Mean Reversion Trading Strategy: How It Works and How to Trade It

Search Fintrev

Search our guides, reviews and glossary.

Mean reversion is the idea that a price, return or ratio that has moved unusually far from its average will tend to move back toward that average. A mean reversion trading strategy turns that idea into rules. You measure how far price has stretched from a chosen mean, enter against the stretch when it reaches an extreme, and exit as price returns toward the mean. The method depends on two things the trader controls: which average counts as “the mean,” and how far is “far enough.” It works best when a market is moving sideways. It loses money when a stretch is the start of a new trend rather than a temporary excess.

What reversion to the mean means

In markets, “the mean” can be several different things:

• Price, such as a 20-day or 200-day moving average of closing prices.
• Returns, where a period of unusually high or low returns is expected to be followed by more ordinary ones.
• Valuation ratios, such as a company’s price-to-earnings (P/E) ratio relative to its own history or to its sector.
• Interest rates and spreads, which are often modeled as pulling back toward a long-run level.
• Volatility, where unusually high or low implied volatility is expected to drift back toward normal.
• Relationships between assets, such as the price ratio of two related stocks or two correlated currency pairs.

Why prices revert, and when they don’t

Several explanations support mean reversion. Markets can overreact to news, so a price overshoots what the information justifies. Short-term order imbalances can push a price away from fair value until liquidity providers step in. In ranging markets, the same support and resistance levels keep attracting buyers and sellers, so price oscillates around a central value.

Academic research supports mean reversion over some horizons. In 1985, Werner De Bondt and Richard Thaler found that portfolios of past “loser” stocks later outperformed past “winners” over multi-year horizons, which they attributed to investor overreaction. Later studies by Fama and French and by Poterba and Summers reported long-horizon mean reversion in stock returns, though its strength remains debated. Over short horizons, momentum effects (where winners keep winning for a while) have also been documented, so whether price reverts depends heavily on the time horizon you measure.

The theory breaks down when the average itself moves. If a company loses a major customer, a product fails, or a central bank changes policy, the old average no longer describes where price “should” be. A stock that falls after a genuine deterioration in its business is not necessarily cheap. It may be settling at a new, lower level. A mean reversion trader has to separate a temporary excess from a change in fundamentals, and indicators alone cannot make that distinction.

How to calculate mean reversion: mean, standard deviation and Z-score

The core calculation measures how unusual the current price is. The steps:

1. Choose a lookback period that matches your holding period. Intraday traders may use minutes or hours of data, swing traders days or weeks, investors months or years.
2. Calculate the mean: Mean = Sum of prices / Number of observations
3. Calculate the deviation of the current price: Deviation = Price - Mean
4. Calculate the standard deviation of the price series to measure its typical variability: Standard deviation = Square root (Sum of squared deviations / (Number of observations - 1))
5. Calculate the Z-score: Z-score = Deviation / Standard deviation

The Z-score expresses the stretch in standard deviation units, which lets you compare a move in a volatile asset with one in a quiet asset. Thresholds of +1.5 or +2 are commonly used to flag a possible overvalued reading and -1.5 or -2 to flag a possible undervalued one. Higher thresholds produce fewer signals at more extreme levels.

In live trading the mean and standard deviation are usually calculated on a rolling window. Each new bar drops the oldest observation and adds the newest, so the mean drifts with price. This matters for exits: if you target “the mean,” the target keeps moving while you hold the trade.

Quantitative traders sometimes model a series as an Ornstein-Uhlenbeck process, a variable pulled toward a long-run level at a certain speed, and estimate a half-life: the expected time for a deviation to shrink by half. A long half-life ties up capital for a long time, which affects whether a setup is worth taking at all.

Indicators for a mean reversion strategy

Most mean reversion tools are variations on “distance from an average.”

• Moving averages. The moving average is the mean. A price beyond a threshold above it may be read as overbought, and one below it as oversold.
• Bollinger Bands. A middle band (often a 20-period simple moving average) sits between two outer bands, commonly two standard deviations away. A close outside a band is a statistical stretch, and the middle band is the natural reversion target. A narrowing “squeeze” signals low volatility and a possible larger move that can break away from the mean, so it is a warning as much as a setup.
• Relative Strength Index (RSI). It runs from 0 to 100. Readings above 70 are conventionally overbought and below 30 oversold. Short-period versions, such as a 2-period RSI, react quickly and suit short-term signals.
• Stochastic oscillator. It compares the close with the high-low range over a lookback period, often 14. Readings above 80 are overbought and below 20 oversold.
• MACD. Built from the difference between two exponential moving averages, it tracks momentum. Signal-line crossovers and divergences from price can suggest a move away from the mean is losing strength.
• Pivot points and support and resistance. These mark levels where reversals have occurred or may occur, which helps place entries and stops.

