Trading
RSI Explained: Momentum, Extremes, and False Signals
Learn what RSI measures, why overbought is not an automatic sell signal, and how to evaluate divergence without hindsight.
The Relative Strength Index, usually called RSI, describes the balance of recent upward and downward price changes on a scale from zero to 100. It is a momentum indicator calculated from price history. It does not measure how cheap a business is, how many buyers exist, or the probability that the next bar will rise.
That distinction changes how you read the familiar shaded areas above 70 and below 30. They identify relatively strong recent movement under a chosen calculation. They are starting points for observation, not automatic instructions to buy or sell.
What goes into the calculation
RSI compares smoothed gains with smoothed losses over a selected period. The standard expression is 100 minus 100 divided by one plus the ratio of average gain to average loss. The loss input uses the magnitude of downward changes. Fidelity's guide explains the formula and the conventional 70 and 30 reference levels.
In a hypothetical calculation, an average gain of 0.60 and an average loss of 0.30 produce a ratio of two and an RSI near 66.7. These are illustrative inputs, not a trading forecast. The calculation also has edge cases when gains or losses are zero; platform handling and initialization can affect early values.
A common setting is fourteen periods. On a daily chart, that refers to daily price changes; on an intraday chart, it refers to that chart's bars. Shortening the period typically makes the display more sensitive. Changing the session or price series changes the inputs, so record those settings when comparing results.
Why overbought can stay overbought
Fidelity notes that RSI can remain at extreme readings during strong trends. A high number describes the strength of the recent advance; it does not establish that the advance must end. Repeatedly shorting because RSI exceeds a threshold can mean repeatedly fighting a trend.
Imagine a hypothetical stock rising from 40 to 44, pausing, then climbing to 49. RSI could spend much of that sequence above 70. Someone who treats every high reading as a reversal prediction has turned a description of strong momentum into an unsupported timing claim.
Read the chart first. Is price making directional progress or oscillating within a range? Is an extreme reading appearing after a sudden gap? How far away is the level that would invalidate a proposed trade? These questions supply context that the oscillator alone cannot provide.
Divergence is a comparison, not a countdown
A bearish divergence means price makes a higher high while RSI makes a lower high. A bullish divergence reverses that relationship at lows. This disagreement can prompt closer review, but it cannot tell you when a reversal will occur or whether a tradable reversal will occur at all.
Define which swings count before testing divergence. For example, a swing high identified only after two subsequent lower highs cannot be used as if it were confirmed at the original peak. Otherwise, a historical chart makes the setup appear earlier and easier than it was in real time.
A hypothetical pair of price highs at 80 and 82, with RSI readings of 75 and 68, illustrates the comparison. Price could still continue to 88. The divergence has described weaker momentum between selected observations; it has not placed a ceiling on price.
Build a small research exercise
Choose one market, one bar interval, and a fixed session definition. Write a rule for identifying an RSI event, then record every occurrence over the selected period. Avoid collecting only attractive reversals. Include signals that produce sideways movement, continued trends, or immediate adverse moves.
Keep the entry rule separate from the RSI observation. If the rule waits for a completed candle, record the next realistic opportunity to execute. Specify an exit and a holding horizon before measuring outcomes. A five-bar recovery and a three-month rally answer different research questions.
Compare the same rule with and without the RSI condition. Include spreads and fees, and preserve a later sample that was not used to choose settings. If trying many thresholds creates one impressive result, that result needs stronger follow-up evidence, not greater confidence simply because it looks precise.
Connect momentum to a complete plan
RSI cannot choose an appropriate position size or ensure a loss stays within a budget. Investor.gov explains that a stop order can execute at a different price from its trigger, especially in fast markets. The number on an oscillator does not change that execution constraint.
A useful RSI note is specific: the reading, settings, market context, event definition, and what happened next. Over time, those records can reveal whether RSI adds information to your process. Keep its role proportionate to the evidence, and resist treating a familiar threshold as a universal market rule.