{fullwidth}
Technical Analysis of the Financial Markets — John J. Murphy
Macro Overview & Strategic Value
Chapter 12 (contributed content on Japanese candlestick charting) introduces a data-visualization and pattern-recognition system that uses identical OHLC inputs to bar charts but encodes the open-close relationship far more visually, and adds a library of short-term reversal and continuation patterns rooted in centuries-old Japanese trading practice. The core thesis is that candle patterns are direct visual snapshots of trader psychology unfolding in real time — a two- or three-day sequence literally shows the moment sentiment shifts, which is a more granular behavioral signal than most bar-chart patterns provide.
This matters to a practitioner because candlesticks are explicitly presented as compatible with, not a replacement for, every other tool in the book — they can be layered onto existing trend/oscillator analysis for earlier signal generation. The chapter’s most important operating constraint (a candle pattern is meaningless without first establishing the prevailing trend) reinforces a theme repeated since Chapter 5: no pattern has interpretive value in isolation from trend context.
Structurally, the chapter’s closing “filtered candle pattern” innovation (candle signals combined with oscillator presignal zones) is a direct, explicit multi-indicator confirmation technique — extending the confirmation/divergence principle from Chapter 6 and the oscillator mechanics from Chapter 10 into a concrete, combinable signal-filtering system.
Core Concepts & Mechanics
- Candlestick construction from standard OHLC data — The “body” spans open-to-close (white/hollow if close > open, black if close < open); “shadows/wicks” above and below show the day’s high-low extremes — identical inputs to a bar chart, but the open-close relationship is far more immediately visible.
- Long Day / Short Day body classification — Classifies days purely by the size of the open-close body (ignoring shadows entirely), giving a simple visual read on how decisively a session closed relative to its open.
- Spinning Tops — Small bodies with shadows longer than the body itself, regardless of color; interpreted as pure indecision days rather than directional signals.
- Doji lines and subtypes — Open equals (or nearly equals) close; the Long-legged Doji signals strong indecision, the Gravestone Doji (long upper shadow only) is bearish, and the Dragonfly Doji (long lower shadow only) is bullish — showing that shadow asymmetry, not just body size, carries directional meaning.
- Trend-prerequisite rule for pattern validity — A bullish reversal pattern is structurally impossible to validly identify inside an uptrend, and vice versa for bearish reversals in a downtrend; a simple moving average (e.g., ~10-period) is recommended as the practical trend filter needed before applying any candle pattern.
- Multi-day reversal patterns (Dark Cloud Cover / Piercing Line, Evening Star / Morning Star) — Two- and three-day sequences where an initial long trend-confirming candle is followed by a decisive counter-move closing beyond the prior body’s midpoint (or, for stars, a gapped small-body day followed by a strong reversal candle) — encoding a specific, sequential psychological turn rather than a single-candle read.
- Rising/Falling Three Methods (continuation pattern) — A long trend-day, followed by 2-3 small-bodied “rest” days contained within the first day’s range, followed by a new long day closing at a fresh trend extreme; functions as an explicit “stay in the trade” signal during a pause rather than a reversal warning.
- Pixel-resolution distortion in computerized charting — On compressed price scales, multiple price levels can collapse into a single screen pixel, creating the visual illusion of Doji patterns that don’t mathematically exist; software using exact price relationships rather than visual rendering avoids this false-signal risk.
- Filtered candle patterns (Greg Morris, 1991) — Candle reversal signals are only considered valid when a companion oscillator (e.g., Stochastics %D) sits in its “presignal area” (>80 or <20); this pairs a leading, discretionary-feeling pattern signal with an objective, rules-based oscillator filter to suppress premature/false candle signals.
Technical Terminology & Reference Table
| Term | Operational Definition |
|---|---|
| Body | Candlestick rectangle spanning open-to-close; white/hollow if close > open, black if close < open |
| Shadow (wick) | Thin line above/below the body showing the period’s high-low range |
| Long Day / Short Day | Classification based purely on body size (open-close distance) |
| Spinning Top | Small body with shadows longer than the body; signals indecision |
| Doji | Open equals (or nearly equals) close; no meaningful body |
| Gravestone Doji | Doji with long upper shadow, no lower shadow; bearish |
| Dragonfly Doji | Doji with long lower shadow, no upper shadow; bullish |
| Star | A small-body day gapping away from the prior long-body day |
| Presignal area | Oscillator zone (e.g., Stochastics %D >80 or <20) preceding an actual buy/sell signal; used to filter candle patterns |
| Reversal pattern | Candle sequence signaling a trend change; only valid against the opposing prevailing trend |
| Continuation pattern | Candle sequence signaling the current trend will resume after a pause |
The Author’s Market Philosophy
The chapter’s model treats candle patterns as unusually direct behavioral recordings — because the pattern is built entirely from the open-close relationship across consecutive sessions, it’s argued to expose the exact moment collective conviction breaks or resumes, more viscerally than a bar chart’s high-low emphasis. Edge generation is framed as combinable rather than standalone: candlesticks are explicitly positioned as compatible with virtually every other tool in the book (moving averages for trend context, oscillators for filtering), reflecting an assumption that no single indicator category is sufficient and that layering different data lenses on the same price action reduces false signals. The insistence that reversal patterns are invalid without a genuine prior trend reinforces the book’s recurring caution against pattern recognition detached from context, treating the trend-first, pattern-second sequence as non-negotiable.
