The short answer
Classify gold’s regime with two daily measurements and one warning sign: ATR(14) as a percentage of price, ranked against the past year (quiet, normal or high volatility), ADX(14) for trend strength (trending, unclear or ranging), and range-expansion days when the day’s true range is at least 1.5 times ATR. Then adjust: smaller size and wider stops when volatility is high, and different setups when the trend reading changes. Regimes describe the backdrop; they never tell you the direction of the next trade.
We tested these rules on COMEX gold futures daily bars from 4 January 2010 to 3 November 2026 (4,068 sessions with full high and low data), the same data source behind our gold volatility tool. The results are below, including the ones that didn’t match the textbook.
The three rules
| Rule | Measure | Regime |
|---|---|---|
| Volatility | ATR(14) ÷ close, as a percentile of the last 252 sessions | Below 25th: low. Above 75th: high. Otherwise normal |
| Trend | ADX(14), Wilder’s method, daily | 25 or more: trending. Below 20: ranging. 20 to 25: unclear |
| Range expansion | Today’s true range ≥ 1.5 × yesterday’s ATR(14) | A warning day: volatility may be stepping up |
All three come from J. Welles Wilder’s 1978 book New Concepts in Technical Trading Systems, which defined true range, ATR and ADX. We fixed the thresholds before testing and did not optimise them. For the indicators themselves, see ATR and Bollinger Bands on gold and the ADX indicator on gold.
Why a percentile rather than a dollar figure? Gold’s price has risen from around $1,100 to the $4,000s since 2010. A $40 day meant something very different at each level. Ranking ATR as a percentage of price against its own recent history adjusts for that.
What the data showed
1. How often each regime occurs
| Volatility regime | Share of days | Trend regime | Share of days |
|---|---|---|---|
| Low | 29% | Trending (ADX ≥ 25) | 41% |
| Normal | 46% | Unclear (20–25) | 21% |
| High | 24% | Ranging (ADX < 20) | 38% |
2. Volatility regimes persist
| Regime today | Same regime 20 sessions later | Average ATR % today | Average ATR % 20 sessions later |
|---|---|---|---|
| Low | 58% | 1.04% | 1.14% |
| Normal | 59% | 1.42% | 1.44% |
| High | 60% | 1.93% | 1.77% |
High-volatility periods stayed elevated a month later on average, while drifting back somewhat; quiet periods stayed quiet more often than not. This is volatility clustering: big moves tend to be followed by big moves, small by small, which Benoit Mandelbrot described for commodity prices in 1963. Practical meaning: if gold has been wild this week, size for a wild next week too.
3. Range-expansion days are a useful warning
Days with a true range of at least 1.5 × ATR made up 14.7% of sessions. Over the next five sessions, the average true range was 1.19 × the prior ATR, against 1.03 × after all days. An unusually large day tends to be followed by larger days than normal.
4. ADX did not make breakouts more reliable
This is the result that runs against common advice. We looked at days when gold closed above its highest close of the previous 20 sessions, and checked whether it was higher 10 sessions later:
| ADX regime on the breakout day | Breakouts | Higher 10 sessions later | Average 10-session change |
|---|---|---|---|
| Trending (ADX ≥ 25) | 296 | 49% | +0.24% |
| Unclear (20–25) | 127 | 53% | +0.42% |
| Ranging (ADX < 20) | 181 | 66% | +0.72% |
| All breakouts | 604 | 55% | +0.43% |
For comparison, 55% of all days were followed by a higher close 10 sessions later, because gold rose strongly over the period. Breakouts out of a low-ADX, ranging market followed through more often than breakouts in an already-strong trend, where much of the move may have happened before the signal. Treat this as one test on one market, not a rule: the samples overlap (consecutive days share future prices), there are no costs in it, and gold’s long uptrend flatters anything bullish.
