What is mean reversion trading?
Mean reversion is the idea that price swings around a fair value, and that extreme moves away from it tend to fade. A burst of selling from a large order, a run on stops or an overreaction to a headline can push price far below its average; once that pressure is spent, price often drifts back. The drawing at the top shows the ideal case: a sharp drop far below the lower band, a reversal candle, then a return to the dashed 20 MA.
Note the word “often”. Reversion can also happen the other way round: price stays down and the moving average slowly falls to meet it, so a trade that waited for a bounce goes nowhere. And when the stretch has a real cause, such as earnings, the old average is simply wrong and price may never come back to it.
- Mean
- The reference average you measure from: VWAP, a moving average or the Bollinger middle band. It moves with every new candle.
- Stretch
- How far price is from the mean, measured with bands, in percent, in multiples of ATR or by an RSI extreme.
- Fade
- Trading against the latest move: buying a sharp drop or shorting a sharp rise, expecting a return toward the mean.
- Falling knife
- A fast, steep drop. Buying it before it stops, “catching a falling knife”, is the classic mean reversion mistake.
Choosing the mean: VWAP, the 20 MA or the Bollinger middle band
The mean is your target, so pick it before the session and keep it. The three common choices for day trading are listed below.
The drawing below shows a mean doing its job on a trend day: price runs above VWAP, comes back to it and closes back above. A fade of the run up ended at VWAP; the move after it belonged to trend traders. That is the rule for every mean reversion trade: the mean is the finish line. More in the VWAP trading strategy guide.
- VWAP (volume-weighted average price): the average price of the session, weighted by volume. It starts fresh each session and uses each bar’s typical price, (high + low + close) ÷ 3, times its volume, so it shows where the day’s volume has traded on average. The natural mean for intraday fades.
- 20-period moving average: on a 5-minute chart, the average of the last 20 closes, about 100 minutes of trading. It works on any timeframe and does not reset overnight.
- Bollinger middle band: by default the same 20-period simple moving average, with bands two standard deviations above and below it, so you get the mean and a stretch gauge in one tool.
How to measure “stretched”: bands, distance and RSI
A stretch only means something next to how far this stock usually moves. The three common gauges are listed below; use two at a time.
Hypothetical 5-minute chart: the 20 MA sits at $50.00 and the bands at $49.20 and $50.80. A burst of selling drives price to $48.50, $1.50 below the mean, with RSI (14) at 24. All three gauges agree that price is stretched; none of them says the selling is over.
Not every close outside a band is a stretch to fade. In the drawing below, the bands narrow into a squeeze, then widen as price closes above the upper band. Bollinger treats closes outside the bands as continuation signals at first, so that is a breakout setup, not a fade. Prefer stretches where the bands were already steady.
- Bollinger Bands (20, 2): a close at or below the lower band is low relative to the last 20 bars. John Bollinger’s own rule is that a tag of a band is a tag, not a signal.
- Distance from the mean in ATR: for example, price more than 2 × ATR (14) below VWAP. Take the number from how far this stock usually stretches, not from a book. See the average true range guide.
- RSI extremes: RSI (14) under 30 or over 70, the oversold and overbought lines used by its creator, J. Welles Wilder Jr. In a strong trend, RSI can stay under 30 for a long time.
Wait for a confirmation candle before you enter
A stretch tells you where to look, not when to act. Price that is “too low” can get much lower, so wait for proof that the selling has stopped: a candle that closes in your direction. That can be a hammer or a bullish engulfing candle at the low (the bearish versions at a high) from the candlestick patterns guide, a close back inside the band after a close outside it, or a close above the previous candle’s high after a run of lower highs.
Location adds weight. A stretch that lands on yesterday’s low, a prior swing low or another support level has a reason to stop there; a stretch into empty space does not.
In the example, the $48.50 low sits on yesterday’s low at $48.55. By then the bands have widened and the lower band is at $48.90. The next 5-minute candle has a long lower wick and closes at $49.00, back inside the band: that close is the entry. Buying the touch at $48.50 would have saved 50 cents, but the same touch on a day when price kept falling would have been a falling knife.
