The Test Bench › Mean reversion › RSI

RSI Explained: How RSI(14) and RSI(2) Work, and What Our Tests Found

How RSI is calculated, what 30 and 70 mean, and why a 2-day RSI uses 10 and 90, with what our tests found.

RSI (the Relative Strength Index) compares a stock’s recent up moves with its recent down moves and scales the result from 0 to 100. The number of days it looks back changes how often it reaches an extreme, and the shortest settings are the basis of many mean reversion strategies.

This page covers:

• How RSI is calculated, with a worked example
• What 30 and 70 mean, and why RSI(2) uses 10 and 90
• RSI(2) vs RSI(14) side by side
• The parts of an RSI rule, and what our tests found

What is RSI?

J. Welles Wilder introduced the Relative Strength Index in 1978. On each day it looks at the change in the closing price, keeps a running average of the gains and a running average of the losses, and compares the two. When recent gains have been larger than recent losses, RSI is above 50; when losses have been larger, it is below 50. A run of nothing but up closes pushes it towards 100, and a run of down closes towards 0.

RSI is computed on any timeframe. Wilder used daily bars and 14 periods, and that remains the default setting on most charting platforms, including GU Analyser.

How is RSI calculated?

1. Take each day’s change in the close. A rise is a gain and a fall is a loss, both recorded as positive numbers; the other is zero for that day.
2. Update the average gain and the average loss with Wilder’s smoothing: yesterday’s average × (n − 1) ÷ n, plus today’s value ÷ n, where n is the RSI period.
3. RS = average gain ÷ average loss.
4. RSI = 100 − 100 ÷ (1 + RS).

One invented day shows why the period matters. Yesterday the average gain was $0.80 and the average loss $0.40, which gives an RSI of 66.7 on any period. Today the stock falls $1.50.

The same fall moves RSI(14) from 66.7 to 60.8 and RSI(2) from 66.7 to 29.6. A 14-day RSI gives today one fourteenth of the weight; a 2-day RSI gives it half.

Why your RSI may differ from another platform

Charting platforms start the two averages in different ways: Wilder began with a simple average of the first 14 days, while others, GU Analyser among them, start from the first day’s change. Readings can therefore differ slightly between platforms, mainly at the start of the data; the difference shrinks as more days are added.

What do 30 and 70 mean on RSI?

Wilder marked 70 and 30 as the boundaries of the usual range for a 14-period RSI. A reading above 70 is called overbought: recent gains have far outweighed recent losses. A reading below 30 is called oversold. The labels describe what has already happened to the price. Whether a stock tends to rise after an oversold reading, or fall after an overbought one, is a separate question that a test answers.

What is RSI(2)?

RSI(2) is RSI with a two-day period. Larry Connors and Cesar Alvarez made it widely known in Short Term Trading Strategies That Work (2008). Because each day carries half the weight, RSI(2) swings across most of its range in a few days, so it is often read with 10 and 90; their tests also used 5 and 95 as stricter levels.

The long side of their rule, on daily bars:

• Trend filter. The close is above its 200-day simple moving average.
• Entry. RSI(2) closes below 10, and the published rule buys on that close. In practice that means an order placed just before the close, on a reading taken from the last price. Our backtest buys at the next day’s open instead, the price a trader can act on once the close is known.
• Exit. The close moves back above its 5-day simple moving average.

The short side mirrors it: below the 200-day average, RSI(2) above 90, covered when the close falls back below the 5-day average. Connors’ published rules use no stop-loss.

Two moments in an oversold spell can be tested: the day RSI enters the zone and the day it leaves it. Buying on entry catches the fall earlier; waiting for RSI to leave the zone waits for the first up close. Our studies measure both.

RSI(2) vs RSI(14)

The RSI formula is the same; the period changes how heavily the latest change affects it. RSI(14) describes roughly the last few weeks, RSI(2) the last couple of days.

Two other indicators build on a short RSI: ConnorsRSI adds the up or down streak and the size of today’s move, and Stochastic RSI places RSI(14) in its own recent range. Mean reversion trading explains the rules all four are used in.

How traders use RSI

• Oversold and overbought zones. Some traders treat a reading below 30 (or 10 on RSI(2)) as a possible buying point and above 70 (or 90) as a possible selling point, usually with other conditions.
• Leaving the zone. Waiting for RSI to cross back above 30, or back below 70, before acting.
• With a trend filter. Taking oversold readings only above a long moving average, and overbought readings only below it.
• The 50 line. Reading RSI above 50 as a sign that gains have outweighed losses recently.
• Divergence. A new low in the price with a higher low in RSI, read as a fall losing strength (and the reverse at a high). Our studies do not test divergence.

RSI strategy rules

An RSI rule states each of these parts:

• The period. 14 is the default; 2 to 5 for short-term rules.
• The zones. 30 and 70 for RSI(14); 10 and 90, or 5 and 95, for RSI(2).
• The signal. The day RSI enters the zone, or the day it leaves it.
• The trend filter. For example the 200-day moving average, or none.
• Execution. The signal day’s close or the next day’s open.
• The exit. A close above the 5-day average, a fixed number of days, an opposite RSI reading, or a stop.
• Trading costs. A short-term rule trades often, so each trade’s cost is paid many times.

