A volatility regime filter is a rule that classifies the market as quiet, normal or extreme, then allows or blocks entries based on that class. Most scalping losses don't come from bad setups. They come from good setups taken in the wrong environment. The filter fixes the environment problem.
This guide covers what the filter measures, how to build it with ATR, which thresholds to start with on 1m to 15m charts, and where it fails.
TL;DR
- A volatility regime filter ranks current volatility against recent history and only allows trades inside a defined band.
- The simplest version: ATR(14) ranked as a percentile over the last 100 bars, trade only between roughly the 20th and 90th percentile.
- It is a gate, not a signal. Pair it with a trend or momentum trigger.
- Tune it per asset and timeframe, then backtest it before you trust it.
- Leverage still magnifies losses. A filter lowers the number of bad trades. It does not cap risk.
Table of Contents
- What a volatility regime filter does
- Why scalpers need it more than swing traders
- The ATR percentile method
- Starting settings by timeframe
- Add a second dimension: trend slope
- Perspective: what we see in practice
- Common mistakes
- How to backtest a regime filter
- Risk: a filter is not a safety net
- Skip the build: regime logic inside our indicators
- FAQ
- Sources
What a volatility regime filter does
Markets cycle between states. Price compresses, expands, spikes, then compresses again. A regime is the state you are in right now. A filter turns that state into a yes or no answer for your strategy.
Three states cover most scalping needs:
- Dead: ranges are tiny. Spreads and fees eat the target. Breakouts fail.
- Normal: ranges are healthy. Targets get reached before stops.
- Extreme: news, liquidations, gaps. Slippage grows and stops get skipped.
A scalper's edge usually lives in the middle state. The filter's job is to keep you out of the other two.
This differs from the classic index filter used in longer-term systems, which defines whether the broader market is bullish or bearish, often with a moving average. That is a trend regime. A volatility regime asks a different question: how big are the moves, not which way are they going. You can run both together.
Why scalpers need it more than swing traders
On a 5-minute chart, a stop of a few ticks is normal. When volatility drops, the same stop becomes noise-sized, and the target is too far away. When volatility spikes, that stop is too tight for the actual swings. Fixed parameters only fit one regime.
Research backs the general idea that volatility is not constant. One arXiv paper on structural clustering of volatility regimes develops a method to find how many distinct volatility regimes exist in a financial time series and uses them for dynamic trading strategies. You don't need machine learning to benefit. A rule-based gate captures most of the value for intraday work.
For the mechanics behind this, see our guide on the role of volatility in scalping and the overview of volatility indicators every trader should know.
The ATR percentile method
The most common build uses Average True Range, because it captures gaps and wicks, not just closes.
Public open-source examples show the pattern. One ATR-based regime filter script on TradingView calculates ATR over 14 periods, ranks it against the last 100 periods on a 0 to 100 percentile scale, and allows trades only when the reading sits between 20 and 90. Under 20 is treated as a sleeping market. Above 90 is treated as erratic spikes such as news events.
The steps:
- Compute ATR(14) on your trading timeframe.
- Compute its percentile rank over a lookback window, for example 100 bars.
- Define the allowed band, for example 20 to 90.
- Allow entries only when the current rank is inside the band.
- Re-check the gate on every bar close, not mid-candle.
Step 5 matters. A reading that flickers in and out of the band while a candle is forming will give you a signal that disappears. Confirm on close, the same discipline a non-repainting signal follows.
Percentile versus ratio
Percentile rank is one way. Another is a ratio: ATR divided by its own moving average. A ratio of 1.0 means volatility matches its recent average. A value of 0.70 means 30% below. Because it is dimensionless, it travels across assets and price levels. Percentile is more robust to outliers. Ratio is easier to read at a glance. Pick one, then keep it consistent across your tests.
Starting settings by timeframe
These are starting points to test, not guarantees. Every asset behaves differently, and session hours change the picture.
| Timeframe | ATR length | Percentile lookback | Allowed band | Notes |
|---|---|---|---|---|
| 1m | 14 | 200 to 300 bars | 25 to 85 | Noisy. Use a longer lookback to stabilize rank. |
| 5m | 14 | 100 to 200 bars | 20 to 90 | The standard baseline. |
| 15m | 14 to 20 | 100 bars | 20 to 90 | Fewer signals. Wider band is fine. |
For deeper settings logic, our post on volatility gating settings with ATR 14 to 20 on TradingView walks through the parameters step by step.
Add a second dimension: trend slope
Volatility alone says the market is active. It does not say the move is tradable. Many traders add a higher-timeframe slope check. One open-source ATR regime script pairs a normalized volatility classifier with a higher-timeframe EMA slope detector and outputs three states: trade, stay out, or extreme. Its notes suggest tightening the ATR average length to 20 to 30 and using a 1H or 4H EMA for scalping.
A practical version of this stack:
- Gate 1: ATR percentile inside 20 to 90.
- Gate 2: 1H EMA slope agrees with your trade direction.
- Trigger: your entry signal on the lower timeframe.