Stop, target and size: the mean is the target
Stop: beyond the extreme, plus a small buffer. The low of the stretch is where the idea is proven wrong; if price trades below it after your confirmation candle, the sellers are back. In the example the stop goes at $48.40, ten cents under the $48.50 low, so the risk is $0.60 a share.
Target: the mean, as it stands when you enter. The 20 MA has slipped to $49.90 while price was falling, so the reward is $0.90, a risk/reward of 1:1.5. That is the catch in mean reversion: the mean sets the target, and the closer it drifts to your entry, the less you can make. If the reward falls below your minimum, skip the trade; do not move the stop closer to make the numbers work. See the risk-reward ratio guide.
Size: risking 1% of a $10,000 account, or $100, allows 166 shares ($100 ÷ $0.60). At the target that is +$149.40, or +1.5R; at the stop, −$99.60, or −1R, plus costs. The position size calculator does the math.
Two exits many traders add: take the trade off at the mean, where the edge you were trading is spent, and use a time stop, such as leaving if price has not moved toward the mean within six 5-minute candles.
UpTickFlow, our invite-only TradingView indicator, follows the same confirmation-first idea, so weigh this accordingly. It does not measure stretches or read Bollinger signals. It waits for a candle to close at a level, draws the entry, stop and target before the trade, scores the setup from 0 to 100 and stays quiet in chop or when there is no room to the target: a second check, not a scanner.
Why mean reversion fails: trends, news and the falling knife
Mean reversion fails often, and a failure starts out looking exactly like the setup.
Hypothetical: the same stock on another day, after a downgrade before the open. It drops below its lower band to $48.50, prints a small green candle, and you buy at $48.80 with a stop at $48.30. The stop is hit. Two more “stretched” buys at $47.60 and $46.70 are stopped out too, as in the drawing below. At 1% risk each, that is about 3% of the account: a bad day, but a planned one. A trader with no stop who kept averaging down is still holding at $45.00, waiting for a mean that has moved away.
A common rule: after two failed fades in the same direction, stop fading that stock for the day. Most failures come from three causes:
- Trends. In a strong trend, price can ride a band for hours: every close looks stretched and the next one is lower. Bollinger notes that prices can walk up the upper band and down the lower band.
- News. Earnings, a downgrade or a Fed decision (the statement comes out at 2:00 p.m. ET) can reset what a stock is worth. The old average is no longer fair value, and price owes it no return.
- Falling knives. Buying a fast drop before any candle closes back up, often with a limit order resting at a band, puts you in front of sellers who are not finished.
The ADX filter: trade mean reversion in ranges, not trends
Since trends break mean reversion, check the trend first. The Average Directional Index (ADX), introduced by J. Welles Wilder Jr. in 1978 and usually calculated over 14 periods, measures trend strength from 0 to 100 without saying which way. A common reading: above 25 is a trend; below 20 is a weak trend or none.
The drawing below shows both regimes. On the left, price trends and ADX climbs above 25: fades there get run over. On the right, price chops sideways and ADX sinks below 20: stretches from a flat mean tend to come back. See the ADX indicator and range trading strategy guides.
ADX lags, and a quiet range can end in a breakout, so use it with three more checks:
- The slope of the mean: a flat 20 MA or VWAP suits fades; a steep one means a trend, whatever ADX says.
- The higher timeframe: a 5-minute stretch into 1-hour support has room to bounce; one against a 1-hour downtrend usually does not. See multi-timeframe analysis.
- The clock and the calendar: skip the first minutes after the open, the minutes around a Fed statement and stocks with fresh news.
Mean reversion vs trend following
The two styles make opposite bets. Mean reversion buys weakness and sells strength, expecting a return to the average. Trend following buys strength and sells weakness, expecting the move to continue, as in the pullback and breakout strategies.
Neither is better. They suit different conditions, and mixing them on the same chart without a filter, fading the very moves you would otherwise follow, is a common way to lose on both.
- Entry: after a stretch away from the mean, versus on a break or a pullback toward it.
- Target: the nearby mean, versus letting winners run to the next level or trailing a stop.
- Where each fits: ranges with a low ADX, versus trends with a rising ADX.
- How it feels: mean reversion tends to bring frequent small wins and, without stops, rare large losses; trend following, frequent small losses and rare large wins.