What our RSI tests found

The event studies compared the days after a signal with ordinary days for the same S&P 500 stocks, 2006 to 2026; the backtest ran the signals as one trading rule.

• RSI(2) (event study, 10 and 90, next 5 days): in stocks above their 200-day average, the days after an oversold reading returned 0.19 percentage points more than ordinary days; in stocks below it, the days after an overbought reading returned 0.33 percentage points less than ordinary days. Four of its 8 signal tests showed a clear difference. Read the study
• RSI(14) (event study, 30 and 70, next 20 days): in stocks above their 200-day average, the days after an oversold reading returned 0.65 percentage points more than ordinary days, but the test did not show a clear difference; in stocks below it, the days after an overbought reading returned 1.60 percentage points less than ordinary days. Three of its 8 signal tests showed a clear difference. Read the study
• Backtest (one dip-buying rule: buy above the 200-day average, sell after the first close above the 5-day average): RSI(2) ranked first of the four entries by median total return per stock over 2006 to 2026 (+45%); RSI(14) ranked fourth (+5%). Read the study

Our RSI tests

Each study states the signal, the stocks, the period and what it compares the signal with. The results show what happened in the tested sample, and each study states what the test cannot establish.

Want to test an RSI rule?

1. Add RSI as a custom signal with the period you use: 14, 2 or any other.
2. Build the condition: RSI crossing below 10, crossing back above 30, or any other zone.
3. Run an event study to see what followed the signal, or a backtest with an entry, an exit and a cost on each trade.

Common questions

What is RSI?

RSI, the Relative Strength Index, compares the average size of a stock’s up moves with the average size of its down moves over a set number of periods and scales the result from 0 to 100. J. Welles Wilder introduced it in 1978 with 14 periods, which is still the default.

How is RSI calculated?

Each day’s change in the close is split into a gain or a loss. Wilder’s smoothing keeps a running average of each: yesterday’s average weighted by (n − 1) ÷ n plus today’s value weighted by 1 ÷ n. RS is the average gain divided by the average loss, and RSI = 100 − 100 ÷ (1 + RS).

What do 30 and 70 mean on RSI?

They are Wilder’s conventional zones for a 14-period RSI. A reading below 30 is called oversold and a reading above 70 overbought. Both describe the recent balance of up and down closes; neither predicts the next move by itself.

Is an RSI below 30 a buy signal?

It marks a stock whose recent losses have outweighed its gains; it does not by itself predict a rise. Many traders take it only in a long-term uptrend, or wait for RSI to cross back above 30. In our test on 483 S&P 500 stocks above their 200-day average, the 20 days after RSI(14) fell below 30 returned 0.65 percentage points more than ordinary days, but the test did not show a clear difference.

What does an RSI of 50 mean?

The average gain and the average loss over the period are equal. Above 50, recent gains have been larger; below 50, recent losses have. Some traders use 50 as a dividing line between up and down phases.

What is the best RSI setting?

There is no single best setting. 14 periods is the common default. Shorter settings react faster and reach the extremes more often: a 2-period RSI is often read with 10 and 90 in place of 30 and 70.

What is RSI(2)?

A 2-period RSI on daily closes, popularised by Larry Connors and Cesar Alvarez in Short Term Trading Strategies That Work (2008). Their rule buys a reading below 10 while the stock is above its 200-day moving average and sells when the close moves back above its 5-day moving average.

What is the difference between RSI(2) and RSI(14)?

The RSI formula is the same; the period changes how heavily the latest change affects it. RSI(14) gives each new day one fourteenth of the weight and changes slowly; RSI(2) gives it half and swings across most of its range in a few days. RSI(14) is usually read with 30 and 70, RSI(2) often with 10 and 90.

What is RSI divergence?

A price making a new low while RSI makes a higher low, or a new high with a lower RSI high. Traders read it as the move losing strength. It is a separate signal from the 30 and 70 zones, and our studies do not test it.

Which timeframe is best for RSI?

RSI works the same way on any bar length, from minutes to weeks; the period counts bars, not days. Wilder used daily bars, and our tests use daily bars on S&P 500 stocks.

Is RSI a mean reversion indicator?

RSI itself measures recent momentum; a strategy needs rules around it. Traders use extreme readings as part of a mean reversion rule, buying an oversold reading and selling on the rebound, and also use RSI as a momentum or trend filter, reading RSI above 50 as strength. Short settings such as RSI(2) are mostly used for mean reversion.

Does RSI work?

It depends on the rule. In our tests on S&P 500 stocks, RSI(2) below 10 in stocks above their 200-day average was followed by a median 5-day return 0.19 percentage points more than ordinary days; RSI(14) below 30 in stocks above their 200-day average was followed by a median 20-day return 0.65 percentage points more than ordinary days, but the test did not show a clear difference. In a separate backtest of one dip-buying rule, RSI(2) had a median total return per stock over 2006 to 2026 of +45%, against +5% for RSI(14); both trailed buying and holding on most stocks. These results apply to the tested stocks, period and rules; they do not show that RSI predicts future returns in general.

Can you backtest an RSI strategy?

Yes. In GU Analyser you add RSI with the period you want as a custom signal, build the entry and exit rules on a visual canvas, and run a backtest or an event study on daily bars, with no code.