What Is the Relative Strength Index (RSI)? Formula and How to Use It

Search Fintrev

Search our guides, reviews and glossary.

The relative strength index (RSI) is a momentum oscillator that measures the speed and size of recent price changes and expresses the result as a number between 0 and 100. It compares the average gain on up periods with the average loss on down periods over a look-back window, which is 14 periods by default. Traditionally, a reading of 70 or above is treated as overbought and a reading of 30 or below as oversold. The indicator works the same way on a share price, a currency pair or a futures contract.

J. Welles Wilder Jr. introduced the RSI in his 1978 book New Concepts in Technical Trading Systems. It was first calculated by hand for commodities and futures. Once charting software became widely available, it was applied to stocks, currencies and almost every other traded market, and it is now a standard indicator on retail and professional trading platforms.

What the RSI measures

The RSI answers a narrow question: over the last n bars, how large were the up moves compared with the down moves? It does not measure price level, value or volume. A stock trading at a record high and a currency pair trading at a ten-year low can both show an RSI of 60 if their recent gains have outweighed their recent losses by the same proportion.

Because it uses only closing-price changes, the RSI rises when up closes become more frequent or larger, and falls when down closes become more frequent or larger. A run of small gains can lift it gradually. A single large drop can pull it down sharply, because one big loss raises the average loss.

The word “overbought” causes the most confusion. On the RSI it means only that buying pressure has dominated recent price changes to an unusual degree. It does not mean the asset is overvalued or that a decline is due. The same applies to “oversold.” Both labels describe momentum, and momentum can stay extreme for as long as a trend persists.

How the RSI is calculated

The RSI formula has two layers. The visible part is short:

RSI = 100 − [100 ÷ (1 + RS)]

where RS = Average gain ÷ Average loss over the chosen number of periods.

The work lies in how the two averages are built.

Step 1: Measure each period’s change

For every bar, subtract the previous close from the current close.

• If the change is positive, it counts as a gain, and the loss for that bar is zero.
• If the change is negative, its absolute value counts as a loss, and the gain for that bar is zero.
• An unchanged close counts as zero for both.

Losses are entered as positive numbers. This keeps RS positive and the RSI inside 0 to 100.

Step 2: Calculate the first averages

For the first RSI value you need 14 price changes, which means 15 closes. Add up all gains over those 14 periods and divide by 14. Then add up all losses and divide by 14. Both sums are divided by the full period count, including the bars where that side was zero. A market with ten up days and four down days is therefore not averaged over ten and four days respectively.

Step 3: Smooth the averages going forward

After the first value, Wilder’s method does not drop the oldest bar and re-average. It carries the previous average forward and blends in the newest bar:

• New average gain = [(Previous average gain × 13) + Current gain] ÷ 14
• New average loss = [(Previous average loss × 13) + Current loss] ÷ 14

For a period length other than 14, replace 13 with n − 1 and 14 with n.

This smoothing gives the newest bar a weight of 1/14 and lets older bars fade gradually instead of falling out all at once. Mathematically, it is an exponential moving average with a smoothing factor of 1/n. A 14-period Wilder average behaves like a 27-period standard EMA, because a standard EMA uses 2 ÷ (N + 1), and 2 ÷ 28 equals 1/14.

Worked example (hypothetical)

The figures below are invented to show the arithmetic. They are not drawn from any real market.

Initial value. Over 14 bars, a price closes higher on 8 bars with gains totaling 4.20, and lower on 6 bars with losses totaling 2.10.

• Average gain = 4.20 ÷ 14 = 0.30
• Average loss = 2.10 ÷ 14 = 0.15
• RS = 0.30 ÷ 0.15 = 2.0
• RSI = 100 − [100 ÷ (1 + 2.0)] = 100 − 33.33 = 66.67

Next bar: a gain of 0.50.

• Average gain = [(0.30 × 13) + 0.50] ÷ 14 = 4.40 ÷ 14 ≈ 0.3143
• Average loss = [(0.15 × 13) + 0] ÷ 14 = 1.95 ÷ 14 ≈ 0.1393
• RS ≈ 2.256
• RSI ≈ 100 − (100 ÷ 3.256) ≈ 69.29

Following bar: a loss of 0.60.

