Technical Analysis

What does technical analysis actually claim?

Technical analysis organizes market data into a trading plan. Price indicators transform bar history into trend, momentum, and volatility readings. Volume profiles use trade-level or lower-timeframe detail to locate activity across prices. Breadth indicators add participation across a group of securities. Each layer answers a different question about the setup.

Start with four decisions: identify the trend, map the price levels, measure the size of recent moves, and assess participation. Choose tools that answer those questions, then define the entry, target, and exit before opening the position. Connect the chart with options flow, dealer levels, and payoff analysis to build the complete trade.

What is the raw material, and what does a bar hide?

A candlestick chart compresses every trade inside an interval into four numbers. The real body is the absolute distance between open and close, the upper shadow runs from the top of the body to the high, and the lower shadow from the bottom of the body to the low. A bar with an open of $100, a high of $105, a low of $95, and a close of $102 does not reveal whether the high came before the low, how long price stayed at each level, or where volume traded.

Candlestick patterns state how those four measurements compare across one to three consecutive bars. An engulfing pattern requires the second body to cover the first and close in the opposite direction; a doji requires open and close at or near the same price. The name tells you the shape. Before you trade one, define its confirmation rule and test it on the market and timeframe you trade. Check two chart settings before measuring off a bar: the session decides which prints build it, and split adjustment restates older prices, so an adjusted chart’s historical levels differ from what printed that day.

How does a chart get turned into structure?

Market structure is the first layer and the one that makes the rest consistent. It reads the chart as an ordered sequence of swing highs and swing lows: ascending peaks and troughs define an uptrend, descending peaks and troughs define a downtrend, and a sequence that alternates without direction defines a range. A swing point is a definition, commonly a bar whose high exceeds the highs of a set number of bars on both sides. A two-bar rule marks many small pivots and a ten-bar rule few large ones, so the rule has to be stated first.

Support and resistance attach price areas to those pivots. Support is an area below the market where declines have repeatedly lost force, resistance an area above where advances have. Both are better mapped as zones than exact prices: a stock that reversed at $50.04, $49.91, and $50.12 has shown interest around $50, not three separate levels. When a level breaks, its role commonly reverses, which is the mechanic behind the retest. A different framework appears in support and resistance versus order blocks.

A trendline does that job diagonally, connecting rising swing lows or falling swing highs to express direction and pace. Its slope is the price change between two pivots divided by the time between them. Fix its anchors in advance, because redrawing them whenever price crosses the line makes almost any past move look orderly and the tool impossible to test.

A breakout is price leaving a boundary defined before the move. Confirmation costs something either way: a closing rule, a distance rule in percent or in ATR, a volume rule, or a retest rule each trade earlier entry for more evidence, so set the rule and the stop for a failed break before the move. Fibonacci retracement generates candidate levels differently, plotting lines at 23.6%, 38.2%, 50%, and 61.8% of one completed swing. In an uptrend the level is the swing high minus the ratio times the range. The 61.8% and 38.2% ratios come from the Fibonacci sequence; the 50% line is a trend-analysis convention, and the projection version is covered in retracement versus extension.

How does the moving-average family work, and where does it fail?

A simple moving average is the arithmetic mean of the last n selected prices. Every observation carries equal weight, and the oldest drops out when a new one enters, so the average moves both when a new price arrives and when an old one leaves. A large drop in the SMA can be caused entirely by a high close rolling out of the window while nothing unusual happens today.

An exponential moving average updates recursively: new EMA = prior EMA + [2 ÷ (n + 1)] × (new price - prior EMA). An SMA updates when the newest price enters and the oldest leaves. Their different weighting produces different turning points and crossings. Compare the selected periods against the actual price path, as shown in SMA versus EMA.

A moving average crossover fires when a fast average moves from one side of a slow average to the other, so every crossover confirms a move that already happened. Its failure mode is the range: in a sideways market the two averages sit near each other and cross repeatedly in both directions, and nothing in the construction distinguishes a range from a trend, so read ADX before acting on a cross. The golden cross and death cross are the same construction at conventional settings, the 50-day crossing the 200-day on daily closes, above for the golden cross and below for the death cross. In a representative death-cross case the signal prints only after a 16.8% decline is complete; in a golden-cross case price can already sit 27% above the faster average when the cross confirms. The pair is compared in golden cross versus death cross.

What do the oscillators actually measure?

The Relative Strength Index converts the balance of recent gains and losses into a bounded 0 to 100 reading. Relative strength is average gain divided by average loss, and RSI equals 100 minus 100 divided by (1 + RS). Equal average gain and loss produce 50, which gives the centerline a direct interpretation. RSI compares a security with its own recent changes, not a benchmark.

