Forex Trading13 min read

RSI Indicator: Settings, Strategies and What Our Data Shows

RSI indicator guide: Wilder's formula, the 14 setting, overbought and oversold levels and divergence, with tests on four markets of what RSI can do.

Last Updated: 2026-11-01
RSI Indicator: Settings, Strategies and What Our Data Shows

This site contains affiliate links. We may earn a commission if you sign up through our links. How we research

The RSI indicator (Relative Strength Index) is a momentum oscillator that compares the size of recent up moves with recent down moves and shows the result on a scale from 0 to 100. J. Welles Wilder created it in 1978, the default setting is 14 periods, and readings above 70 are called overbought and below 30 oversold. In our tests on four markets, the classic "buy when RSI comes back above 30" signal mostly landed within a few points of the share of all bars that went up anyway, so this guide treats RSI as a way to describe momentum, not as a reversal signal.

Disclosure: OnForex may earn a referral commission from Exness links on this page. This is general education, not investment advice.

What the RSI shows (and what it does not)

RSI answers one question: over the last N bars, were the price moves mostly up or mostly down, and by how much? A high reading means closes have been rising more than they have been falling. A low reading means the opposite. A reading near 50 means up and down moves have been about equal.

It does not tell you that price is "too high" or "too low". It has no idea about value, news or levels. It only measures how one-sided recent moves were, which is why a strong trend can keep RSI above 70 or below 30 for a long time.

RSI formula

Wilder's calculation has three steps. A gain is a bar's rise in closing price from the previous close (zero if the close fell) and a loss is the size of its fall (zero if the close rose).

  1. Average gain and average loss. For the first 14 bars take the simple average of gains and of losses. After that, smooth them: new average = (previous average x 13 + current value) / 14. This is Wilder's smoothing, a slower cousin of the exponential moving average.
  2. Relative strength (RS) = average gain / average loss.
  3. RSI = 100 - 100 / (1 + RS).

A calculation with numbers

Say the 14-bar average gain on EURUSD 4H is 6 pips and the average loss is 4 pips.

  • RS = 6 / 4 = 1.5
  • RSI = 100 - 100 / (1 + 1.5) = 100 - 40 = 60

Now the next bar closes 10 pips higher (a gain of 10, a loss of 0).

  • New average gain = (6 x 13 + 10) / 14 = 6.29 pips
  • New average loss = (4 x 13 + 0) / 14 = 3.71 pips
  • RS = 6.29 / 3.71 = 1.69, so RSI = 100 - 100 / 2.69 = 62.9

One strong bar moved RSI by less than 3 points, because 13/14 of the old average stays. That smoothing is why RSI is a gradual measure and why a single candle rarely pushes it from 50 to 80.

RSI indicator settings

The 14-period RSI with levels at 70 and 30 is the default on MT4, MT5 and TradingView. The inputs you can change are the period and the levels.

PeriodCharacterTypical use
7 to 9Fast, reaches 70 and 30 often, many false extremesShort-term timing; some traders use 80/20 levels to cut noise
14Wilder's default, balancedMost charts and most tutorials
21 to 25Smooth, rarely reaches the extremesSlower trends, higher timeframes

Changing the period changes how often RSI reaches 70 and 30, not whether the readings predict anything. Because there is no "correct" value, the rule is the same as for any indicator: if you change a setting, test it on past charts and watch out for choosing the one that merely fitted the past best (see our backtesting guide).

In our 1-hour data with the 14 setting, RSI was above 70 on 5.5% (EURUSD), 6.0% (GBPUSD), 7.3% (USDJPY) and 9.1% (XAUUSD) of bars, and below 30 on 5.0%, 5.2%, 5.3% and 4.4%. Gold and USDJPY rose during the period, so they spent more time overbought than oversold.

Reading the chart

EURUSD 4-hour chart with an RSI(14) panel showing RSI above 70 at the start, then dipping below 30 several times while price fell steadily
EURUSD 4H, 21 Aug - 24 Sep 2026, RSI(14). RSI starts near 77, then falls below 30 around 28 Aug and again from 14 Sep. Price kept falling from about 1.1700 to about 1.1385 while RSI stayed low.

The chart shows the main behaviour. At the left edge RSI is near 77, an overbought reading, and price then drifts lower. Around 28 August RSI falls below 30 (about 24); price bounces to roughly 1.1640 by 9 September, so a buyer at that first oversold reading would have seen a gain. On 14 to 17 September RSI is below 30 again, and this time price kept going from about 1.1550 to about 1.1380. By the right edge RSI is back near 25 and still falling.

Two points to notice:

  • Oversold stayed oversold. During a decline, RSI spends long stretches near or below 30 without a lasting turn.
  • The bounces were weak. In the downtrend, RSI bounces reached only about 63 to 67 and then rolled over. Traders often read that as a sign the trend is intact: in a downtrend RSI tends to hold below the 60s, and in an uptrend it tends to hold above the 40s. This is a widely used rule of thumb that we did not test.

