RSI: Wilder smoothing, persistent extremes and divergences
The Relative Strength Index measures the balance of recent positive and negative close-to-close changes.
Before this lesson
These lessons provide background; you can still read this page now.
What the indicator measures
The Relative Strength Index measures the balance of recent positive and negative close-to-close changes. Its 0–100 scale is not a probability scale. RSI 70 does not mean a 70% chance of a decline, and a low reading does not establish that an asset is economically cheap.
Calculation and units
RS = average gain / average loss; RSI = 100 − 100/(1+RS). TradingHub initializes averages from N price changes, then uses Wilder smoothing: new average = (previous average × (N−1) + current component)/N. This differs from simply recomputing a short rolling arithmetic ratio each bar.
Worked educational example
Given smoothed average gain 2 and average loss 1, RS is 2 and RSI is approximately 66.67. These are assumed averages for arithmetic, not a historical result. If the next bar is unfinished, both its change and the resulting RSI can move before close.
Reading it without overclaiming
Overbought and oversold describe the oscillator’s selected thresholds. Strong trends can maintain extreme readings, and a divergence can persist without a reversal. Define which price pivots and closed bars are compared; choosing pivots after a move makes a pattern look more decisive than the information available earlier.
How TradingHub implements it
The built-in default is period 14 with levels 70 and 30. Fourteen changes require fifteen close observations. In a completely flat series the native charts return 100 when average loss is zero; Studio’s RSI template returns 50 when both averages are zero. Treat this edge convention as an implementation difference, not strong momentum.
Comparisons that matter
Stochastic locates a close inside a recent high/low range, whereas RSI summarizes gains and losses. MFI also uses volume. None supplies an independent guarantee merely because it agrees with RSI. Match history and smoothing when comparing a built-in RSI with a Studio instance.
A reproducible chart review
Open the existing Trading chart and choose the same market, interval and closed observation time as your comparison. Record every calculation parameter before changing one at a time. Match source fields, history length and the bar window. Native indicators calculate from the supplied visible slice; Indicator Studio calculates from its available chart snapshot. A difference after panning may therefore be a window or warmup effect rather than a new market event.
Limits and next lesson
The illustration uses constructed candles, not a dated market observation or a backtest. A live candle can change until it closes. Missing history, volume units and implementation conventions limit comparisons. No indicator guarantees a direction, an execution price or protection from loss. Use the related lessons to separate definitions, observed values and possible explanations; a correct quiz records learning progress, not trading competence.
Knowledge check
What does RSI above its upper threshold establish?
- A guaranteed next-bar decline
- A numerical probability of profit
- The oscillator is above the chosen threshold, not a guaranteed reversal
Show explanation
Thresholds describe the calculation. Persistence, trend conditions and pivot selection limit directional interpretations.
Interactive checking needs JavaScript; the lesson and explanation remain readable.
Primary sources and scope
Educational scenarios are not execution guarantees, forecasts or personal recommendations. Check current product rules at their source.
Content revised: . Editorial responsibility: TradingHub. No independent professional certification is claimed.