The more gates you stack, the fewer trades you get. That is the point, but check that you still have enough samples to evaluate the system. Our piece on multi timeframe scalping signals covers how to lock in 1:1 to 1.5:1 risk to reward with that kind of confluence.
Perspective: what we see in practice
We build short-term tools, and the same pattern shows up again and again. Traders blame the signal when the real issue was the regime. A signal tuned for expansion fires in a dead tape, hits its stop, and gets labeled broken.
Here is how we think about it:
- The filter comes first. Volatility gating is a core part of how our indicators are designed, alongside confluence and divergence detection. Signals are confirmed on candle close and do not repaint.
- Adaptive beats fixed. Fixed pip or tick thresholds break when conditions change. Percentile and ratio methods adapt on their own.
- Transparency matters. Our scripts are open-source. If you don't understand why a parameter is set, you can read it. Blindly following output is how traders get hurt.
- Fewer trades, cleaner trades. A good gate feels boring. If it never says no, it isn't doing its job.
Our Edge Finder is built around regime awareness, and our team's rule of thumb is simple: check the regime before the signal, every time.
— Tran
Common mistakes
Over-optimizing the band
Testing every band from 5 to 95 and picking the best backtest result is curve fitting. Choose a sensible band, test neighbors, and look for a plateau. If 20 to 90 works and 18 to 92 and 22 to 88 also work, the result is stable. If only one pair works, it is noise.
Too short a lookback
A 20-bar lookback makes percentile rank jumpy. The filter flips constantly. Start at 100 and move up for lower timeframes.
Ignoring session effects
Volatility has a daily rhythm. The open of the U.S. stock session is not the same as the late afternoon. The SEC notes that after-hours trading can bring greater price fluctuations for stocks with limited trading activity, and that news announced then may have a bigger effect on prices. A percentile computed across a full 24 hours may mislead for equity-linked instruments. Test the filter inside the session you actually trade.
Treating the filter as a signal
The gate says the environment is acceptable. It does not say buy or sell. You still need an entry trigger and a stop.
Hard-coding the filter and never reviewing
Regimes shift over months. Review the band quarterly, or after a major change in market structure.
How to backtest a regime filter
A fair test compares the same strategy with and without the gate:
- Run your entry logic on at least several hundred trades, ungated.
- Add the filter, same stops, same targets.
- Compare win rate, average R, max drawdown and trade count.
- Split the sample: tune on one period, verify on another.
- Include fees and spread. Scalping edges are thin, and costs decide outcomes.
A good filter usually cuts trade count, lifts average R and shrinks drawdown. If it only shrinks trade count, it isn't helping. For a full process, read our guide on how to backtest trading strategies.
Risk: a filter is not a safety net
Scalping is fast, and many retail traders use leverage. The SEC's forex bulletin warns that leverage magnifies small price moves, so you risk losing all your initial capital and may lose more than you deposited. Only trade money you can afford to lose. Size positions so one failed trade is small, even in a normal regime.
Extreme regimes are where this bites hardest. The upper edge of your band exists to keep you out of those moments. If you trade forex, review our forex risk management tips for retail traders. If you are weighing how to hold other assets through turbulent stretches, a guide to assets for volatile markets covers the longer-term side, which is a separate decision from intraday trading. Talk to a licensed financial adviser about retirement money specifically.
Automated rules add their own risk. The SEC's day trading guidance stresses the extreme risk of rapid price changes. Know your parameters. Test on a demo account first.
Skip the build: regime logic inside our indicators
If you'd rather not code and maintain the gate yourself, our TradingView suite includes volatility gating, multi-timeframe confluence, divergence detection and projected TP/SL. Signals are non-repainting and confirmed on candle close. You can also route alerts to Discord by webhook.
- Edge Finder and its regime indicator page for regime-aware setups.
- Smart Scalping Signals for buy and sell signals on 1m to 15m.
- Momentum Oscillator as a trigger to pair with the gate.
Plans are $79 per month, $799 per year, or $1,999 one-off for lifetime access. See the membership options for details. More practical shortcuts live on our Trading Tricks & Tips page.
FAQ
What is a volatility regime filter?
It is a rule that measures current volatility, usually with ATR, compares it to recent history, and allows or blocks trades based on the result. It keeps you out of dead and chaotic markets. It does not tell you direction.
What ATR settings work best for a regime filter?
ATR(14) is the common default, with 20 to 30 sometimes used for scalping filters. Rank it over a 100-bar window or longer. Treat these as starting points and test them on your own asset and session.
Does a volatility filter work in crypto and forex?
The logic applies to any market with measurable ranges. Crypto trades around the clock and forex has session cycles, so test the lookback against the hours you trade. Remember that leverage amplifies losses in both.
Can a regime filter repaint?
It can if it reads values from a candle that has not closed. Evaluate the gate on bar close and your signals will stay fixed once printed.
Sources
- Volatility Regime Filter (ATR based), TradingView
- ATR Regime Filter Pro, TradingView
- Structural clustering of volatility regimes for dynamic trading strategies, arXiv
- After-Hours Trading: Understanding the Risks, SEC
- Foreign Currency Transactions, SEC
- Day Trading: Your Dollars at Risk, SEC