• Average gain ≈ (0.3143 × 13 + 0) ÷ 14 ≈ 0.2918
• Average loss ≈ (0.1393 × 13 + 0.60) ÷ 14 ≈ 0.1722
• RS ≈ 1.695
• RSI ≈ 100 − (100 ÷ 2.695) ≈ 62.89

A 0.50 gain lifted the reading by about 2.6 points. The slightly larger 0.60 loss pulled it down by about 6.4 points. The response is uneven because the RSI is driven by the ratio of the two averages. When the average loss is small, adding to it changes the ratio far more than adding the same amount to the larger average gain.

Why readings rarely reach 0 or 100

The RSI reaches 100 only if the average loss is exactly zero, meaning no down closes at all inside the smoothed window. It reaches 0 only if there are no up closes. The smoothing keeps traces of earlier bars in both averages, so extreme readings near the ends of the scale tend to appear only during persistent one-directional moves.

Why two charts can show different RSI values

Wilder’s smoothing never fully forgets the starting point. The value you see depends partly on how much price history the platform loaded before the first average was calculated. A platform that starts from a longer history may print a slightly different number than one that starts from a shorter history. The gap usually narrows as more bars accumulate.

Some platforms also offer RSI variants that use a simple or standard exponential average instead of Wilder’s smoothing. These variants produce different readings from the same prices. Differences in session times and closing prices can also change the result. Forex in particular trades around the clock, so the daily close a platform uses affects every daily bar.

How to read the RSI

The indicator is usually plotted in a separate panel beneath the price chart, with horizontal reference lines at set levels.

The two bands: 70 and 30

The two horizontal lines most platforms draw by default are the 70 level (overbought) and the 30 level (oversold). Wilder proposed these thresholds. A reading above 70 says recent gains have heavily outweighed losses. At exactly 70, the average gain is about 2.33 times the average loss. At 30, the average loss is about 2.33 times the average gain.

Reaching a band is an observation, not an instruction. Some traders treat the line crossing back below 70 or back above 30 as the signal, rather than the first entry into the zone. The crossing back shows that the extreme momentum has started to ease.

The 50 midline

At 50, average gains and losses are equal. Readings above 50 indicate that gains have been larger on balance, and readings below 50 indicate the reverse. Some traders use crossings of 50 as a simple momentum filter. In that approach, they favor long setups while the RSI holds above 50 and short setups while it holds below 50.

Reading RSI ranges inside trends

The bands become less useful once a market starts trending. A strong uptrend seldom pulls the RSI down to 30, and a strong downtrend seldom lifts it to 70.

Market technician Constance Brown, CMT, argued that the levels themselves should shift with the trend. In an uptrend, the effective oversold zone sits much higher than 30, often around 40 to 50. In a downtrend, rallies tend to stall near 50 to 60 rather than 70. Commonly cited working ranges are:

• Uptrend: the RSI tends to oscillate between roughly 40 and 80. Pullbacks to about 40 to 50 act as momentum support.
• Downtrend: the RSI tends to oscillate between roughly 20 and 60. Rallies to about 50 to 60 act as momentum resistance.

These ranges also help with spotting trend changes. Suppose an uptrend’s RSI fails to reach 70 on several consecutive swings and then drops below 30. That combination suggests the uptrend is breaking down. In a downtrend, the mirror case is an RSI that stops reaching 30 and then rallies above 70, which suggests the decline has lost its grip. In a sideways range, adjusting the levels is usually unnecessary, and the standard 70 and 30 apply more directly.

RSI trading signals

Each of the techniques below reads the RSI line differently. None of them guarantees an outcome, and each can produce false signals.

Overbought and oversold crossings

The simplest approach is to buy when the RSI moves below 30 and back up, and to sell or short when it moves above 70 and back down. This works best in a sideways market where price keeps returning to the middle of a range.

In a trend, the same approach has a structural problem. Selling every move above 70 during a strong uptrend means repeatedly trading against the dominant direction, and price can keep climbing long after the reading first turned overbought. Even when a countertrend reversal does arrive, it may be short-lived, because a strong trend often returns to test its previous extreme.