The stochastic oscillator answers a different question on the same scale: where the latest close sits inside the high-low range of the lookback window. %K is 100 times the close minus the lowest low, divided by the highest high minus the lowest low, and %D is a short simple moving average of %K. Because RSI measures change and the stochastic measures position in range, they disagree in informative ways, examined in RSI versus the stochastic oscillator.

MACD is the family outlier because it is unbounded. The MACD line is the 12-period EMA minus the 26-period EMA, the signal line a 9-period EMA of that line, and the histogram MACD minus signal. A value above zero means only that the fast EMA sits above the slow one, and no fixed level means the same thing across two instruments.

Overbought and oversold are labels applied to a reading, not measurements of value. RSI above 70 or below 30, and stochastic above 80 or below 20, are conventions chosen by the indicator authors. In a sustained trend nearly every lookback window contains mostly gains and closes near the top of its range, so the calculation correctly keeps returning a high number for as long as the trend persists. Divergence compares consecutive price swing points with the matching indicator swing points: regular bullish divergence pairs a lower price low with a higher indicator low, and regular bearish divergence pairs a higher price high with a lower indicator high. The value is read at the bar of the price extreme, not the nearest oscillator peak, and momentum can diverge for long stretches inside a trend that keeps running.

How do volatility tools set the scale of a move?

Average True Range is the scale everything else borrows. True range is the greatest of the current high minus low, the absolute distance from the high to the prior close, and the absolute distance from the low to the prior close. Including the prior close is what captures overnight gaps that a high-low range misses. ATR is direction blind by design: a sharp rally and a sharp decline raise it identically.

Bollinger Bands place an envelope a chosen number of standard deviations of closing price around a simple moving average, conventionally 20 periods and two deviations, so a few extreme closes expand them abruptly. Keltner Channels build the same envelope from different material: an exponential moving average offset by a multiple of ATR, commonly a 20-period EMA and a 2.0 multiplier. ATR-based width tracks the size of recent bars including gaps, so it changes more smoothly through a sustained trend. The pair is compared in Bollinger Bands versus Keltner Channels.

The Average Directional Index answers the question the others cannot: is there a trend at all. DX is 100 times the absolute gap between +DI and −DI divided by their sum, and ADX is a Wilder-smoothed average of DX, with above 25 conventionally a strong trend and below 20 trendless. ADX carries no direction; the +DI and −DI lines do. Its cost is the double smoothing, which confirms a trend well after it began.

Where does volume put the levels?

Volume profile bins executed volume by price rather than by time, a histogram of where business actually got done. The point of control is the price row holding the most volume in the profile. The value area is the band around it containing a configured share of that volume, commonly 70%, built by expanding outward from the point of control and adding the heavier neighboring row until the target is reached. Its edges are the value area high and low, and all three levels move when the session window, row height, percentage setting, or data feed changes.

VWAP collapses the same information into one line: cumulative price times volume divided by cumulative volume from the session start. A 600-share trade influences it six times as much as a 100-share trade at the same price. Anchored VWAP uses identical math from a bar the trader selects, typically an earnings gap or a high-volume reversal day, which turns the line into the average price paid since that event. Because the sums never reset, it grows less responsive as volume builds.

Relative volume qualifies participation rather than location: current volume divided by average comparable-period volume. Comparable does the work: equity volume is heavier near the open and close, so an intraday reading compares the same time of day across prior sessions. An RVOL of 2.40 means 2.4 times normal participation for that point in the session, with no directional content.

A worked example: one chart read end to end

XYZ has a 20-day simple moving average of $100.00 and a 20-day standard deviation of closes of $2.70. The prior 20-day EMA is $100.40 and the prior 14-period ATR is $2.20. Yesterday’s close was $101.20. Today the stock trades a high of $104.30 and a low of $101.00 and closes at $104.00, clearing a resistance shelf at $103.50 that had capped two prior rallies.

Volatility scale. True range is the greatest of $104.30 − $101.00 = $3.30, |$104.30 − $101.20| = $3.10, and |$101.00 − $101.20| = $0.20, so TR = $3.30. Updated ATR = [($2.20 × 13) + $3.30] / 14 = $31.90 / 14 = $2.28.

Trend baseline. The EMA smoothing factor is 2 / (20 + 1) = 0.0952. New EMA = $100.40 + 0.0952 × ($104.00 − $100.40) = $100.40 + $0.34 = $100.74.

Envelopes. Bollinger upper = $100.00 + (2 × $2.70) = $105.40 and lower $94.60, a width of $10.80. Keltner upper = $100.74 + (2 × $2.28) = $105.30 and lower $96.18, a width of $9.12. The Bollinger Bands now sit entirely outside the Keltner Channels, so no squeeze is present; a squeeze is the reverse, bands contracting inside the channels.

