02 · Technical Indicators in Depth

Technical indicators are second-order processing of "price + volume": moving averages are averages of price, MACD is the difference of moving averages, KDJ is the percentile position within a range... Understanding "what raw material each indicator uses and how many processing steps it takes" matters far more than memorizing formulas — it determines whether you can see through when an indicator will distort in a given market.

💡 One-Sentence Summary

One-sentence summary: all indicators lag. Indicators are 100% accurate only when "describing history"; they offer zero guarantees when "predicting the future".

---

1. Trend Indicators

1.1 MA (Moving Average)

Formula:

MA(n) = (C₁ + C₂ + ... + Cₙ) / n     (C = close)

Default parameters: 5 / 10 / 20 / 60 / 120 / 250-day averages (software defaults are often 5/10/20/60). The longer the period, the smoother and the more lagging.

Core usage:

Usage	Description
Golden/death cross	Short MA crossing above long MA = golden cross (bullish); crossing below = death cross (bearish)
Bullish/bearish alignment	Short > medium > long (e.g., 5>10>20>60) = bullish alignment, uptrend; the reverse = bearish alignment
Support/resistance	In an uptrend, a retest of the MA that holds = support; in a downtrend, a bounce off the MA = resistance
Price-MA deviation	Price stretched far from the MA = excessive bias; short-term mean reversion likely (read together with the outer Bollinger Bands)

Common mistakes:

• Moving-average golden and death crosses get repeatedly slapped in ranging markets (cross up and price falls, cross down and price rises), because a range has no trend to follow in the first place.
• Moving averages are a "follow-trend tool", not a "bottom-fishing tool": in a bearish alignment, every bounce to the MA is an exit point, not a buy point.
• Short-period MAs (e.g., 5/10) cross constantly on high-volatility instruments — huge signal noise; use longer periods.

1.2 EMA (Exponential Moving Average)

Formula:

EMA(n) = EMA(prev) + α × (today's close − EMA(prev))
α = 2 / (n + 1)

Default parameters: 12 / 26 / 50 (crypto markets often watch EMA 20/50/200).

MA vs. EMA:

Item	MA	EMA
Weighting	All prices equally weighted	Recent prices weighted more
Reaction speed	Slow	Fast
Lag	Stronger	Weaker
Noise	Somewhat less	More
Best for	Large-degree slow trends	Medium/short-term, turning-point-sensitive instruments (e.g., crypto)

Core usage:

• Golden/death cross usage is the same as MA, but signals come slightly earlier;
• EMA20 often serves as the short-term trend lifeline; EMA50/200 as the bull/bear divide (price above EMA200 = long-term bullish trend).

Common mistakes:

• Assuming "faster" means "more accurate": speed is bought with more false signals — fast and accurate cannot coexist.
• EMA 200 is watched by huge numbers of traders in A-shares/crypto, producing a "self-fulfillment" effect — but self-fulfillment also fails (at the start of a trend change, everyone cuts at once, creating an overshoot).

1.3 BOLL (Bollinger Bands)

Formula:

Middle band MB = MA(20)
Upper band UP = MB + 2 × SD    (SD = 20-day standard deviation of closes)
Lower band LO = MB − 2 × SD

Default parameters: 20 days, 2 standard deviations (20/2).

Core usage:

Usage	Description
Three bands	Middle band = 20-day MA (direction); upper/lower bands = the statistical band containing ~95% of price action
Expansion/squeeze	Expansion (bands widening) = volatility increasing, a move is starting; squeeze (bands extremely narrow) = volatility compressed, a regime change is near ("a Bollinger squeeze always precedes a big move")
Band-exit reversion	Price above the upper band → overbought, usually reverts to the middle band; below the lower band → oversold, usually reverts to the middle band
Middle-band support	In a trend, price rides the middle band; retests that hold it = healthy trend

Common mistakes:

• "Touch the upper/lower band = sell/buy" is the biggest mistake: in a strong trend, price rides along the upper/lower band for extended stretches (one-sided squeeze markets); fading it gets slapped repeatedly. The bands are "statistical boundaries", not "trading signals".
• After a squeeze, a big move is "guaranteed" but you don't know which way — other signals must set the direction.
• Bollinger Bands stay permanently squeezed on low-volatility instruments, reducing their value.

---

2. Momentum Indicators

The essence of momentum indicators: quantifying "the speed of the rise/fall". They do not answer "where price is", but "whether the rise/fall still has legs".

2.1 MACD (Moving Average Convergence Divergence)

Formula:

DIF = EMA(12) − EMA(26)
DEA = EMA(9, DIF)
MACD histogram = 2 × (DIF − DEA)

Default parameters: 12 / 26 / 9.

