Studies

Study series · 24 testsNot this time · all 24

CRT (Candle Range Theory) across four timeframes, NAS100 and gold

One rule, written down before any test, run on 4 candle sizes, 2 markets and 3 reward-to-risk floors: 24 tests. 12 of 24 averaged above zero out of sample, none graded promising. Every test is shown below with equal weight.

Tests
24
Markets
NAS100, Gold
Candles
Daily, 4H, 1H, 15m
Data
Oct 2021 – Oct 2026
Out of sample
May 2023 – Oct 2026
Published
26 Oct 2026
The rule

What was tested

Short
C1 highsweep beyond C1 highC1 low untouched: one side onlycloses back inside ✓stop · C2 hightarget · C1 lowenter shortnext 5-minute openend of C3: out ifneither was hitriskrewardreward ÷ risk = 1.29at least 1 (main rule): take itC1the rangeC2the sweepC3the tradeillustration, not data
Long
C1 lowsweep beyond C1 lowC1 high untouched: one side onlycloses back inside ✓stop · C2 lowtarget · C1 highenter longnext 5-minute openend of C3: out ifneither was hitriskrewardreward ÷ risk = 1.29at least 1 (main rule): take itC1the rangeC2the sweepC3the tradeillustration, not data
  1. 1
    C1 sets the range

    One candle on the New York clock, built from 5-minute bars. Its high and its low are the range.

  2. 2
    C2 sweeps exactly one side…

    The next candle trades beyond C1’s high (for a short) or its low (for a long), strictly; one tick is enough. It must not go beyond the other side. Both sides swept: no trade.

  3. 3
    …and closes back inside the range

    C2 closes below C1’s high and above C1’s low. A close outside the range, or exactly on its edge: no trade.

  4. 4
    Entry at the next 5-minute open

    A market order at the open of the first 5-minute bar after C2 closes. The setup is only known once C2 has closed.

  5. 5
    Stop at C2’s wick, target at C1’s other side

    Stop at C2’s extreme, no padding. Target at C1’s opposite extreme. If one 5-minute bar touches both, the stop counts.

  6. 6
    Out at the end of C3 if neither is hit

    C3 is the candle after C2, the one the trade is entered in. The trade closes at the close of its last 5-minute bar.

  7. 7
    Skip it if reward to risk at C2’s close is below the floor

    Distance from C2’s close to the target, divided by the distance to the stop. The main rule needs 1:1; two trials need 1:2 and 1:4. The target is never moved.

Which candles count

All New York time. One position at a time. Every test trades on 5-minute bars.

CandleCandles start atPairs tradedOut at the end of C3Most a day
Daily17:00 to 17:00 (Sunday evening is Monday’s candle)Every pairNext 17:001 (follows from the rules)
4H17:00, 21:00, 01:00, 05:00, 09:00, 13:00Key times: C2 is the 05:00 or the 09:00 candle13:00 or 17:002
1HOn the hourC1 and C2 inside 01:00 to 13:001 hour after entry2
15m:00, :15, :30, :45C1 and C2 inside 01:00 to 13:0015 minutes after entry2

CRT (Candle Range Theory) is popularised by Romeo (@Romeotpt on X). Its sweep-and-reclaim idea builds on “Turtle Soup”, from Linda Raschke and Larry Connors’ book Street Smarts (1995). The rules here are IDFX’s stated interpretation, written down before any test. They are not Romeo’s own rules.

Out-of-sample results

The main rule, 1:1

Before trading costs Out-of-sample

24 tests were run (4 timeframes x 2 markets x 3 reward-to-risk floors). With that many, one or two can look good by chance alone even if none has an edge. All are listed with equal weight; read any positive result as '1 of 24'.

NAS100
Gold
Daily
Daily · NAS100Not this time−0.210Rper trade−36.1Rtotal172trades 0 of 5 windows upshuffled price as good: 35 of 40This test Daily · GoldNot this time−0.134Rper trade−22.9Rtotal171trades 2 of 5 windows upshuffled price as good: 27 of 40This test
4H
4H · NAS100Not this time+0.185Rper trade+39.9Rtotal216trades 4 of 5 windows upshuffled price as good: 4 of 40This test 4H · GoldNot this time+0.065Rper trade+18.8Rtotal291trades 4 of 5 windows upshuffled price as good: 27 of 40This test
1H
1H · NAS100Not this time−0.022Rper trade−31.0Rtotal1,419trades 2 of 5 windows upshuffled price as good: 29 of 40This test 1H · GoldNot this time+0.035Rper trade+51.8Rtotal1,466trades 4 of 5 windows upshuffled price as good: 10 of 40This test
15m
15m · NAS100Not this time−0.012Rper trade−21.9Rtotal1,788trades 2 of 5 windows upshuffled price as good: 28 of 40This test 15m · GoldNot this time+0.007Rper trade+13.3Rtotal1,785trades 4 of 5 windows upshuffled price as good: 4 of 40This test

Each curve is the test windows stitched end to end, in R, on the same time axis. "Windows up": test windows with a positive average. "Shuffled price as good": how many of the engine's reshuffled price series did as well or better; the fewest for any of the 24 tests was 4.