Combining a statistical stretch (a Z-score or band touch) with an oscillator reading and a nearby price level gives more confirmation than any single signal. It also produces fewer trades, and confirmation does not remove the risk of a trend continuing.

How to trade mean reversion: building the rules

A mean reversion system needs rules for five decisions.

1. Market filter. Decide when the strategy is allowed to trade. A common approach is to require a flat or gently sloping long-term moving average, or a low reading on a trend-strength indicator, so that you fade moves in ranges rather than in trends. Some traders trade only in the direction of the larger trend, buying dips in an uptrend instead of selling rallies.
2. Entry trigger. Define the extreme, for example a close below the lower Bollinger Band, a Z-score below -2, or a 2-period RSI under 10. Many traders wait for a sign that the move is stalling, such as a reversal candle (a hammer, doji or engulfing pattern, explained in the guide to candlestick charts), rather than buying while price is still falling.
3. Profit target. The mean itself (the moving average or middle band) is the obvious target. Some traders take partial profits at the mean and hold the rest toward the opposite band.
4. Stop-loss. Place the stop at a level where the reversion idea is clearly wrong, such as beyond a recent swing extreme or a Z-score of -3. A time stop (exit if price has not reverted within a set number of bars) limits how long capital stays tied up. Averaging down into a losing position instead of honoring the stop can turn one bad trade into a large one.
5. Position size. Because the stop is often further from entry than the target, size the position from the stop distance, not from conviction. A position size calculator converts account risk and stop distance into a lot size.

Trading against a move is sometimes called “fading” it. When price reaches a significant level and reverses, the fade buys the rejection of support or sells the rejection of resistance, with the stop on the other side of the level.

Mean reversion in day trading

Day traders open and close positions within the same session, so they apply mean reversion to intraday moves. Typical tools include short-period moving averages on minute charts, intraday Bollinger Bands and fast RSI or stochastic settings. The volume-weighted average price (VWAP) is another commonly used intraday mean, since it reflects the average price paid during the session weighted by volume.

The trade-off is noise. Short time frames generate more signals, and more of them are false, because small random moves easily push an oscillator to an extreme. Costs also weigh more heavily: a spread that is trivial against a 200-pip swing can consume a large share of a 10-pip intraday reversion. High-frequency mean reversion algorithms that profit from tiny, short-lived deviations depend on speed and low costs that retail traders rarely have.

Mean reversion swing trading

Swing traders hold positions for several days to a few weeks. They usually measure the mean with longer moving averages, such as the 20-day or 50-day, and look for a pullback that stretches price well away from that average before turning.

A typical swing setup combines:

• A larger trend filter, for example price above the 200-day moving average when looking for long trades.
• A short-term pullback to an oversold RSI or stochastic reading.
• A reversal candlestick at or near a support level.
• A Fibonacci retracement level as an additional reference. The levels most often watched are 38.2%, 50% and 61.8% of the prior move.

Swing trading spreads costs over larger moves than day trading. It also exposes positions to overnight and weekend gaps, which can jump past a stop-loss so the fill is worse than the stop price.

Win rate and expectancy for swing traders

Mean reversion rules often place the profit target (the mean) closer to entry than the stop. That payoff shape tends to produce many small wins and fewer but larger losses. A high win rate alone does not make a system profitable. Expectancy measures what each trade earns on average:

Expectancy = (Win rate × Average win) - (Loss rate × Average loss)

Hypothetical example: a system wins 70% of trades with an average gain of $100 and loses 30% with an average loss of $200. Expectancy = (0.70 × 100) - (0.30 × 200) = 70 - 60 = $10 per trade before costs. If round-trip costs average $12 per trade, the same system loses money despite winning seven times out of ten. Lower the win rate to 60% with the same averages and expectancy before costs becomes (0.60 × 100) - (0.40 × 200) = -$20. Because small changes in win rate have large effects when losses exceed wins, test rules on historical data and on a paper trading simulator before risking money.

