OPEN-SOURCE SCRIPT
Bollinger Bands Unbound [Probalist Essentials]
A Bollinger Bands that actually tells you when not to fade the band. Most BBand scripts show a band touch and leave you to guess whether it's a reversal or the first bar of a trend walk. This one answers that. The Band-Walk Guard flags every stretch where price keeps closing outside the band — trend mode, fade suppressed — and the Reversion Map measures, on your actual chart history, where confirmed fades went: what fraction tagged the basis, what the median reversion distance was, how wide the 1σ spread is, and whether the recent regime still matches the long-run.
Built for traders who want more than a line on a chart. The squeeze regime, the walk length, the basis-reach rate, the distribution panel — each answers a specific question about volatility state and mean-reversion quality that the default indicator leaves you to figure out yourself.
🟡 WHY THIS VERSION
Most Bollinger scripts treat every band touch as a trade. This one flags the band walk so you know when it's a trend, not a fade. The Reversion Map goes further: it counts how often your own chart's confirmed fades actually tagged the basis, shows the median reversion distance and a 1σ spread as a bell curve right of price, and splits bull from bear (they're different animals in a drifting market). You get a Wilson-backed verdict — not a fixed 60% bar, but a proper statistical lower bound — plus recent vs long-run form so you can see when the edge has changed regime. The edge clock shows when backed evidence last fired, not just whether it 'exists'. It keeps score honestly: small n prints 'gathering', wide bands stay wide, no numbers are hidden.
🟡 HOW IT WORKS
Bollinger Bands: basis = SMA(src, len) — or EMA via the Basis MA type input (the bands stay basis ± mult × stdev either way); upper/lower = basis ± mult × stdev(src, len). %B = (src − lower) / (upper − lower). Bandwidth = (upper − lower) / basis × 100, expressed as a percentage of price so it's scale-free across any instrument.
The Band-Walk Guard counts consecutive confirmed closes outside each band. A walk counter ticking upward is trend state — the outer band edge ignites in the heat ramp colour so you see it immediately. The Reversion Map fires only when price confirms back inside: a confirmed re-entry after a close beyond the lower band is a bullish fade; beyond the upper band is bearish. Each confirmed fade's outcome is recorded i_horizon bars later (the distance in %return, signed in the signal's direction), building two rolling arrays (bull/bear) of up to 200 outcomes. From those arrays: win-rate, median, 16th–84th percentile (≈1σ), Wilson lower bound, and recent-form over the last 60 outcomes. The bell curve drawn right of price is the kernel-smoothed empirical density of those returns — what actually happened, not a fitted model. Squeeze uses a rolling bandwidth percentile: self-calibrating, never an absolute number, works across all volatility regimes. With the optional RSI confluence filter on, a fade only counts when RSI was at an extreme on the excursion bar — and because the outcome arrays rebuild over the full history, toggling the filter shows like-for-like what the classic BB+RSI double-confirm does to the numbers on this chart.
🟡 KEY FEATURES
🟡 HOW TO USE
🟡 PAIRS WELL WITH
A simple trend filter (price vs a longer moving average, or ADX slope direction) tells you which side of the mean-reversion to favour — the Reversion Map treats bull and bear fades separately, but a trend context makes even those reads sharper. Volume (relative volume or OBV) helps validate whether a fade has the force to carry: a low-volume re-entry into the band after a walk is weaker context than one with a volume spike. For timing confluence, a momentum oscillator (RSI or Stochastic) reaching oversold/overbought at the same time as a fade signal gives a second read on exhaustion. Key daily or weekly support/resistance levels provide a structural backdrop — a lower-band fade off a major S/R is meaningfully different from one in open air. None of these gate the signal here; they're the context a trader builds around it.
Bollinger Bands Unbound gives you the two answers the default version never does: whether a band excursion is a walk (stay out) or a confirmed fade (measure it), and where that fade actually went on your own chart. The statistics are descriptive — your history, counted honestly. Use the walk guard to stay out of trend mode, the Reversion Map to understand what confirmed fades have been worth, and the evidence-gated alerts when the record actually supports a bias.
Open source under MPL-2.0. The probability layer describes past signals on your chart — it is a measurement, not a prediction, and nothing here is financial advice.
