Candle Range Theory (CRT) on Gold: A Precise Definition and a Mechanical Test

The short answer

Candle Range Theory (CRT) says that when a candle trades beyond the high or low of the previous candle and then closes back inside its range, price is likely to travel to the opposite end of that range next. We wrote it as exact rules and tested it on gold:

• 4-hour candles, two years (October 2024 to October 2026): 818 trades, 41.1% winners, −$1,817 per ounce after costs, profit factor 0.69.
• Daily candles since 2015: 744 trades, 43.7% winners, −$449 per ounce, profit factor 0.91.

Every version we tried lost, including different targets and holding periods. That does not mean discretionary CRT traders lose; it means the candle pattern on its own did not carry an edge on gold in these samples.

CRT, defined precisely

CRT uses three consecutive candles on the same timeframe:

• Bearish CRT: candle 2’s high is above candle 1’s high, and candle 2 closes below candle 1’s high (but above its low).
• Bullish CRT: candle 2’s low is below candle 1’s low, and candle 2 closes above candle 1’s low (but below its high).
• If candle 2 sweeps both ends (an outside bar), there is no clear signal; we skipped those.

The logic is the same as a classic liquidity sweep: stops beyond the obvious high are triggered, the move fails, and the market rotates the other way. CRT simply defines “obvious high” as the previous candle’s extreme.

Exact test rules

How we tested it

The results

4-hour CRT (two years)

Daily CRT (since 2015)

Where the main 4-hour trades ended

What the results say, in plain English

• The sweep-and-close-back pattern did not predict the next candle. On H4, the stop beyond the sweep was hit almost twice as often as the far end of the range.
• The midpoint target made it worse, not better. Even a target halfway across the range lost more, because the stop was large relative to the shorter target.
• Holding longer helped a little. Allowing three candles cut the loss by about 40%, but still lost in both years.
• Daily CRT was nearly random. It made money from 2015 to 2020 (profit factor 1.15) and lost after 2021. Over eleven years it was close to a coin flip minus costs.
• The problem is not costs. At zero cost the main H4 version still lost $1,572 per ounce.

Why CRT traders say this misses the point

Most CRT educators do not trade every candle 2 that closes back inside. They add filters such as:

1. Higher-timeframe direction. Only take bearish CRT setups when the daily or weekly trend is down. See our multi-timeframe analysis lesson.
2. Key levels. The sweep must take a meaningful high or low, such as the previous day’s high or a weekly level, not just any candle.
3. Time. Setups that form around the London or New York open, similar to the Power of 3 framework.
4. Lower-timeframe confirmation. A market structure shift on M15 or M5 after the sweep, rather than entering at the next candle’s open.

Each of these could change the result. They also add judgement and hindsight risk, which is why it is important to journal setups before they play out. Our tests of the previous day high and low and the Judas swing cover two of those filters mechanically.

When CRT fails

• Trending markets. In a strong trend, candle 2 sweeps the high, closes back inside, and candle 3 simply continues up.
• News candles. A data release inside candle 2 can create a long wick that means nothing about the next four hours.
• Different candle alignment. A 4-hour candle starting at 00:00 GMT gives a different range from one starting at 18:00 New York time.
• Large sweeps. When candle 2’s wick is long, the stop is far and the risk per trade is high.

Risk sizing

The average stop on the 4-hour version was about $18.26 per ounce. Risking $100:

• $100 / ($18.26 + $0.30) is about 5.4 ounces (0.05 lots).
• The worst losing run was 12 trades, about $1,200 at that size.

Calculate size for each setup with the position size calculator.

Checklist before you trade a CRT set-up

• Fix your candle timeframe and alignment. New York-aligned 4-hour candles, broker candles or daily candles: choose one and do not switch.
• Write the higher-timeframe bias first, before candle 2 closes. A CRT set-up against that bias is, by your own rules, not a trade.
• Check what candle 2 swept. Was it just the previous candle’s high, or a meaningful level such as the previous day’s high? Record it.
• Measure the stop. If candle 2’s wick is unusually long, the risk may not justify the trade.
• Avoid news candles. Mark candles that contain CPI, payrolls or a Fed decision and judge them separately in your journal.
• Score every set-up, not just the clean ones. The mechanical version lost largely because of the many set-ups that look valid and then fail.

Test it yourself

Pick one timeframe and one candle alignment, write down your CRT filters before you look at the chart, then mark every candle 2 for three months and record what candle 3 did. Compare your filtered results with the unfiltered numbers above. If your filters turn a 0.69 profit factor into something well above 1.0 over a large sample, you have found real context. The strategies tested hub lists our other mechanical tests.

Practise it with other traders

CRT is easy to see in hindsight and hard to apply live. In the XAUUSD Global community, traders post their CRT setups before the third candle closes, so everyone can judge them honestly. The first month is free, then $19.99 a month.

Educational content only, not financial advice. Trading gold with leverage carries a high risk of loss.

Frequently asked questions