How RSI Is Computed and What It Actually Measures
RSI compares the average gain on up days to the average loss on down days over a specified lookback period (typically 14 days for daily charts). The result is expressed on a 0-100 scale. A reading of 70 means up-day average gains have been approximately 2.3 times larger than down-day average losses over the period -- the stock has been trending up strongly. A reading of 30 means the reverse -- down-days have dominated meaningfully. RSI does not measure price levels or valuations; it measures the internal momentum of recent price action, specifically how one-sided the gains or losses have been.
The 14-period default is not sacred. Shorter periods (7-9 days) produce a more volatile RSI that triggers overbought/oversold readings more frequently -- useful for shorter-term timing. Longer periods (21-25 days) produce a smoother RSI that triggers readings less frequently but with higher conviction when it does reach extremes. Different timeframes also matter: RSI on weekly charts is a very different tool from RSI on daily charts. A stock that is overbought (RSI above 70) on a daily chart but deeply oversold (RSI below 30) on a weekly chart may be experiencing a short-term countertrend bounce within a longer downtrend -- a context that daily-only analysis would miss.
RSI = 100 - (100 / (1 + RS))
RS = Average Gain over N periods / Average Loss over N periods
Standard: N = 14 periods