Built for traders who want more than a line on a chart. The squeeze regime, the walk length, the basis-reach rate, the distribution panel — each answers a specific question about volatility state and mean-reversion quality that the default indicator leaves you to figure out yourself.
🟡 WHY THIS VERSION
Most Bollinger scripts treat every band touch as a trade. This one flags the band walk so you know when it's a trend, not a fade. The Reversion Map goes further: it counts how often your own chart's confirmed fades actually tagged the basis, shows the median reversion distance and a 1σ spread as a bell curve right of price, and splits bull from bear (they're different animals in a drifting market). You get a Wilson-backed verdict — not a fixed 60% bar, but a proper statistical lower bound — plus recent vs long-run form so you can see when the edge has changed regime. The edge clock shows when backed evidence last fired, not just whether it 'exists'. It keeps score honestly: small n prints 'gathering', wide bands stay wide, no numbers are hidden.
🟡 HOW IT WORKS
Bollinger Bands: basis = SMA(src, len) — or EMA via the Basis MA type input (the bands stay basis ± mult × stdev either way); upper/lower = basis ± mult × stdev(src, len). %B = (src − lower) / (upper − lower). Bandwidth = (upper − lower) / basis × 100, expressed as a percentage of price so it's scale-free across any instrument.
The Band-Walk Guard counts consecutive confirmed closes outside each band. A walk counter ticking upward is trend state — the outer band edge ignites in the heat ramp colour so you see it immediately. The Reversion Map fires only when price confirms back inside: a confirmed re-entry after a close beyond the lower band is a bullish fade; beyond the upper band is bearish. Each confirmed fade's outcome is recorded i_horizon bars later (the distance in %return, signed in the signal's direction), building two rolling arrays (bull/bear) of up to 200 outcomes. From those arrays: win-rate, median, 16th–84th percentile (≈1σ), Wilson lower bound, and recent-form over the last 60 outcomes. The bell curve drawn right of price is the kernel-smoothed empirical density of those returns — what actually happened, not a fitted model. Squeeze uses a rolling bandwidth percentile: self-calibrating, never an absolute number, works across all volatility regimes. With the optional RSI confluence filter on, a fade only counts when RSI was at an extreme on the excursion bar — and because the outcome arrays rebuild over the full history, toggling the filter shows like-for-like what the classic BB+RSI double-confirm does to the numbers on this chart.
🟡 KEY FEATURES
- Band-Walk Guard: counts consecutive outside closes and flags trend state with an ignited band edge. During a walk no fade can fire by definition — a fade requires the confirmed close back inside, the very bar that ends the walk — so the guard shows you why the tool is silent
- Reversion Map: empirical forward-return distribution per direction (bull/bear fades tracked separately), shown as a kernel-smoothed bell curve right of price
- Basis-reach rate: % of past fades that tagged the basis within the horizon, displayed as a compact label on the basis line — the natural mean-reversion target with its own base rate
- Squeeze detection: self-calibrating bandwidth percentile (default 10th over 120 bars), squeeze shading, release-breakout alert
- Weighted signal dots: excursion depth (in rolling stdev units) drives dot size via self-calibrating 60th/90th percentile thresholds — tier 3 dots are the deep outlier fades
- Wilson-backed verdict: lower bound of win-rate tested against 50% at user-chosen confidence (80/90/95%); recent form (last 60 outcomes) vs long-run both shown
- Edge clock: stamps when backed (●) and leaning (◐) evidence last appeared on an actual signal — tells you if the edge is live or stale
- Probalist heat ramp: %B mapped to ±100 drives the fill colour from strong blue (price at lower band) through mint neutral to orange/amber (price at upper band)
- Optional RSI confluence filter — the classic BB+RSI double-confirm, built in. With the filter on, the Reversion Map measures the filtered signal, so you can A/B on your own chart whether the famous combo actually improves the outcomes.
- Reversion-complete markers: a small ✓ on the basis line the moment a live fade tags its mean-reversion target — the exact event the basis-reach rate counts, so the entry dot and the exit ✓ tell the whole arc on the chart.
