 2 — why does a horizontal level give the lowest-risk entry?

Because it tells you where the stop goes and puts it close, so the risk is the thickness of the zone rather than the size of the swing. That is a statement about geometry, and the course’s chart for this rule happens to have the geometry measured on it.

Take the two printed tags at face value. A zone 4.41 wide that is 3.39% of price puts the top of the buy zone at 4.41 ÷ 0.0339 = 130.09 and its bottom at 125.68. A zone 4.07 wide that is 2.85% puts the bottom of the sell zone at 142.81 and its top at 146.88. Now price the trade the labels describe — buy low, sell high:

The point is in the ratio of two widths. The swing between the zones is 9.8% of price; the buy zone is 3.39%. Because the stop is measured across the zone and the reward across the swing, the zone comes out at 4.41 ÷ 12.72 = 34.7% of the reward, and the risk with its half-dollar buffer at 38.6%. Enter halfway up the swing instead and the nearest defensible stop is still under the same zone, so the risk grows by the distance you climbed while the reward shrinks by the same amount — the ratio collapses from both ends at once. “Lowest risk” in the course’s sentence means exactly this: the zone is the only place where the stop’s natural home is also close.

Rule 3 — why does a longer range make the trend after it more durable?

The course states it as an observation with one example, and this lesson will not pretend to more than that. But the example is a good one, and there is a reason the observation is plausible that is worth spelling out.

Why would a longer range produce a more durable trend? The mechanical answer is that a range is where positions are built. Every bar spent between 3,200 and 4,200 is a bar in which someone bought and someone sold at those prices, and the longer it lasts, the more of the eventual holders bought inside it. When price finally leaves, those holders are the ones who are not selling on the first pullback — they are in profit and were patient enough to sit through four months of nothing. A range that lasted a week has built no such base.

One refinement from the trading notes this site draws on, restated in our own words: not every range is the same kind of range. Some are quiet because nothing is happening — neither side has force. Others are quiet because two timeframes are pulling against each other and cancelling. The second kind tends to resolve harder, because the losing side is already committed. Telling them apart is a matter of reading the larger frames, which is Lesson 21’s job; rule 3 only tells you that the length is worth noticing. And a figure worth holding loosely: one school of the same notes puts markets in ranges roughly 60% of the time and in trends 30–40%, which if even approximately right means rule 3 describes the majority of any chart, not an exception.

What this lesson will not do is attach a probability to the rule. A structureless series has no such effect to measure, and any number offered would be an invention. The course gives one example; the lesson gives the mechanism; the reader gets to look at the next range with both in mind.

Rule 4 — what do tight candles at the average actually buy you?

A cheap stop, which is the same thing rule 1 was refusing to give you. Rules 1 and 4 are one rule about one average seen from two distances: far away, the stop under the average is several candles and the rule says wait; close up, with the candles tight, the stop is about one candle and the rule says this is the place.

Back to the November 2024 chart, three days on. By 12:00 UTC on 9 November the 20-period average had climbed to 75,823, and the candle at that hour spanned 76,273 to 76,542 — $269, about a quarter of an average candle ($1,049 over the previous 20 bars, $960 over the previous 100), closing $497 above the average. That is a narrow-range candle on the average, which is where the course’s trend method says to buy the retest. The next candle broke its high.

Now put the trade you were tempted into at the spike next to the one the rules waited for. The stop goes half an average candle under the 20-period average in both, and the target is the same 80,000 — a round number above both entries, chosen only so the two can be compared; any common target tells the same story, because what changed is the stop.

Both trades won — neither stop was hit — and that is what makes the comparison fair: same direction, same target, same outcome. Waiting cost $2,193 of entry. It saved $4,619 of stop, because the stop’s natural home under the average climbed $6,812 in the 22 bars between the two entries. With risk fixed in dollars, the narrow-candle entry bought 4.9 times the coins, so the same correct call paid 2.88R instead of 0.97R. Lesson 42 has the arithmetic behind the last column. (The candle unit is the 100-bar average at each entry: $855 at the spike, $960 at the narrow candle.)

Do tight candles really come before bigger moves?

This is the one part of the rules a real chart can test directly, so we did. On the same BTC, ETH and SOL four-hour data, we took every candle whose high-to-low range was under 0.6 of the previous 20 candles’ average and whose close sat within half a standard deviation of the 20-period average — the setup above — and measured how far price travelled over the next six candles, from the highest high to the lowest low. Medians, in average candles of the previous 20 bars; the number of starting candles is in brackets.

In dollars the course’s sentence is backwards: after a narrow candle the next six candles moved 14–19% less than after an ordinary one, and after a wide candle more. Quiet follows quiet and loud follows loud — volatility clusters — and the shuffled column, where it cannot, shows no difference at all. Measured against the narrow candle itself, though, the next six candles covered 4.2–4.4 of its ranges against 2.6–2.7 for an ordinary candle: about 1.6 times as much. That is the version of rule 4 a trader can use. The move after a tight candle is not bigger; it is bigger relative to the stop a tight candle lets you place. The November trade is that sentence in one example: the same move, three times the pay per dollar risked.

Rule 5 — why does a long candle body become support?

The course’s reasoning: a long body on heavy volume is a record of positions opened at every price inside it, and those positions are what defend the area when price returns. The zone is drawn across the body — open to close — not across the wicks, because the wicks are prices that were visited and abandoned within the bar, and few positions live there.