Core usage:

Usage	Description
Golden/death cross	DIF crossing above DEA = golden cross (bullish); below = death cross (bearish). Golden crosses above the 0 line are most reliable (bull market); death crosses below the 0 line are most dangerous (bear market)
0 line	DIF above 0 = medium-term bullish; below = medium-term bearish
Histogram	Shrinking bars = momentum fading (trend may be nearing its end); red bars turning green = bull/bear handover
Top divergence	Price makes a new high but DIF's peak is lower than the previous one = top divergence (rising momentum exhausted, bearish; see below)
Bottom divergence	Price makes a new low but DIF's trough is higher than the previous one = bottom divergence (falling momentum exhausted, bullish)

Top divergence sketch:
  price       ↗ ↗
        ╱╲ ╱╲          ← price makes a new high
DIF     ╱  ╲╱
      ╱        ← DIF peak lower (didn't follow) → top divergence

Common mistakes:

• Divergence can persist for a long time ("divergence can diverge again"): in strong trends, divergence signals are often simply steamrolled — never treat them as an absolute reversal basis.
• Golden/death crosses appear constantly in ranging markets — high signal noise. MACD suits trending markets, not sideways ones.
• Watching only the histogram and ignoring the DIF/DEA level misses the critical context of where 0 line sits.

2.2 KDJ (Stochastic Oscillator)

Formula:

RSV = (C − L₉) / (H₉ − L₉) × 100        (relative position within 9 periods)
K = K(prev) × 2/3 + RSV × 1/3
D = D(prev) × 2/3 + K × 1/3
J = 3K − 2D

Default parameters: 9 / 3 / 3 (RSV period 9; K, D smoothing 3).

Core usage:

Usage	Description
Overbought/oversold	K, D values > 80 = overbought; < 20 = oversold
Golden/death cross	K crossing above D = golden cross (bullish), more reliable at lows (<20 zone); K crossing below D = death cross (bearish), more reliable at highs (>80)
J value	J > 100 = extremely overbought; J < 0 = extremely oversold; when J is stuck ("frozen"), refer to K/D
Divergence	K/D top divergence bearish; bottom divergence bullish

Common mistakes:

• Freezing (sideways pegging): in strong trends, KDJ stays pegged in overbought/oversold zones for a long time (K/D stuck above 80 or below 20); trading "sell overbought, buy oversold" here gets harvested by the trend repeatedly.
• KDJ is extremely sensitive to short-term price — the noisiest intraday signals. It suits short-term trading, not medium/long-term direction calls.
• A low golden cross is not a sufficient condition for a bottom — in a slow bleed, KDJ can produce several low golden crosses followed by death crosses in a row.

2.3 RSI (Relative Strength Index)

Formula:

RS = average gain over n periods / average loss over n periods
RSI = 100 − 100 / (1 + RS)

Default parameters: 14 (6 / 14 / 24 as a common trio).

Core usage:

Usage	Description
Overbought/oversold	RSI > 70 = overbought; < 30 = oversold
50 divide	RSI > 50 = bulls in command (strong); < 50 = bears in command
Divergence	Top divergence (price new high, RSI not) bearish; bottom divergence bullish
Range reference	RSI hovering between 40–60 long term = ranging market; signals stop working

Common mistakes:

• Freezing, same disease as KDJ: in one-sided moves, RSI stays above 70 or below 30 for long stretches and "overbought/oversold" signals fail completely (a strong stock's RSI can sit above 90 for a month).
• RSI divergence in trends often "diverges again and again" — same as MACD divergence; use it with price structure.
• Treating 30/70 as absolute thresholds is dogma: high-volatility instruments (e.g., crypto) routinely range far wider than 30–70.

2.4 CCI (Commodity Channel Index)

Formula:

TP = (H + L + C) / 3
CCI = (TP − SMA(TP)) / (0.015 × MD)     (MD = mean absolute deviation of TP)

Default parameters: 14.

Core usage:

• CCI > +100: entered the strong zone (overbought, but strength can persist); CCI < −100: weak zone.
• Falling back below +100 from above = long exit signal; rising back above −100 from below = short exit signal.
• CCI is dimensionless and oscillates around 0 with no fixed ceiling/floor — it expresses "extremeness of deviation from the mean" better than RSI.

Common mistakes:

• CCI > 100 does not necessarily mean a top: in trending markets CCI can stay above 200 for a long time — fading it is fatal.
• CCI (14) is sensitive to price noise; too short a period produces messy signals.