🟡 HOW TO USE
- Band walk in readout? That's trend state. The outer band edge ignites — use it as a filter, not a fade setup. Wait for the walk to end and a confirmed re-entry before looking for a fade.
- Fade signal (gold dot below price = bullish, blue dot above = bearish) fires on confirmed close back inside the band. Dot size = excursion depth; larger dots = deeper, statistically rarer excursions.
- Check the Reversion Map rows in the readout: 'n' below 20 = gathering, don't weight the stats. Once n grows, median and the 1σ band tell you how far fades typically carried and how consistent they were.
- Basis-reach rate on the basis line: if 70%+ of past lower-band fades tagged the basis within 20 bars, the basis is the realistic mean-reversion target on this chart — a number to weigh, not an instruction.
- Squeeze shading (green tint) = bandwidth at the bottom decile. A squeeze release with a close outside the band often starts a new walk, not a fade — the readout flags it.
- Evidence-gated alerts fire only when Wilson lower bound > 50% — that's a statistical bar, not a 50/50 flip. A backed signal with a small median (e.g. +0.15% on a 5-minute chart) may not survive round-trip costs; the edge read is about direction bias, not net P&L.
- Combine with a trend filter above (e.g. price vs 200 SMA) to take only same-direction fades — bullish fades when above the 200, bearish when below. Reduces signal count, improves composition.
- Non-repainting by construction: the confirmation bar closes, then the signal fires. Realtime mode (confirm=0) updates on the unconfirmed bar — note the repaint risk if you use it.
- Curious whether the BB+RSI double-confirm helps on your chart? Turn on the RSI confluence filter and compare the readout stats with the filter off — the arrays rebuild over the full history, so the comparison is like-for-like.
🟡 PAIRS WELL WITH
A simple trend filter (price vs a longer moving average, or ADX slope direction) tells you which side of the mean-reversion to favour — the Reversion Map treats bull and bear fades separately, but a trend context makes even those reads sharper. Volume (relative volume or OBV) helps validate whether a fade has the force to carry: a low-volume re-entry into the band after a walk is weaker context than one with a volume spike. For timing confluence, a momentum oscillator (RSI or Stochastic) reaching oversold/overbought at the same time as a fade signal gives a second read on exhaustion. Key daily or weekly support/resistance levels provide a structural backdrop — a lower-band fade off a major S/R is meaningfully different from one in open air. None of these gate the signal here; they're the context a trader builds around it.
Bollinger Bands Unbound gives you the two answers the default version never does: whether a band excursion is a walk (stay out) or a confirmed fade (measure it), and where that fade actually went on your own chart. The statistics are descriptive — your history, counted honestly. Use the walk guard to stay out of trend mode, the Reversion Map to understand what confirmed fades have been worth, and the evidence-gated alerts when the record actually supports a bias.
Open source under MPL-2.0. The probability layer describes past signals on your chart — it is a measurement, not a prediction, and nothing here is financial advice.
Skrypt open-source
W zgodzie z duchem TradingView twórca tego skryptu udostępnił go jako open-source, aby użytkownicy mogli przejrzeć i zweryfikować jego działanie. Ukłony dla autora. Korzystanie jest bezpłatne, jednak ponowna publikacja kodu podlega naszym Zasadom serwisu.
Wyłączenie odpowiedzialności
Informacje i publikacje nie stanowią i nie powinny być traktowane jako porady finansowe, inwestycyjne, tradingowe ani jakiekolwiek inne rekomendacje dostarczane lub zatwierdzone przez TradingView. Więcej informacji znajduje się w Warunkach użytkowania.
Skrypt open-source
W zgodzie z duchem TradingView twórca tego skryptu udostępnił go jako open-source, aby użytkownicy mogli przejrzeć i zweryfikować jego działanie. Ukłony dla autora. Korzystanie jest bezpłatne, jednak ponowna publikacja kodu podlega naszym Zasadom serwisu.
Wyłączenie odpowiedzialności
Informacje i publikacje nie stanowią i nie powinny być traktowane jako porady finansowe, inwestycyjne, tradingowe ani jakiekolwiek inne rekomendacje dostarczane lub zatwierdzone przez TradingView. Więcej informacji znajduje się w Warunkach użytkowania.