What RSI actually calculates

RSI compares the average size of up closes to the average size of down closes over a period, usually 14, and expresses the result on a 0-100 scale. If the average gain equals the average loss, RSI sits at 50. If there have been only gains, it goes to 100. Only losses, to 0.

So RSI is a ratio of directional strength, smoothed and rescaled. It is not a measure of how expensive something is, and it has no relationship to fair value.

Why 'overbought' is the wrong word

The labels come from a range-trading world: in a sideways market, RSI near 70 does often coincide with the top of the range. But trends are not ranges. In a strong uptrend, up moves are persistently larger than down moves, so RSI sits high — often above 70 for weeks — while price keeps climbing.

This is why shorting every RSI-over-70 reading in a trend is one of the most reliable ways for a beginner to lose money. The indicator is not broken; the reading is being applied to a regime it does not describe. RSI extremes are far more actionable in ranges than in trends.

Better uses for RSI

As a regime filter. In an uptrend, RSI tends to hold above 40 and often bounces off 50. In a downtrend it tends to stay below 60. That asymmetry is more durable information than the 70/30 thresholds.
For momentum divergence. Price makes a new high, RSI makes a lower high — the thrust behind the move is weaker than before. Genuine, but as with MACD, it can persist for a long time.
For failure swings. RSI pushes above 70, pulls back, then fails to exceed its prior high on the next push while price does. A cleaner early warning than the raw threshold.

The 50 line is underrated

If you only take one thing from this lesson: watch whether RSI is spending its time above or below 50. That single habit captures most of the trend context that the 70/30 lines are usually misused for, and it does not tempt you into counter-trend trades.

Reference page with the formula: /indicators/rsi.

How to check this yourself

Put RSI(14) on a strong trending market and shade every stretch where it stayed above 70. Now look at what price did during those stretches. Then do the same on a ranging market. The two pictures will disagree completely, and that disagreement is the lesson: the same number means different things in different regimes.