2.5 WR (Williams %R)

Formula:

WR = (Hₙ − C) / (Hₙ − Lₙ) × 100        (position of the close within the n-period range)

(Mirror image of KDJ's RSV: high WR = close near the bottom of the range = oversold; the same logic applies to terminals that plot it as 0 ~ −100.)

Default parameters: 14 (10 / 6 as a common two-line setup).

Core usage:

• WR > −20 (or > 80, depending on the terminal's sign convention) = overbought; WR < −80 (or < 20) = oversold.
• Two-line golden/death crosses follow KDJ logic, but WR is more sensitive and more volatile.

Common mistakes:

• High sensitivity brings high noise: WR overbought/oversold signals can fire several times a day — trading them directly = working for the commission.
• Same freezing problem as KDJ/RSI; fails in one-sided markets.

---

3. Volume Indicators

3.1 OBV (On Balance Volume)

Formula:

Bullish close today: OBV = OBV(prev) + today's volume
Bearish close today: OBV = OBV(prev) − today's volume
Unchanged close:     OBV unchanged

Default parameters: None (a cumulative value, summed from the start date).

Core usage:

• OBV rising in step with price = volume-price cooperation, healthy trend.
• OBV top divergence: price makes a new high but OBV doesn't = the rise lacks volume support; be alert for a top.
• OBV flattening/rising first at the bottom while price still falls = accumulation signs ("volume leads price").

Common mistakes:

• OBV is cumulative; its magnitude depends on history — read only "shape and slope", never absolute values.
• OBV divergence has the same "diverge then diverge again" problem; wait for price-structure confirmation.

3.2 VWAP (Volume Weighted Average Price)

Formula:

VWAP = Σ(price × volume) / Σ(volume)     (cumulative for the day/period)

Default parameters: Intraday cumulative (resets each trading day/settlement cycle); terminals usually show today's VWAP by default.

Core usage:

• The "today's cost basis" favored by institutional flows: price above VWAP = today's buyers collectively in profit (bullish lean); below = in loss (bearish lean).
• Intraday traders use VWAP as the bull/bear line: retest that holds above VWAP = go long; break below VWAP = flip short.
• The "settlement price" of crypto perpetuals is usually computed as a weighted average over some window — same principle as VWAP.

Common mistakes:

• VWAP resets daily; a single day's VWAP alone has little reference value. Multi-period VWAP (weekly/monthly) needs separate setup — do not mix them.
• VWAP is a "cost reference", not "support/resistance"; it generates no buy/sell points by itself — it is descriptive statistics only.

---

4. Volatility Indicators

4.1 ATR (Average True Range)

Formula:

TR = max(today's high − today's low, |today's high − prev close|, |today's low − prev close|)
ATR(n) = n-period average of TR

Default parameters: 14.

Core usage: ATR measures "how much this instrument moves per day on average recently" — the cornerstone of stop-loss and position sizing:

Usage	Formula	Description
Volatility stop	Stop price = entry ∓ k × ATR (k usually 2–3)	Stop distance adapts to volatility: not swept by noise, still protective
Position sizing	Position size = per-trade risk amount / (k × ATR)	Converts "how much to lose" into "how much to buy"
Breakout entry	Enter when close breaks entry + k × ATR	Volatility breakout method (e.g., the Turtle system)
Regime gauge	Rising ATR = expanding volatility; falling ATR = compressing volatility	Expanding marks a trend launching / turning intense; compressing marks consolidation / a possible breakout ahead

Common mistakes:

• ATR is only a "volatility ruler" with no direction — it never tells you up or down, only "how big each step is".
• An ATR stop is "volatility-adaptive", not "loss-proof": in extreme conditions a single candle can far exceed 3×ATR (e.g., wick-hunt moves); position management remains the last line of defense.

---

5. Other Indicators

5.1 SAR (Stop And Reverse)

Formula:

SAR(today) = SAR(prev) + AF × (EP − SAR(prev))
AF: acceleration factor, starts at 0.02, +0.02 on each new high/low, capped at 0.2
EP: the extreme of the current trend (highest/lowest point)

Default parameters: 0.02 / 0.2.

Core usage:

• SAR dots below price = bullish trend with support points below; SAR above price = bearish trend.
• A trend-following tool: SAR flipping from below to above (or vice versa) = reversal signal, often used as a trailing take-profit/trailing stop — as price keeps rising, the stop keeps ratcheting up.
• Compared with moving averages, SAR is better at "locking in profit" than at "catching turns".

Common mistakes:

• In sideways chop, SAR dots flip back and forth and get slapped repeatedly — disabled in ranging markets.
• SAR only suits instruments and timeframes with clear trends; the 0.02 parameter stops too tight on low-volatility instruments.