Systemic & Portfolio Integration
The mandatory moving-average trend filter before applying any candle pattern directly reuses the trend-identification infrastructure from Chapter 9, while the filtered-candle-pattern technique is a concrete implementation of the multi-indicator confirmation principle first flagged in Chapter 6 and built out through Chapter 10’s oscillators. Because candle signals often trigger earlier than equivalent price-based indicator signals, they function well as an early-warning layer within a broader systematic framework that still relies on trend and oscillator tools for final confirmation and risk sizing.
Important Formulas, Data, or Initial Examples
- Dark Cloud Cover mechanics: Day 1 = long white body confirming an uptrend; Day 2 opens above Day 1’s high, then closes below the midpoint of Day 1’s body — a purely rule-based midpoint threshold, no numeric price example given.
- Evening Star mechanics: Day 1 = long white body; Day 2 gaps up, small body (a “Star”), staying above Day 1’s body; Day 3 gaps down below the Star and closes below the midpoint of Day 1’s body.
- Rising Three Methods mechanics: Day 1 = long white body; Days 2-4 = small bodies (at least two black) contained within Day 1’s range (“period of rest”); Day 5 = new long white day closing at a fresh high.
- Stochastics %D filter thresholds: presignal zone defined as %D > 80 (overbought) or < 20 (oversold); a candle pattern occurring outside this zone (e.g., %D at 65) is explicitly disregarded under the filtering method.
- Pattern library size: Japanese literature catalogs roughly 40 reversal candle patterns and 16 continuation candle patterns; the chapter’s closing reference table lists dozens of named single-, two-, three-, four-, and five-candle patterns split into bullish/bearish reversal and continuation categories.
Active Recall Evaluation
- Why is it structurally impossible, according to this chapter, for a valid bullish reversal candle pattern to occur within an established uptrend?
- Explain the mechanical difference between how a human visually identifies an Evening Star pattern versus how a computer program identifies the same pattern, and why that difference matters for consistency.
- How does the pixel-resolution distortion problem create false Doji signals, and what specifically resolves this issue?
- Describe how the filtered candle pattern technique combines a candle signal with an oscillator’s presignal area — what specific failure mode is this filter designed to prevent?
- Why does the Rising Three Methods pattern function as a “stay in the trade” signal rather than a new entry signal, based on its underlying five-day structure?
Answer Key (spoiler)
- A bullish reversal pattern is defined specifically as marking the turning point from a downtrend into a new uptrend; if the market is already in an uptrend, there is no downtrend left to reverse, so any candlestick sequence resembling the bullish shape in that context isn’t actually functioning as a reversal signal — the same visual shape only carries reversal meaning when it appears against the opposite prevailing trend it’s meant to reverse.
- A human interpreting a chart visually will often accept a “close enough” version of the pattern (e.g., accepting a close only near, not exactly below, the required midpoint, or allowing a missing gap), applying subjective judgment to imperfect real-world data; a computer program instead requires an explicit, exact mathematical rule (a precise price relationship) to flag the pattern, meaning it will reject cases a human might subjectively still call valid, trading flexibility for total consistency.
- When a large price range is compressed onto a limited number of screen pixels, several distinct price levels can end up mapped to the exact same pixel row, making the open and close appear visually identical (a false Doji) even though the underlying numeric prices actually differ; this is resolved by using software that identifies patterns from the exact numeric open/high/low/close values rather than from the rendered pixel image.
- The technique only treats a candle reversal pattern as valid if it occurs while a companion oscillator (e.g., Stochastics %D) is already sitting in its overbought (>80) or oversold (<20) presignal zone — meaning conditions are already ripe for an oscillator-based signal to eventually fire; this filters out candle patterns that appear when the oscillator is in a neutral, mid-range zone (e.g., 65), which are considered premature or unsupported by the broader momentum context and more likely to be false signals.
- The pattern’s five-day structure shows an initial strong trend day, followed by several small-bodied days that stay contained within that first day’s range (a pause, not a reversal), and then a resumption of new-high closes on the final day; because the “rest” days never actually break the established range or reverse the initial move, the pattern confirms the underlying trend was never actually threatened, giving a trader confidence to remain positioned through the pause rather than treating it as a new, separate entry opportunity.