Mean Reversion Trading: What It Is, Connors' RSI(2) Rules, and Every Trade on the S&P 500 ETF Since 2021

Mean reversion is the idea that a price that has moved unusually far from its average will tend to move back toward it. A mean reversion trader buys after a sharp drop and sells after a sharp rise, betting on the return to normal rather than on the continuation of the move. It is a hypothesis about a market, not a law: some markets and periods revert, others trend. This page runs one published mean reversion strategy, Larry Connors' RSI(2), over five years of the S&P 500 ETF and lists what each trade did.

How to measure "far from the mean"

Every mean reversion rule needs two things: a mean, and a way to say how far price is from it.

• Distance in standard deviations. Bollinger Bands sit two standard deviations from a 20-day average; a close outside them is a large move for that market at that time.
• A short oscillator. The Relative Strength Index on two days instead of fourteen reacts to a couple of closes and can reach readings below 10 or above 90 within days.
• Distance from a moving average. The gap between price and its 5-day or 20-day average, in percent.

The mean itself can be short (5 days) or long (200 days). A drop toward a long average inside an uptrend is what trend traders call a pullback: the same move read by two styles.

A mean reversion formula

There is no single mean reversion formula, but most rules reduce to a z-score: how many standard deviations the close sits from its average over the same window. Bollinger Bands draw the same idea as lines, the 20-day average plus and minus two standard deviations, so a close on the upper band has a z-score of +2 and a close on the lower band −2. The RSI(2) below measures the stretch differently, from the last two days of gains and losses, but asks the same question.

The RSI(2) strategy

StockCharts' ChartSchool describes Connors' 2-period RSI strategy in four steps:

1. Trend. Look for longs only when the market is above its 200-day simple moving average, and for shorts only when it is below.
2. Signal. Buy when RSI(2) moves below 10, sell short when it moves above 90. Connors tested buying levels from 0 to 10.
3. Entry. ChartSchool notes that a trader can enter at the close or wait for the next open. This page enters at the close of the signal day.
4. Exit. In his S&P 500 example, Connors exits longs on a move above the 5-day SMA and shorts on a move below it.

The test below applies those rules to daily closes, one position at a time, using daily closes from January 4, 2021, so that the first 200-day average falls on October 18, 2021; it starts flat on that day and runs to October 28, 2026. It uses dividend-adjusted prices and no commissions, spreads or taxes. No position was open on October 28, 2026.

Every trade, October 2021 to October 2026

Long trades lasted 3.2 sessions on average and 10 at most; their median gain was 0.85%, and 46 of them compounded to 33.03%. The 12 shorts, ten of them in 2022, compounded to 0.11%: their gains were wiped out by two losses, in April 2025 and April 2026. Over the same five years, holding the ETF gained 84.65%, while the strategy was in a position on 200 of the 1,263 sessions. A short return is measured as (entry − exit) / entry.

The last 250 sessions, trade by trade

The chart below the introduction covers October 30, 2025 to October 28, 2026; the strategy was flat on the first day of the window. It took 14 trades, 13 long and one short.

The short of April 6, 2026 shows the risk of the method. The ETF was below its 200-day average after the March sell-off, RSI(2) rose to 91.09 and the rule sold at 655.61. The market did not revert: it kept rising, and the first close below the 5-day average came eleven sessions later, on April 21, at 700.53. That one trade lost nearly as much as the six winning longs before it had made together (7.32% added up).

Is it good to trade against the trend?

Mean reversion trades against the last few days, not necessarily against the trend. Connors' 200-day filter makes the long trades buy dips in an uptrend, a form of buying on weakness, and that side produced 35 winners out of 46 here. The short side, which sold rallies in downtrends, won 7 of its 10 trades in 2022, about 9.6% added up, then gave nearly all of it back on two shorts, in April 2025 (−2.22%) and April 2026 (−6.85%), when the market turned up from below its average. Fifty-eight trades on one market are not enough to say how the strategy behaves elsewhere: the ChartSchool article itself presents it as a basis for developing a strategy, not as one to use out of the box.