Technical Indicators Explained: What They Measure and What They Cannot

What an indicator actually is

A technical indicator is arithmetic applied to a price series. A 50-day moving average is the mean of the last 50 closes. RSI is a ratio of average gains to average losses. MACD is the difference between two exponential averages. None of these observes anything the price chart does not already contain; they reorganise it so that a particular property becomes easier to see or to code.

That framing resolves most confusion. Indicators do not predict. They summarise. Their value is that a summary can be tested mechanically, whereas "the trend looks strong" cannot.

The four categories

A well-constructed strategy typically uses at most one from each relevant category: for example, a moving average for trend context, ATR for stop distance and sizing, and possibly RSI for timing. Adding a second momentum oscillator alongside the first adds a correlated, delayed copy of the same information.

Lag, smoothing, and the trade-off

Every indicator computed from past data lags. Smoothing reduces noise and increases lag; shortening the lookback reduces lag and increases false signals. There is no setting that removes both, and every claim of a lag-free indicator is a claim about how the lag has been redistributed rather than eliminated.

Simple moving average of length N
   Average lag  = (N - 1) / 2 periods
   SMA(50)      = about 24.5 periods of lag

Exponential moving average with smoothing alpha
   Equivalent N = 2/alpha - 1
   EMA(50)      = similar lag, weighted toward recent data

Faster response always means more whipsaws.
The choice is not lag versus no lag.
It is lag versus false signals.

Turning an indicator into a rule

1. 1

   State what the indicator measures

   For example: "RSI(2) below 10 means the last two days were unusually weak relative to recent history."

2. 2

   State why that should matter

   Connect it to a market mechanism: short-term overreaction, liquidity demand, or trend persistence. Without a mechanism, you are testing a shape.

3. 3

   Define the action precisely

   Entry condition, on which bar, at what price. "Buy when RSI is low" is not a rule; "buy at the close when RSI(2) closes below 10" is.

4. 4

   Define the exit and the stop

   The indicator rarely provides both. Most strategies use the indicator for entry and something else, such as a time stop or ATR stop, for exit.

5. 5

   Test across a parameter range

   Vary the lookback and threshold. A genuine effect shows a plateau; a fitted one shows a spike at one setting.

6. 6

   Test across instruments and periods

   If the rule works only on one instrument in one decade, it is an artefact. Effects driven by participant behaviour should appear broadly.

The five common indicator mistakes

1. Stacking correlated indicators. RSI, Stochastic, and MACD frequently agree because they measure overlapping properties. Three confirmations from one category is one confirmation.
2. Optimising the settings. Searching for the parameter that performed best is overfitting. Prefer conventional values and require robustness across neighbours.
3. Using oscillators as standalone signals in trends. Overbought readings persist for months in strong trends. Oscillator extremes need a trend filter or a reversion context.
4. Ignoring what the indicator cannot see. No indicator knows about earnings tomorrow, a merger announcement, or a liquidity crisis.
5. Confusing an indicator with an edge. The indicator is a measurement device. The edge, if any, comes from the market behaviour the measurement captures.