The Relative Strength Index (RSI) is a momentum oscillator. It compares the average size of recent up-closes with recent down-closes and puts the result on a scale from 0 to 100. J. Welles Wilder introduced it in his 1978 book New Concepts in Technical Trading Systems. His default lookback was 14 periods, and he read values above 70 as overbought and below 30 as oversold. RSI shows how fast price has been moving. It does not tell you what an asset is worth, and in a strong trend it can stay above 70 or below 30 for weeks or months.
How RSI is calculated
RSI is built in three steps.
- Separate gains and losses. For each period, a close above the previous close counts as a gain. A close below it counts as a loss, recorded as a positive number.
- Average them. The first Average Gain and Average Loss are simple averages over 14 periods. After that, Wilder's smoothing applies:
Average Gain = [(previous Average Gain × 13) + current gain] / 14 Average Loss is updated the same way. This works like an exponential moving average.
- Convert to the 0–100 scale.
RS = Average Gain / Average Loss RSI = 100 − 100 / (1 + RS)
If Average Loss is zero, RSI is 100. If Average Gain is zero, RSI is 0.
Worked example. Average Gain = 1.2 and Average Loss = 0.8 give RS = 1.5. RSI = 100 − 100 / 2.5 = 60.
Why two charts can show different RSI values
Each value depends on the previous one, so the result depends on how much history went into the calculation. StockCharts says its charts use at least 250 prior data points. A spreadsheet that starts from fewer bars will give slightly different numbers.
How traders read RSI
Overbought and oversold: 70 and 30
Wilder treated readings above 70 as overbought and below 30 as oversold. Some traders use 80 and 20 instead, which produces fewer signals. Short-term traders sometimes pair a 2-period RSI with 80/20 levels. In a 2000 AAII Journal article, Wayne A. Thorp noted that the right extremes differ from one security to another. He also wrote that crossing a level is a reason to watch for a possible reversal, not a signal in itself.
Divergence
Wilder described two kinds of divergence:
- Bearish: price makes a higher high while RSI makes a lower high.
- Bullish: price makes a lower low while RSI makes a higher low.
Thorp calls divergences some of RSI's most powerful signals. StockCharts warns that they mislead in strong trends: an uptrend can produce many bearish divergences before it actually tops.
Failure swings
A failure swing is a pattern in RSI alone, without reference to price:
- Bearish: RSI rises above 70, pulls back, rallies again without reaching 70, then falls below its previous low.
- Bullish: RSI falls below 30, bounces above it, pulls back but stays above 30, then breaks above its previous high.
RSI ranges in trends
In Technical Analysis for the Trading Professional, Constance Brown says RSI tends to stay between 40 and 90 in bull markets and between 10 and 60 in bear markets. In practice, the 40–50 zone often acts as support in uptrends and the 50–60 zone as resistance in downtrends. These ranges change with the RSI setting, the strength of the trend and volatility.
Her mentor, Andrew Cardwell, described reversals, which mirror divergences:
- Positive reversal: RSI makes a lower low, usually between 30 and 50, while price makes a higher low.
- Negative reversal: RSI makes a higher high, usually between 50 and 70, while price makes a lower high.
Cardwell saw bearish divergences as something that happens in bull markets and bullish divergences as something that happens in bear markets.
Turn average gain and loss into RSI
Enter an Average Gain and an Average Loss to see the RS ratio converted to RSI on the 0–100 scale.
| Average Gain over the lookback (Wilder's default is 14 periods) | Average Loss over the same lookback, as a positive number | RSI |
|---|---|---|
| 1.2 | 0.8 | 60 |
| 1.2 | 0.8 | 60 |
| 0.8 | 1.2 | 40 |
- Average Gain and Average Loss are your own inputs; work them out with simple averages over 14 periods first, then Wilder's smoothing.
- If Average Loss is zero, RSI is 100; this calculator needs a loss above zero, and an Average Gain of zero gives RSI 0.
- Readings above 70 are traditionally called overbought and below 30 oversold. That label describes momentum, not value.
This only converts two averages into RSI: it does not compute the averages from price history, explain why charts differ, or say whether a reading will lead to a reversal.
Where RSI fails
- Sustained trends. Thorp writes that RSI's "usefulness breaks down when prices are in a sustained up- or downtrend." His example shows RSI staying near 70 for months while price kept rising. Overbought readings work best when price moves sideways in a range.
- It only uses past prices. RSI is calculated from past closes. Anything not yet reflected in price, such as news, is invisible to it.
- The settings change the signals. A 9-period RSI with 80/20 levels and a 14-period RSI with 70/30 levels will flag different moments on the same chart.
- One indicator is not enough. Both StockCharts and AAII recommend confirming RSI with other tools, such as moving averages or price structure.
Quick reference
| Item | Standard setting |
|---|---|
| Creator | J. Welles Wilder, 1978 |
| Default period | 14 |
| Scale | 0–100 |
| Overbought / oversold | 70 / 30 (some use 80 / 20) |
| Smoothing | Wilder's (previous × 13 + current) / 14 |
| Signal types | Levels, divergence, failure swings, trend ranges |
Limitations: this guide describes how RSI is defined and how it is usually interpreted. It does not test whether RSI signals are profitable, and it is not trading advice.