Three details decide whether the rule is being applied or merely quoted. The first is where the zone ends: a close below 69,011 is the rule failing, and the stop belongs just under the body for exactly that reason. The second is what that does to the trade. A body 3.6 candles tall puts the stop up to 3.6 candles away; this entry came 29% of the way into the body, so the stop sat 2.8 average candles below it and the trade paid only 1.07R to the obvious target. The body defines the stop; it makes it cheap only when price comes back deep into the body, near its open. The third is volume. The course pairs the long body with volume, and on four-hour majors the pairing turns out to be automatic: of the 70 long bodies in the count below, the quietest still traded 1.6 times its 20-bar average volume and the median 4.1 times. A long body on ordinary volume is rare enough that the volume clause filters almost nothing.

How often does the body hold?

We counted, on the same BTC, ETH and SOL four-hour data. A long bullish body is one at least three times the average range of the previous 20 candles. We kept the cases where price first moved away — at least half a body above the close — and then came back into the body within 60 candles, as in the screenshot. “Held” means price went on to a new high above the pullback’s starting point before any candle closed below the body’s open. Of 70 such returns since 2020, 41 held — 58.6%. The same candles shuffled into random order twenty times produced 1,989 cases and a hold rate of 64.5%. Split by volume into thirds — 23, 24 and 23 cases from quietest to loudest — the 70 held 61%, 50% and 65% of the time: no pattern.

So the measurable part of rule 5 is the geometry, not the defence. A long body tells you where a stop can live — under its open — and how far away that is, before you enter. On this evidence it does not make that stop less likely to be hit than chance would. Seventy cases is a small sample and the gap to the shuffled figure is inside its noise; the honest summary is no detectable edge either way.

When are these rules wrong?

Rule 1’s real numbers come from six mostly trending years. On BTC, ETH and SOL since 2020 the average did about all of the closing. In a long sideways market price does more of it, and fading a spike can work — but that edge belongs to recognising the range, which Lesson 16 and Lesson 20 cover; it does not belong to the gap. The spread is wide, too: in a quarter of the 20-period episodes the average did less than 57% of the work. The gap alone, in either regime, is not a signal.

The thresholds depend on the bar size chosen. Two sigma at 0.4% per bar gave 1.98% for the 20-period average in the simulation; on BTC since 2020 the same percentage definition gives 6.2%, because BTC’s bars are bigger and its trends longer. In candles the two agree far better — 3.1 against 3.7 — which is why the candle column is the one to carry to a real chart.

Rule 2 assumes the zone holds. The course’s own chart ends with the lower zone breaking. The stop under the zone is not decoration; it is the part of the rule that makes the failure cost one zone’s thickness instead of a swing.

Rule 3 is a tendency with an example, not a law with a rate. Nothing on this page claims how often a long range is followed by the move the course describes, because nothing here measured it.

Rule 4 is wrong in dollars. Tight candles were followed by smaller moves, not bigger ones (see the rule 4 section). The rule survives only as a statement about the move relative to the stop; read it any other way and you will size up for a breakout that is, typically, quieter than average.

Rule 5 did not beat chance in our count. 41 of 70 returns into a long body held, against 64.5% for shuffled candles. Use the body to place the stop, not as a reason to expect the bounce.

What are the most common mistakes with these rules?

What else do people ask?

Educational content only — not financial advice, and not a trade recommendation. The five rules and the four TradingView screenshots are from Part 9 of our own slide course; the only figures taken from those screenshots are the ones printed on them (the two measurement tags on the LTCUSD chart, 4.07 / 2.85% and 4.41 / 3.39%, and the label range : 132 days), and levels described as “read from the scale” are approximate. The three BTC/USDT screenshots were captured from the TradingPrimer live chart on 6 October 2026 (Binance candles, times in UTC); prices described as an open, high, low or close are the candle’s own; stops and targets are derived from them as stated; the 20-period average is a simple moving average. Simulation: a Gaussian random walk of 400,000 bars at σ = 0.4% per bar, seed 20260905, simple moving averages, “far” = two standard deviations of (close − MA) ÷ MA, an episode running from the first bar beyond +2σ to the first bar back inside ±0.5σ, the split = the average’s rise divided by the total gap closed, reported as medians; the half-life column is ln 2 ÷ −ln(1 − 2/(n+1)). Real data: Binance spot BTC/USDT, ETH/USDT and SOL/USDT four-hour closed candles, 1 January 2020 (11 August 2020 for SOL) to 26 September 2026; the same episode definitions with the gap measured in average candles (mean high − low of the previous 100 bars); control = the same candles reshuffled 20 times (seeds 3700–3719), keeping each candle’s shape relative to the previous close. Rule 4 test: range under 0.6× the previous 20 candles’ average and |gap| under 0.5σ of the 20-period gap; outcome = highest high minus lowest low of the next six candles. Rule 5 test: bullish body ≥ 3× the previous 20 candles’ average range; price first reaches the close plus half a body, then returns to the close within 60 candles; “held” = a new high above the pre-return peak before any close below the open. R:R is reward ÷ risk and break-even win rate 1 ÷ (R + 1) throughout. Every historical frequency on this page — the gap shares, the six-candle ranges, the hold rates — is reported against a shuffled control, not as a strategy’s win rate. No win rate for any of the five rules is quoted, because none was measured. Published 5 September 2026; updated 6 October 2026 with the real-chart measurements and screenshots.