Mean reversion in forex

Currency pairs often spend long periods trading within ranges, especially during quieter sessions, which makes them a natural market for mean reversion. Forex traders use the same moving averages, bands and oscillators as equity traders, plus a few market-specific tools.

Pivot points are calculated from the previous session’s high (H), low (L) and close (C):

• Pivot (P) = (H + L + C) / 3
• First resistance (R1) = (2 × P) - L
• First support (S1) = (2 × P) - H

Price that stretches to R1 or S1 and stalls may be faded back toward the pivot.

Session timing matters too. Ranges are often defined during lower-activity periods, while the overlap of major trading centers brings more volume and more breakouts. A range fade that looks sound in a quiet session can fail when the next session opens.

Currency correlations offer a relative-value version. When two pairs that have historically moved together diverge, a trader may buy the underperformer and sell the outperformer, expecting the gap to close. Correlations shift with interest rate policy and economic conditions, so a historical relationship can weaken without warning.

Central bank decisions and major data releases, such as employment or inflation reports, can push a pair to a new level that never reverts. Many mean reversion traders avoid holding range trades through scheduled high-impact events.

Pairs trading and statistical arbitrage

Pairs trading applies mean reversion to the relationship between two assets instead of to a single price. You identify two related instruments, such as two companies in the same industry, and track the ratio or spread between their prices. When the spread moves to an extreme Z-score, you go long the relatively cheap asset and short the relatively expensive one, then close both legs when the spread returns to its mean.

Because one leg is long and the other short, the position is partly hedged against moves in the overall market. What remains is the risk that the relationship itself breaks down. Correlation (two prices moving together) is not enough on its own. Quantitative traders often test for cointegration, a statistical property meaning that a combination of the two prices tends to return to a stable level over time. Statistical arbitrage extends the same idea to baskets of many securities, typically run by algorithms.

Worked example: a stock stretched above its mean

Hypothetical example. Company XYZ has averaged a closing price of $50 over the past 200 trading days, with a standard deviation of $5. After a strong earnings report, the stock jumps to $70.

Z-score = (70 - 50) / 5 = 4

A Z-score of 4 places the price four standard deviations above its 200-day mean, far beyond the common +2 threshold. A mean reversion trader might read this as overextended and consider a short position, with a target near the mean and a stop above the post-earnings high.

Suppose that over the next few weeks the excitement fades and the stock drifts back to about $52. The short captures most of the move from $70 toward the mean. Note that the 200-day average is also rising during this period because the new, higher prices enter the window, so the target moves up while the trade is open.

The same setup carries a clear risk. If the earnings report reflected a lasting improvement in the business, the $50 average is outdated and the stock may hold above $70 or keep rising. The Z-score measures how unusual a move is, not whether it is justified.

Mean reversion vs. trend following

• Trend following assumes a strong move will continue. It buys strength, sells weakness and earns most from a smaller number of large winners.
• Mean reversion assumes an extreme move will be reversed. It sells strength, buys weakness and earns from many modest winners.

Each works best where the other struggles. Some traders run both, or use a regime filter to choose between them, because a strategy that fades every rally gives back gains during a persistent trend.

Choosing a time frame and asset

No single time frame is best. The choice depends on your holding period, the time you can spend watching markets, your cost structure and the asset. Intraday mean reversion produces more opportunities with smaller moves and higher relative costs. Swing and position time frames produce fewer, larger moves with overnight risk. Longer horizons shade into mean reversion value investing, which buys assets whose valuation ratios, such as P/E, sit well below their historical norms and expects them to normalize over months or years.

Mean reversion is commonly applied to stocks, forex pairs, commodities, ETFs and bonds. Liquid instruments that tend to trade within ranges give the method the most room; assets prone to sustained trends or sudden repricing need stricter filters and stops. Testing a rule set on each instrument shows how often its extremes have actually reverted over your chosen horizon.

FAQs

The Bottom Line

Traders who fade extremes in ranging stocks, forex pairs or other liquid markets need firm rules for filters, entries, targets, stops and position size. The method fails when a stretch marks a real change in fundamentals, and indicators cannot tell the difference. Because many small wins can be erased by fewer large losses and costs, test any rule set on historical data and a simulator before risking money.