5.2 Ichimoku (One-Glance Equilibrium Cloud)

Formula:

Conversion line   = (9-period high + 9-period low) / 2
Base line         = (26-period high + 26-period low) / 2
Leading span A    = (conversion line + base line) / 2      (plotted 26 periods ahead)
Leading span B    = (52-period high + 52-period low) / 2   (plotted 26 periods ahead)
Lagging line      = today's close plotted 26 periods back

Default parameters: 9 / 26 / 52.

Core usage:

Component	Meaning
Cloud (between A/B)	Support/resistance zone: price above the cloud = bull market; below = bear market; inside = range
Cloud thickness	Thick cloud = strong support/resistance (hard to cross); thin cloud = easy to cross
Conversion/base lines	Like a 9-period and a 26-period MA: short crossing above long = bullish; entangled = range
Lagging line	Compared with price 26 periods ago: above = bullish, below = bearish

Common mistakes:

• The cloud has 5 components; beginners take "price above the cloud" as the only signal, ignoring cloud thickness and lagging-line verification — single-condition usage = random signals.
• "Price crossing the cloud" lags badly: in fast markets the cloud often can't keep up with price, and by the time the signal fires, most of the move is over.

---

6. Indicator Combination Advice

6.1 How to Combine

💡 Basic Principle

Basic principle: keep only one indicator per dimension, and use different dimensions to cross-verify.

Dimension	Recommended	Example
Main chart (price panel)	One MA/EMA set (e.g., EMA 20/50/200) + BOLL middle band	Judge trend direction and trend health
Sub-panel 1 (momentum)	MACD or RSI, pick one	Judge momentum strength, find divergence
Sub-panel 2 (volume)	Volume (OBV)	Verify the authenticity of breakouts and divergences

Example recommended combo:

Main chart: MA20 + MA60 (trend direction) + BOLL (volatility range)
Sub-panel: MACD (12/26/9, momentum + divergence)
Sub-panel: volume / OBV (volume confirmation)

Workflow: main chart sets direction → MACD times the entry → volume judges signal authenticity → ATR sets the stop distance.

6.2 Why More Indicators Means More Losses

1. Signals contradict each other: when KDJ is oversold (buy) but MACD just death-crossed (sell), you are stuck dithering — and dithering is where losses begin.
2. Compounded lag: every indicator lags; stacking 10 indicators = 10 layers of lag — the "signal" you finally see is old news.
3. Multiplying noise: every indicator produces false signals; the more indicators, the smaller the intersection and the more the contradictions, until only "feel" decides — which puts you right back at gut-feel trading.
4. Curve-fitting trap: when you cherry-pick entries that "happen to satisfy 5 indicators at once", you are overfitting history — such conditions almost never recur in the future.
5. Decision paralysis: more indicators = slower decisions = bigger slippage = missing the stop-loss point while hesitating.

💡 Simple Beats Complex

A simple system you can execute long-term beats a theoretically perfect complex system. Most consistently profitable traders use just one main-chart indicator + one sub-panel indicator + volume.

---

Limitations and Pitfalls (Common Indicator Diseases)

1. Lag cannot be eliminated: all indicators are computed from historical prices; "today's" indicator signal reflects "yesterday's" momentum. By the time a golden cross prints, price has usually already risen well off the bottom.
2. Indicators fail wholesale in ranging markets: MA, MACD, KDJ, RSI produce dense false signals in sideways tape — that is their nature (they assume "a trend exists").
3. Parameters hold no magic: changing RSI to 7, 9, or 21 gives you no edge, just a different lag/sensitivity mix; what actually decides P&L are exit rules and position sizing.
4. The self-fulfillment illusion: once an indicator is widely used, its signal can "self-fulfill" short term (everyone buys the golden cross) — but when it fails (at trend changes), the damage is amplified just the same.
5. Indicators can never answer "why": a MACD golden cross means nothing on a stock with a blown earnings report. Landmines built from fundamentals and news are invisible to indicators — technical analysis handles price, not value.

🔄 Divergence Is Not a Reversal Button

Divergence only says "momentum is fading"; it does not guarantee "price will reverse" — after a divergence, price can keep making new highs/lows for a long time. Treating divergence as a reversal signal is the same as treating a golden cross as a must-rise signal: using a lagging indicator as a prediction tool.

⚠️ Risk Warning

All indicators are computed from historical prices and lag by nature; overbought can become more overbought, oversold more oversold. No single indicator constitutes a trading signal, and parameter optimization overfits easily.

Further Reading

• 03 · Volume-Price Analysis
• 04 · Advanced Candlesticks
• 01 · Candlestick Patterns