Momentum. With a 14-period average gain of $0.62 and average loss of $0.28, RS = 2.2143 and RSI = 100 - [100 / 3.2143] = 68.89, still below the conventional 70 line. The 14-period stochastic uses a high of $104.30 and a low of $95.60: %K = 100 × ($104.00 − $95.60) / ($104.30 − $95.60) = 96.55, well above 80. The two disagree on the same bar because they measure different things. MACD reads $102.80 − $100.30 = $2.50 against a signal of $1.90, a histogram of $0.60.

Trend strength. With +DI at 32 and −DI at 14, DX = 100 x |32 − 14| / (32 + 14) = 39.13, and the smoothed ADX sits near 27, above the 25 threshold with direction pointing up.

Break quality. Breakout distance = ($104.00 − $103.50) / $103.50 × 100 = 0.48%. In volatility units that is $0.50 / $2.28 = 0.22 ATR, so a 0.5 ATR filter would not yet call this confirmed.

Location. Yesterday’s profile held 2,000,000 shares. The $99.90 to $100.00 row held 260,000 shares, or 13% of the session, making it the point of control. A 70% value area requires 1,400,000 shares, and expanding outward reaches 1,430,000 by the time the band spans $99.60 to $100.20. Session VWAP is $103.10, so the close sits $0.90 above it. A Fibonacci retracement of the $95.60 to $104.30 swing puts the 38.2% level at $104.30 − (0.382 × $8.70) = $100.98 and the 50% level at $99.95.

What the arithmetic says together. The trend read is positive and confirmed by ADX, the break is real but shallow in ATR terms, and the stochastic is at an extreme while RSI is not. Three frameworks place a reference near $100: the 20-day SMA at $100.00, the point of control row at $99.90 to $100.00, and the 50% retracement at $99.95. A trader placing invalidation one ATR below the broken level uses $103.50 − $2.28 = $101.22, giving $2.78 of risk per share, so a $109.00 objective is $5.00 of reward, or 1.80 times the initial risk.

How does this read on a live tape?

None of the above arrives as a finished picture. The bar in progress has no fixed shape: its close, high, low, and pattern label can all change before the interval ends, so a rule that depends on the close is not knowable until the close prints, and an oscillator recomputed inside a bar can cross a threshold and cross back.

Read location first. Session VWAP updates with new trades, while a completed prior-session value area supplies fixed reference levels. Track where price sits relative to each and how it responds at the level. Then use relative volume to assess participation: a rejection at value-area high with 2.4 RVOL describes a different setup from the same candle at 0.6 RVOL.

Connect the chart with the options chain. At a resistance shelf, compare the price response with call activity, dealer concentrations, and the next target. Track the strength of each rejection and the time spent inside the zone as the level is tested. InsiderFinance technical analysis puts buy and sell signals, stops and three targets on your chart, and the options profit calculator shows each contract’s payoff at the target you choose. The chart defines the setup; the selected contract, expiration and premium define the payoff.

What are the standard mistakes, and why is each one wrong?

Treating an overbought reading as a sell signal. Read a persistent extreme as trend strength, and let price structure mark the turn. The calculation returns a high number precisely because the trend is persisting, so fading it fights the property the indicator was built to measure.

Drawing the level after the move. Mark the boundary before the move. A breakout has meaning only against a boundary defined in advance, and a trendline redrawn to keep a thesis intact will always fit the past and never be testable.

Ignoring the volatility scale. Measure a break in ATR units. A $0.50 break is large in a stock with a $0.30 ATR and negligible in one with a $2.28 ATR, so a fixed percentage filter applies one rule where the same number means different things.

Assuming a profile level is a fixed property of the market. Match the profile settings before comparing levels. The point of control moves with row height, session window, and data feed, so two traders can watch genuinely different levels while both are correct about their own configuration.

Mistaking one regime for the other. Identify the regime before choosing the bet. Mean reversion and momentum trading are opposite bets on the same reading: a z-score of +2.0 is the fade signal for one and the confirmation for the other. Evidence for reversion is strongest over multi-year horizons and in specific short-horizon regimes.

What to learn next

The efficient order is bottom up, because each layer supplies inputs for the next. Start with the bar in candlestick chart and the named shapes in candlestick patterns, then turn a sequence of bars into a stateable direction with market structure, support and resistance, and trendline.

The smoothing family is one concept with variations: simple moving average, exponential moving average, moving average crossover, and the long-horizon cases in golden cross and death cross. The oscillators follow the same shape: RSI, the stochastic oscillator, and MACD, read through overbought and oversold and divergence.

Volatility sets the scale: Average True Range, then Bollinger Bands and Keltner Channels, then ADX for the trend-or-range question. Finish with location and participation: volume profile, point of control, value area, VWAP, anchored VWAP, and relative volume, with Fibonacci retracement as another level generator and breakout as the event that tests them all.