Every test

All 24, with equal weight

Before trading costs Out-of-sample
TestFloorTradesWin ratePer tradeTotalProfit factorWindows upShuffled as goodVerdict
Daily NAS1001:1 (main)17233.7%−0.210R−36.1R0.640 of 535 of 40Not this time
Daily Gold1:1 (main)17135.1%−0.134R−22.9R0.752 of 527 of 40Not this time
4H NAS1001:1 (main)21642.1%+0.185R+39.9R1.314 of 54 of 40Not this time
4H Gold1:1 (main)29141.6%+0.065R+18.8R1.104 of 527 of 40Not this time
1H NAS1001:1 (main)1,41939.5%−0.022R−31.0R0.962 of 529 of 40Not this time
1H Gold1:1 (main)1,46640.5%+0.035R+51.8R1.054 of 510 of 40Not this time
15m NAS1001:1 (main)1,78837.8%−0.012R−21.9R0.972 of 528 of 40Not this time
15m Gold1:1 (main)1,78538.2%+0.007R+13.3R0.994 of 54 of 40Not this time
Daily NAS1001:28524.7%−0.347R−29.5R0.500 of 532 of 40Not this time
Daily Gold1:28732.2%−0.039R−3.4R0.922 of 515 of 40Not this time
4H NAS1001:210830.6%+0.103R+11.1R1.133 of 59 of 40Not this time
4H Gold1:214438.9%+0.171R+24.7R1.274 of 515 of 40Not this time
1H NAS1001:287633.9%−0.037R−32.2R0.942 of 538 of 40Not this time
1H Gold1:297534.7%+0.019R+18.6R1.012 of 518 of 40Not this time
15m NAS1001:21,75433.2%+0.012R+20.6R1.014 of 520 of 40Not this time
15m Gold1:21,74833.0%+0.050R+86.8R1.063 of 55 of 40Not this time
Daily NAS1001:43212.5%−0.545R−17.4R0.331 of 523 of 40Not this time
Daily Gold1:42729.6%+0.408R+11.0R1.483 of 58 of 40Not this time
4H NAS1001:42516.0%−0.317R−7.9R0.612 of 525 of 40Not this time
4H Gold1:44233.3%+0.215R+9.0R1.303 of 519 of 40Not this time
1H NAS1001:434324.8%−0.145R−49.7R0.801 of 535 of 40Not this time
1H Gold1:435024.0%−0.068R−23.7R0.882 of 524 of 40Not this time
15m NAS1001:41,20322.2%−0.100R−119.9R0.881 of 532 of 40Not this time
15m Gold1:41,18522.5%+0.053R+62.6R1.042 of 513 of 40Not this time

The 1:2 and 1:4 rows are trials around the main rule: the same rule, skipping setups that offer less than that reward to risk. Every row is out of sample and before trading costs.

Real trades

The rule on real candles

NAS100 · 2026-08-11 · 09:00 · short · target hit, +2.12R
21:0001:0005:0009:0013:00C1C2C3sweep: 32.42 beyond C1 high5-minute barsC2 high · stop29,773.41C1 high29,740.99C2 close29,719.29C1 low29,605.16exit29,605.16
NAS100 · 2026-10-22 · 09:00 · long · stop hit, −1.00R
21:0001:0005:0009:0013:00C1C2C3sweep: 1.63 beyond C1 low5-minute barsC1 high29,453.54C2 close29,323.59C1 low29,245.59C2 low · stop29,243.96exit29,243.96

Real out-of-sample trades from the 4H NAS100 test (engine run run_7b0882a7cc), picked to show the rules and not the result: the most recent target and stop with a clean entry. Of its 216 trades, 76 hit the target, 116 the stop and 24 closed at the end of C3.

Limitations

  • Before trading costs. Spread, commission and slippage would push every average down, not up.
  • 24 tests. Any single good-looking row can be luck; read each one against all 24.
  • Out of sample is May 2023 – Oct 2026, not the whole data period.
  • One reading of CRT. Other candle times, entries, stops or targets are different strategies and were not tested.
  • One broker’s price feed. Another feed’s candles differ slightly.
  • 5-minute bars: a bar that touches both the stop and the target counts as a loss.
  • A backtest. Not live trading, and not a funded account.

The data

All 24 tests in one file, published as it was written, with the engine run behind each row. Each main-rule test also has its own page and file.

/studies/crt.json