Technical indicators are formulas applied to past prices and volume. Each one compresses a window of candles into a single number that answers one narrow question. The most common beginner mistake is to treat them as independent opinions, so that five indicators pointing the same way feel like five confirmations. Most of them are built from the same closing prices and often say the same thing in different units.

This guide sorts the common indicators by the job they do, works one example for each, and names the trap that comes with it. Each section links to a deeper guide for readers who want to turn an indicator into a testable rule.

Five jobs, not twenty indicators

Job Indicator Question it answers
Trend Moving averages, such as EMA 20 and EMA 50 Which way has price been drifting?
Momentum RSI(14), MACD(12, 26, 9) Is the move speeding up or fading?
Stretch RSI zones, Bollinger Bands (20, 2) Is price far from its recent norm?
Volatility ATR(14) How large is a normal candle right now?
Participation Volume, VWAP How much trading stands behind the move?

John Bollinger’s published rules for his bands put the principle plainly: “If more than one indicator is used the indicators should not be directly related to one another. For example, a momentum indicator might complement a volume indicator successfully, but two momentum indicators aren’t better than one.” Pick one indicator per job and keep the jobs separate.

Moving averages: the trend filter

A moving average smooths closing prices over a set number of candles. A simple moving average (SMA) gives each close in the window the same weight. An exponential moving average (EMA) is the other common type; MACD, below, is built from two of them.

Worked example on invented four-hour values: close 63,100, EMA20 62,400 and EMA50 61,300. Price is above both averages and the faster average is above the slower one, which describes an uptrend context. Distance adds information: with ATR at 1,200, price sits (63,100 − 61,300) ÷ 1,200 = 1.5 ATR above the EMA50, stretched but not extreme.

Trap: averages lag by construction. In a sideways market they cross back and forth with little follow-through. The moving-average crossover guide shows how to test settings without picking the best-looking pair after the fact.

moving-average crossover research workflowA research protocol fixes the moving-average type, source, periods, crossing event, direction and exits before testing. It then respects warm-up and completed-bar timing, models later entries and instrument costs, records every candidate, and reserves later data for one declared evaluation.MAKE TIMING AND TRIAL COUNT PART OF THE RULE1 · SpecifySMA or EMA · sourceperiods · direction2 · Confirmwarm-up · closed barsignal timestamp3 · Modellater entry · exitfees · spread · carry4 · Log allcandidate pairsincluding rejects5 · Evaluatereserved later dataonce · then retireIf the holdout changes a choicemark it used · begin a new evaluationA protocol can reduce avoidable bias; it cannot establish a profitable edge
A research protocol fixes the moving-average type, source, periods, crossing event, direction and exits before testing. It then respects warm-up and completed-bar timing, models later entries and instrument costs, records every candidate, and reserves later data for one declared evaluation. Illustrative teaching graphic · not live market data

RSI: momentum on a 0 to 100 scale

StockCharts’ ChartSchool explains that the Relative Strength Index (RSI) oscillates between zero and 100. It compares average gains with average losses over a window, commonly 14 candles:

  • RS = average gain ÷ average loss
  • RSI = 100 − 100 ÷ (1 + RS)

Worked example: an average gain of 1.2 and an average loss of 0.8 give RS = 1.2 ÷ 0.8 = 1.5, and RSI = 100 − 100 ÷ (1 + 1.5) = 60.

ChartSchool puts the usual zones this way: “When the RSI is above 70, it generally indicates overbought conditions; when the RSI is below 30, it indicates oversold conditions.” Those labels describe a stretch, not a direction. A divergence, where price makes a higher high while RSI makes a lower high, says momentum is fading; it does not say when, or whether, price turns.