For this reason, many traders take bullish signals mainly in uptrends and bearish signals mainly in downtrends. In an uptrend, the bullish signal is often an RSI pullback to the 40 to 50 area that turns higher. In a downtrend, it is a bounce to 50 to 60 that turns back down.

Using the RSI to exit

RSI crossings can manage an existing position as well as open new ones. A short-term trader holding a long position might exit when the RSI falls back below 70, accepting a smaller gain rather than sitting through a likely pullback. A short-term short seller might cover when the RSI climbs back above 30. Exits are a separate decision from entries and usually sit alongside a price-based stop.

Divergence

A divergence occurs when price and the RSI disagree about the direction of momentum.

• Bullish divergence: price makes a lower low, but the RSI makes a higher low. Price fell further, but the selling behind the second low was weaker than before.
• Bearish divergence: price makes a higher high, but the RSI makes a lower high. The new price peak was reached with less upward momentum.

Divergence is one of the most-watched RSI signals, but it can mislead in a durable trend. A long uptrend can print several bearish divergences in a row while price keeps rising. Traders who use divergence typically wait for price confirmation before entering a countertrend trade, such as a break of a recent swing level or a trendline. Divergence also shows up in chart patterns: a lower RSI peak on the right shoulder of a head and shoulders pattern is a common example.

Positive and negative reversals

Reversals are a less familiar price and RSI relationship that runs the opposite way to divergence. Divergences hint that a trend is weakening. Reversals are read as signs that the existing trend is likely to continue.

• Positive reversal (bullish): the RSI makes a lower low while price makes a higher low. Momentum dipped more than price did, yet price held above its prior low.
• Negative reversal (bearish): the RSI makes a higher high while price makes a lower high. Momentum pushed higher, yet price failed to exceed its previous high.

The two concepts are easy to mix up:

In a divergence, the indicator lags behind price. In a reversal, price lags behind the indicator.

Failure swings (swing rejections)

A failure swing looks only at the RSI line’s own highs and lows. It does not need any comparison with price. Wilder treated failure swings as strong standalone signals of a trend change. Because the indicator must break its own support or resistance, they add evidence beyond a simple crossing.

A bullish failure swing has four parts:

1. The RSI falls below 30.
2. The RSI rises back above 30 and forms a short-term peak.
3. The RSI pulls back but stays above 30.
4. The RSI breaks above the peak from step 2. This is the signal.

A bearish failure swing mirrors it:

1. The RSI rises above 70.
2. The RSI drops back below 70 and forms a short-term trough.
3. The RSI bounces but stays below 70.
4. The RSI breaks below the trough from step 2. This is the signal.

Failure swings often appear together with a divergence on the price chart. Like other RSI signals, they are more reliable when they agree with the longer-term trend. A bearish failure swing inside a broader downtrend, for example, is less likely to be a false alarm than one in a strong uptrend.

Range-bound trading

When price moves sideways between clear support and resistance, a trader might buy near support with the RSI below 30 and sell near resistance with it above 70, using the RSI to confirm what the price levels suggest. Ranges eventually break, so this approach is usually paired with tight stop-losses just beyond the range.

Choosing RSI settings

Period length

Fourteen periods is the default and gives a balanced response to price changes. Changing it changes the indicator’s character:

• Shorter periods (about 5 to 9): the line reacts faster and reaches the bands more often. Short-term and day traders use these settings to catch quick momentum shifts, at the cost of more noise and more false signals.
• Default (14): suited to swing and position trading on daily charts and widely used on intraday charts.
• Longer periods (about 21 to 30): the line moves slowly and rarely reaches the bands. These settings suit traders focused on major trends.

The period and the chart timeframe work together. RSI(14) on a 5-minute chart covers just over an hour of trading. On a weekly chart, it covers about three months. A scalper on a one-minute EUR/USD chart and a position trader on a weekly chart can both use “RSI 14” and be measuring completely different things.

70/30 or 80/20?

Some traders tighten the thresholds to 80 and 20 so that only more extreme readings count. This suits strongly trending instruments, or short period lengths that would otherwise touch 70 and 30 constantly. Others widen the bands for longer periods, which rarely reach 70 or 30 at all. A practical approach is to set the bands where readings are genuinely unusual for the specific instrument and timeframe being traded.