Our data: do overbought and oversold readings work?

We tested the textbook signals with RSI(14), looking at whether price was higher or lower a fixed number of bars later, and compared that with a baseline: the share of all bars that were higher (or lower) the same number of bars later, with no signal at all. A signal only adds something if it beats its baseline clearly.

Source: OnForex analysis of hourly prices from Yahoo Finance, January 2024 - October 2026 (gold: June 2024 - October 2026, COMEX gold futures). 4H bars are resampled from hourly data. No spreads or slippage are included.

1-hour charts, 24 bars (1 day) later

SignalEURUSDGBPUSDUSDJPYXAUUSD
Baseline: price higher47.7%51.5%56.5%53.8%
RSI crosses back above 30: higher56.0% (n=248)51.9% (n=266)57.6% (n=256)56.8% (n=177)
Baseline: price lower52.3%48.5%43.5%46.2%
RSI crosses back below 70: lower53.9% (n=243)44.9% (n=268)49.1% (n=273)47.0% (n=300)
First close above 70 (momentum): higher49.8%54.7%53.1%56.9%

4-hour charts, 24 bars (4 days) later

SignalEURUSDGBPUSDUSDJPYXAUUSD
Baseline: price higher48.7%49.8%58.0%57.1%
RSI crosses back above 30: higher47.5% (n=81)50.7% (n=74)54.2% (n=59)57.5% (n=40)
Baseline: price lower51.3%50.2%42.0%42.9%
RSI crosses back below 70: lower46.9% (n=64)48.5% (n=67)50.0% (n=88)50.0% (n=100)
First close above 70: higher48.4%62.1%49.4%51.0%
First close below 30: lower57.0%46.6%44.1%32.5%
Bar chart for EURUSD 4-hour data comparing the share of price rises after RSI crossing back above 30, a lower Bollinger Band close, a MACD cross and two candlestick patterns with the baseline share
EURUSD 4H, January 2024 - October 2026: after RSI crossed back above 30, price was higher 24 bars later 48% of the time, against 49% for any bar. The other classic buy signals sit close to their baselines too.

How to read these numbers

  • The classic "buy back above 30" beat its baseline clearly in one cell. On EURUSD 1H it was 8.3 points above the baseline (56.0% vs 47.7%). On GBPUSD, USDJPY and gold 1H it was 0.4 to 3.0 points above, and on the 4H table it was within about 4 points either way. With 40 to 270 cases per cell, a gap of a few points is noise.
  • "Sell back below 70" did not show an edge. GBPUSD 1H was 3.6 points below its baseline, and the 4H cells on EURUSD and GBPUSD were also below.
  • Gold punished fading strength. Gold had RSI above 70 on about 9% of 1-hour bars because it trended up. The first 4H close below 30 on gold was followed by a lower price four days later only 32.5% of the time, against a baseline of 42.9%: selling "oversold" there went against a rising market.
  • A strong 4H cell can be luck. GBPUSD "first close above 70" at 62.1% stands out, but neighbouring cells sit near 50% and a table of this size will always produce a few cells far from 50%.
  • Baselines are not 50%. USDJPY and gold rose over the period, so "price higher" was already above 50% for them. Compare a signal with its own baseline, never with 50%.

The honest conclusion: RSI levels describe momentum. They were not a reliable reversal signal in our data, and any edge would have to come from something else you add, such as trend, location and risk control. We did not test discretionary context.

Four ways traders use RSI

1. Overbought and oversold in a range

When price is moving sideways between a clear support and resistance, RSI near 70 at resistance or near 30 at support can help time a fade. The reason to trade is the level; RSI is a minor confirmation. The danger is that the range breaks, and RSI then stays extreme. Use closes and a stop beyond the range. See support and resistance.

2. RSI inside a trend

In a trend, RSI is used to time entries, not to call tops and bottoms. In an uptrend a dip of RSI toward the 40s that turns up again can mark a pullback; in a downtrend a bounce toward the 50s or 60s that rolls over can do the same. Define the trend first, with market structure or a moving average, then use RSI for timing. Our pullback trading guide shows how.

3. Divergence

A bearish divergence is a higher high in price with a lower high in RSI; a bullish divergence is a lower low in price with a higher low in RSI. The idea is that momentum is fading even though price made a new extreme.

XAUUSD 4-hour chart with price making a higher high while the RSI(14) panel makes a lower high, a bearish divergence
XAUUSD 4H, early to mid Aug 2026: price makes a higher high while RSI(14) makes a lower high. Divergence describes weakening momentum, not a guaranteed reversal.