What to change in each regime
| Regime | Size | Stops and targets | Setups that fit |
|---|---|---|---|
| High volatility | Smaller, so the dollar risk stays fixed | Wider stops; targets scaled to ATR | Fewer trades; avoid tight-stop scalps |
| Low volatility | Normal | Tighter stops are possible, but watch for expansion | Range edges; prepare for breakouts |
| Trending | Normal | Trail stops behind structure | Pullbacks in the trend direction |
| Ranging | Normal or smaller | Targets at the opposite edge | Fades at range edges; watch for a break of the range |
| Expansion day | Reduce until the next sessions settle | Re-measure ATR before new trades | Wait for structure to form again |
Sizing across regimes (hypothetical). A $10,000 account risks 1% ($100). In a normal regime the stop is 1 × ATR = $40, so size = $100 ÷ ($40 × 100) = 0.025 lot. If ATR doubles to $80 and the stop doubles with it, size = $100 ÷ ($80 × 100) = 0.0125 lot, so in practice 0.01 lot. Same dollar risk, half the position.
The risk management for volatile gold markets guide covers volatility-based sizing in more depth, and the ATR stop-loss calculator does the arithmetic.
Where gold stood on 3 November 2026
On the last day of our data, gold’s ADX(14) was about 32 (trending band), and its ATR was about 2.1% of price. That sounds high, and in long-run terms it is, yet it ranked only around the 18th percentile of the previous 252 sessions, because the year to November 2026 had ATR readings between about 1.7% and 4.8% of price. Regimes are relative: “low” here means low for the recent past, not calm. Check the live figures on the volatility tool.
A weekly regime check (five minutes)
- Note ATR(14) and the 20-day average range on the volatility tool.
- Rank today’s ATR against the past year: bottom quarter, middle or top quarter?
- Read daily ADX(14): above 25, below 20 or in between?
- Check the last five sessions for expansion days.
- Check whether gold’s usual macro links still hold; when they break, a regime change is often underway. See gold correlation breakdowns.
- Write the regime and your size and stop adjustment at the top of your weekly plan.
Limits
- Lagging measures. ATR and ADX use the last 14 sessions, so they confirm a change after it starts.
- Thresholds are conventions. 20 and 25 for ADX and the quartiles for ATR are common choices, not laws.
- One market, one period. These results describe COMEX gold from 2010 to 2026. They may not hold in future data.
- Futures, not your provider’s quote. COMEX futures track spot XAUUSD closely, but your platform’s highs and lows can differ slightly.
Sources
- XAUUSD Global analysis of COMEX gold futures (GC) daily bars, 4 January 2010 to 3 November 2026, the data behind our gold volatility tool; sessions without full high/low data were excluded. Rules fixed in advance; no costs included
- J. Welles Wilder Jr. (1978), New Concepts in Technical Trading Systems, Trend Research (true range, ATR, DMI and ADX)
- Mandelbrot, B. (1963), “The Variation of Certain Speculative Prices”, Journal of Business 36(4), 394–419, doi:10.1086/294632
- The sizing example is hypothetical
Educational content only, not financial advice. Past data does not predict future prices. Trading gold with leverage carries a high risk of loss.
Frequently asked questions
What is market regime detection?
Market regime detection means classifying the current market environment with fixed rules, for example trending or ranging, and quiet or volatile, so you can adapt position size, stops and the type of setup you trade. It describes the backdrop; it does not predict the next move.
How do I know if gold is trending or ranging?
A simple rule is ADX(14) on the daily chart: 25 or above is usually read as trending, below 20 as ranging, and in between as unclear. Check it against the chart itself: higher highs and higher lows, or overlapping swings. In our data since 2010, gold was in the trending band on about 41% of days and the ranging band on about 38%.
How do I measure gold's volatility regime?
Compare today’s ATR(14), as a percentage of price, with its own range over the last year. In our rules, below the 25th percentile is a low-volatility regime and above the 75th is high. On 3 November 2026 gold’s ATR was about 2.1% of price, which was only around the 18th percentile of the previous year because that year had been unusually volatile.
Do volatility regimes last?
They tend to persist. In COMEX gold data from 2010, on days in each of our three volatility regimes, gold was in the same regime 20 sessions later about 58% to 60% of the time, well above what chance alone would give for the low and high bands. Volatility clustering of this kind was described by Benoit Mandelbrot in 1963.
Comments
No comments yet.