Skip to main content

Bollinger-Band Reversion

Updated Jun 24, 2026 at 2:51pm

Research Draft Medium 828 words

Bollinger-Band reversion is a swing-trading mean-reversion setup that fades a price stretch out to (or through) the lower Bollinger Band, betting that price snaps back toward the 20-period middle band. Bollinger Bands plot a 20-period simple moving average with an upper and lower band set two standard deviations away, forming an adaptive envelope that contains most price action. Because the bands widen and contract with volatility, an extension to the lower band marks a statistically stretched, "low relative to itself" condition — the raw material for a fade. The middle band is both the mean and the natural profit target.

The setup

The long version: in a sideways or range-bound stock, wait for price to push down to or briefly through the lower band, signalling a stretched, oversold-relative-to-recent-range condition. Do not buy the tag itself. The trigger is a close back inside the band — Bollinger's own rule is that a tag of a band is not a signal in and of itself, and closes outside the bands are initially continuation signals, not reversals. Waiting for the close to re-enter confirms the snap-back is actually beginning and filters out price that is going to keep going.

  • Entry: on or just after the close that re-enters the band from below.
  • Stop: below the extreme of the stretch (the lowest low / wick that tagged or pierced the band). If price makes a new low beyond that point, the range read was wrong and the trade is invalidated.
  • Target: the middle band (20-SMA midline) — the mean the trade is reverting to. A partial scale at the midline with a runner toward the opposite (upper) band is a common variation in clean ranges.

The short side is the mirror image: fade a stretch to the upper band, enter on the close back inside, stop above the high, target the midline. The same Bollinger caveat applies — an upper-band tag is not a sell signal by itself.

Base rates & evidence

The honest framing: this setup has a real edge only in ranging regimes, and is a reliable account-killer in trends. There is no single authoritative win-rate to quote — published figures vary widely by instrument, lookback, and exit rule, so treat any precise number with suspicion. What the source material consistently agrees on is the direction of the edge:

  • In a flat, mean-reverting market, price reaching a band tends to revert toward the 20-SMA (the "Bollinger bounce"). This is where the setup earns its keep.
  • In a trend, price "walks the band" — it repeatedly touches or rides the lower band on the way down (or the upper band on the way up) without reverting. Bollinger states explicitly that in trending markets price can and does walk up the upper band and down the lower band. Fading every tag in that environment produces a string of losers as you keep buying into a falling market.

So the realistic expectation is a setup with a respectable hit rate and a capped, midline-distance reward when the regime filter is correct, and a sharply negative expectancy when it is not. The regime call dominates the outcome more than the entry mechanics do.

Strengths & limitations

Strengths: clear, objective levels (band, midline, extreme); a built-in volatility-adaptive definition of "stretched" that auto-adjusts as volatility changes; a defined, nearby target (the mean) that supports a favorable reward-to-risk in tight ranges; and a confirmation trigger (close back inside) that removes the worst knife-catching entries.

Limitations / the #1 misuse: treating the band tag as a standalone signal and fading strength in an uptrend or buying weakness in a downtrend. This is the canonical way the setup destroys accounts — shorting an upper-band tag while price is walking the band higher, or buying lower-band tags as a stock trends to new lows. Secondary pitfalls: no fixed stop (mean-reversion losers run if you don't cap them), choppy regimes that fake the re-entry close, and over-tight targets that exit before the move pays. The defenses are mandatory: a regime filter, the close-back-inside trigger rather than the tag, and a hard stop beyond the extreme.

System relevance

In the Augustus pipeline this setup is gated to range / non-trending regime before it is allowed to fire. The regime classifier must read the instrument as range-bound (low directional persistence, contained bands) for a Bollinger-reversion long or short to be eligible; in a classified trend the setup is suppressed precisely because of the walk-the-band failure mode. The 20-SMA midline serves as the default mechanical target.

Sources