RSI compared with similar tools

RSI vs. relative strength

These two are often confused because of the shared name.

• The relative strength index measures one security’s own momentum: its recent gains against its own recent losses.
• Relative strength (also called comparative relative strength) compares the price performance of one security with another, such as a stock against its index, usually by dividing one price by the other.

A stock can have a high RSI while underperforming its sector, if the sector rose even faster. The “RS” ratio inside the RSI formula is the gain-to-loss ratio, not this comparison.

RSI vs. MACD

The moving average convergence divergence (MACD) is a trend-following momentum indicator built from moving averages. It is calculated by subtracting the 26-period EMA from the 12-period EMA to form the MACD line. A 9-period EMA of the MACD line, called the signal line, is then plotted on top of it. Crossings of the MACD line over or under the signal line serve as buy and sell triggers.

The two indicators differ in several practical ways:

• Scale: the RSI is bounded between 0 and 100. The MACD is unbounded and expressed in price units, so its values are not comparable across instruments.
• What it measures: the RSI measures recent gains against recent losses. The MACD measures the distance between a shorter and a longer moving average.
• Typical use: the RSI is mainly read for overbought and oversold extremes. The MACD is mainly read for trend direction and crossovers.

Because they measure different things, the two indicators can disagree. The RSI might sit above 70 for an extended stretch, suggesting the move is overextended, while the MACD still shows buying momentum increasing. Both can signal a possible trend change by diverging from price.

RSI vs. stochastic oscillator and Stochastic RSI

The stochastic oscillator also runs from 0 to 100 and is also read with overbought and oversold zones, usually 80 and 20. It works differently, though: it places the current close within the high-to-low range of recent bars, while the RSI compares the sizes of up and down closes.

Stochastic RSI applies the stochastic formula to RSI values instead of to price. The result shows where the current RSI sits within its own recent range. It moves much faster than the RSI and reaches its extremes far more often. It is a derivative of the RSI, not a replacement, and its readings should not be interpreted on the same scale.

Combining the RSI with other tools

The RSI describes momentum only. It says nothing about trend structure, volatility or where meaningful price levels sit. Combining it with tools that cover those gaps is a common way to filter its signals:

• Moving averages define the trend so that RSI signals can be taken in its direction. An example is accepting bullish RSI setups only while price holds above a chosen moving average.
• Support and resistance give oversold or overbought readings a location. An oversold RSI at a well-tested support level carries more weight than one in the middle of nowhere.
• Bollinger Bands add a volatility context. Some traders look for an oversold RSI while price touches the lower band.
• Volume shows participation. An RSI breakout accompanied by heavy volume suggests broader conviction behind the move.
• ADX measures trend strength, which helps decide whether to read the RSI as a range tool or a trend tool.

Limitations of the RSI

It can stay extreme for a long time. A strongly trending asset can hold above 70 or below 30 for many bars. Selling a market just because it is overbought means trading against momentum that is, by definition, still strong.

False signals are common, and true reversals are hard to isolate. A false positive might be a bullish crossover followed by a sudden decline. A false negative might be a bearish crossover followed by a sharp rally. Real turning points are rare compared with the number of times the line crosses its bands.

It lags. Every reading is built from past closes, and Wilder’s smoothing adds further delay. The RSI confirms that momentum has shifted after the fact, rather than predicting the shift.

Settings change the answer. The same chart can look overbought on RSI(7) and neutral on RSI(21). Levels that suit one instrument may not suit another, and a configuration fitted closely to past data may not carry forward.

Context is outside the formula. News events, gaps, thin liquidity and session changes all feed into closing prices. The RSI cannot tell a gain driven by an earnings surprise from one driven by a slow drift, because both look the same once converted into average gains.

The Bottom Line

The RSI suits traders who want a bounded gauge of recent momentum on any market or timeframe. Its readings describe the balance of recent gains and losses, so overbought and oversold zones work best as context, read in the direction of the trend and confirmed by price, support and resistance, or other tools. Period length and platform settings change the numbers, so traders should match them to the instrument they trade.