RSI indicator value to research decisionAn RSI calculation needs enough prior observations under a declared smoothing convention. The illustration separates a computed value, a completed-bar threshold event and a later modeled action; it does not select the threshold or recommend a trade.A BOUNDED INDICATOR VALUE IS NOT YET A TRADING RULE1 · Specifysource · periodsmoothing · warm-up2 · Calculate0–100 valueor unavailable3 · Confirmclosed-bar crossingchosen threshold4 · Modellater entryexit + costsSame RSI line · reversal and momentum rules ask different questionsThresholds are research choices · compare failures and avoid searching only for a winner
An RSI calculation needs enough prior observations under a declared smoothing convention. The illustration separates a computed value, a completed-bar threshold event and a later modeled action; it does not select the threshold or recommend a trade. Illustrative teaching graphic · not live market data

Trap: treating 30 and 70 as instructions. The RSI strategy guide separates the reversal and momentum interpretations and shows how to test each.

MACD: changes in momentum

ChartSchool defines the main line directly: “The MACD line is the 12-day Exponential Moving Average (EMA) less the 26-day EMA.” It adds that “A nine-day EMA of the MACD line is plotted with the indicator”, which is the signal line. In formula form:

  • MACD line = EMA12 − EMA26
  • signal line = EMA9 of the MACD line
  • histogram = MACD line − signal line

Worked example: the MACD line is 61,500 − 61,100 = 400 when EMA12 is 61,500 and EMA26 is 61,100. With a signal line of 350, the histogram is 400 − 350 = 50. A positive histogram means the MACD line is above its own signal line.

Trap: MACD is made of moving averages, so it lags as well.

Bollinger Bands and %B: how stretched?

ChartSchool notes that “The outer bands are usually set 2 standard deviations above and below the middle band”, and the middle band is a 20-period simple moving average.:

  • middle band = SMA20
  • upper band = SMA20 + 2 × standard deviation
  • lower band = SMA20 − 2 × standard deviation
  • %B = (price − lower band) ÷ (upper band − lower band)

Worked example: an SMA20 of 60,000 and a standard deviation of 800 give an upper band of 60,000 + 2 × 800 = 61,600 and a lower band of 60,000 − 2 × 800 = 58,400. With price at 61,200, %B = (61,200 − 58,400) ÷ (61,600 − 58,400) = 0.875, near the upper band.

Bollinger’s rules state: “Tags of the bands are just that, tags not signals.” A close near or beyond a band describes where price is relative to its recent range, not where it goes next. The Bollinger Bands guide tests breakout and reversion as separate ideas.

ATR: the size of a normal candle

ChartSchool’s definition reads: “Developed by J. Welles Wilder, the Average True Range (ATR) is an indicator that measures volatility.” It measures the size of moves, not their direction. Expressed as a share of price it is easier to compare across assets: an ATR of 1,200 on a price of 60,000 is 1,200 ÷ 60,000 = 0.02, or 2% of price.

This guide uses ATR mainly for risk distance: a stop or zone sized in ATR scales with how large recent candles have been. The ATR stops guide turns ATR into a position size.

Volume and VWAP: participation

Volume counts how much traded in each candle. Check the volume unit your data source reports, and compare volume only against recent candles of the same market. VWAP, a volume-weighted average price, has its own VWAP guide, which explains why the reset point changes the line.

Combining indicators without double counting

One way to respect the “one indicator per job” principle is to weight families rather than individual readings. As a worked example, use weights of trend 0.45, momentum 0.30, location 0.15 and structure 0.10. With invented family readings of 0.6, 0.4, 0.5 and +1, the score is 0.6 × 0.45 + 0.4 × 0.30 + 0.5 × 0.15 + 1 × 0.10 = 0.565.

That number is a measure of agreement between families, not a probability that price rises. A score of 0.565 does not mean a 56.5% chance of anything. The TA signal guide explains how such a score becomes a bias label, and where it stops being useful.

What to remember

Indicators summarize the past. Use one per job, write down which question each one answers, and judge them on closed candles. Overbought and oversold zones are places traders look, not reasons to trade. Educational material only; all values above are invented for arithmetic.