We tested RSI(14) divergence using two swings 5 to 40 bars apart (a bearish setup needed the first RSI peak above 60, a bullish one below 40). Price moved in the divergence direction 24 bars later:

EURUSDGBPUSDUSDJPYXAUUSD
1H45.4% (n=174)49.3% (n=209)56.6% (n=212)54.3% (n=175)
4H52.9% (n=51)49.0% (n=51)36.5% (n=52)58.5% (n=41)

That is roughly a coin flip, so divergence is a reason to pay attention, not a trigger. Our divergence guide has more detail.

4. The 50 line and failure swings

Some traders treat 50 as a bias line: RSI above 50 means up moves dominate, below 50 means down moves dominate. Wilder also described a failure swing: RSI rises above 70, falls, bounces without reaching 70 again and then breaks below its previous low. It is a pattern in the indicator, not a promise about price. We did not test the 50 line or failure swings separately.

Worked example: turning an RSI idea into a trade plan

Suppose EURUSD is in an uptrend on 4H, price is above a rising 50-period average, and it pulls back to a prior swing low. RSI dips to about 42 and turns up. The idea is a pullback entry with the trend.

  1. The reason to trade is the trend and the swing low. RSI is the timing cue.
  2. Entry on the close of the bar where RSI turns up. Place the stop below the swing low with a buffer, say 20 pips away.
  3. Risk 1% of a USD 1,000 account = USD 10. Pip value is about USD 10 per standard lot, so size = 10 / (20 x 10) = 0.05 lots.
  4. Target the previous swing high and check the reward is at least 1.5 to 2 times the risk (risk-reward ratio).
  5. The stop is where the idea is wrong. If price closes below the swing low, you are out, whatever RSI says.

Numbers are illustrative. Spreads widen around news and rollover, and stops can fill with slippage. See position sizing and stop loss.

RSI vs stochastic vs MACD

RSIStochasticMACD
MeasuresAverage up moves vs average down movesClose vs recent high-low rangeGap between two moving averages
Scale0 to 1000 to 100Unbounded, around a zero line
SpeedMediumFastSlow
Usual signal70 / 30 levels, divergence80 / 20 crossesSignal-line and zero-line crosses

They are all built from the same price series, so they tend to agree. Stacking three of them gives one piece of evidence, not three. See our stochastic guide and MACD guide.

Common mistakes

  • Selling every reading above 70 or buying every reading below 30. In a trend, RSI can stay extreme for days, as the EURUSD chart shows.
  • Using RSI without a level or a trend. A reading in the middle of nowhere gives no reason to trade.
  • Comparing results with 50% instead of the baseline. Trending markets make "price higher" more or less than 50% on their own.
  • Hunting for the perfect period. Tuning 14 to 11 or 17 on past data usually fits noise.
  • Ignoring the stop. Every RSI idea needs a price where it is wrong, and the size comes from that distance.
  • Trading right before news. Spreads widen and RSI jumps with the first fast candle.

For the wider picture, read our technical analysis guide, then Bollinger Bands, the ADX indicator, candlestick patterns, multi-timeframe analysis and our list of forex strategies.

Practise on a demo account first

Mark structure, FVGs and stops on a free Exness demo account before you risk money. Forex and CFD trading carries a high risk of loss.

Continue to Exness

Forex and CFD trading involves significant risk. You may lose your invested capital.

Risk warning: Forex and CFD trading carries a high risk of loss and is not suitable for everyone. Indicators describe past prices and do not predict future moves. Past test results do not guarantee future results. Never risk money you cannot afford to lose.

Related Articles

Forex Trading2026-11-01

Stochastic Oscillator: Settings, Formula and What Our Data Shows

How the stochastic oscillator works: the %K and %D formula, 14,3,3 settings, fast vs slow vs full, overbought and oversold crosses tested on four markets.

Read More3 min
Forex Trading2026-11-16

MACD Forex Guide: MACD Indicator Settings (12, 26, 9) and Use

MACD forex guide: what the default 12, 26, 9 settings mean, the MACD indicator formula with a worked example, signal line, histogram, divergence and limits.

Read More3 min
Forex Trading2026-11-01

Parabolic SAR: Settings, Formula and Backtest Results

Parabolic SAR explained: the formula, 0.02 / 0.2 settings, how to read the dots, why it whipsaws, and what a flip-every-signal test showed on four markets.

Read More2 min
Forex Trading2026-11-01

ADX Indicator: Settings, +DI/-DI and What Our Data Shows

ADX indicator guide: how ADX and +DI/-DI are calculated, what 20, 25 and 40 mean, how to use ADX to pick a method, and what our data shows on four markets.

Read More3 min
Forex Trading2026-10-31

Ichimoku Cloud Explained: Five Lines, Signals and Settings

Ichimoku cloud explained: the five lines, how to read the cloud, sanyaku kōten and gyakuten, the 9-26-52 settings and what our daily data shows.

Read More2 min
Forex Trading2026-11-20

Heikin Ashi Candles: Formula, Example and How to Read Them

Heikin Ashi candles explained: the formula, a worked four-candle example you can check by hand, how to read the chart, how it differs from normal candles, and its limits.

Read More3 min