CRT Trading: Candle Range Theory and Turtle Soup, With Rules and Data
CRT is a sweep of one candle's range that closes back inside. Here's how to mark it on futures, which systems have written rules, and why the data says to be careful.

CRT (Candle Range Theory) is a sweep-and-return pattern. A reference candle sets a high and a low. The next candle takes out one of those extremes, leaves the other untouched and closes back inside. From there you target the range's 50% level (the midpoint), then the opposite extreme.
At its core, it's Connors and Raschke's Turtle Soup (1995) applied to the range of one specific candle, almost always a 4-hour candle. Futures have one detail that changes everything: CME 4h candles don't open at the same times as the ones the forex crowd uses.
In this guide you'll learn how to mark a CRT step by step, which systems have written rules, the most common mistakes, what the few existing studies say and how to adapt it to a prop firm account. One caveat: there's no accessible primary material from the method's creator, so what follows is the most widely cited version, not the canonical one.
Free tool: if you use NinjaTrader 8, you can download the CRT indicator we built at El Trader Financiado for the community. See the indicator and download it.
What Is CRT (Candle Range Theory)?
The descriptions of public scripts attribute the method to @Romeotpt, on X. The original name they cite is "Candles Are Ranges Theory", and the author calls the reversal that follows the sweep "turtle soup".
In the most widespread version of the method, the definition is this:
- Candle 1 (parent or reference): sets the range with its high and its low.
- Candle 2: sweeps one side without sweeping the other and closes inside the range of candle 1.
- Bearish CRT: candle 2 takes out the high and closes inside. The bullish one is the mirror image.
- Targets: first the range's 50% level; then the opposite extreme.
It's usually read as a sequence of accumulation, manipulation and distribution. Candle 1 is the range (accumulation), the wick that pokes outside is the false move (manipulation) and the run to the other extreme is the real move (distribution). This reading comes from the method's author and the people who popularized it. It's not a measured fact.
CRT and Turtle Soup: are they the same?
They share the idea, but not the rules. The original Turtle Soup, published in Street Smarts (1995), is a mechanical, daily, counter-trend rule on 20-day extremes. The ICT version and the CRT version apply to "any previous high or low" and usually ask for structure confirmation.
Taking the 1995 rule to 4h candles or 5-minute charts is an extrapolation. There's no published evidence that it works the same way on those timeframes.
The discounts our readers use most
See all discountsLucid Trading
Active discount
TakeProfit Trader
Active discount
Apex Trader Funding
Until Nov 25
How to Mark a CRT, Step by Step
1. Pick the right 4-hour candle
CRT communities use the 4h candles that open at 1 AM, 5 AM and 9 AM ET. That's the forex grid.
On CME futures the session opens at 6:00 PM ET, so the native 240-minute candle is offset. On CME index futures, 4h candles open at 2 AM, 6 AM and 10 AM ET.
Many sources use one or the other without saying which. Decide which one you're using and don't mix them.
2. Set the bias on the higher timeframe
The most common timeframe pairs are these:
| CRT timeframe | Entry timeframe |
|---|---|
| Weekly | Daily |
| Daily | 4h |
| 4h | 15 minutes (or 5) |
3. Require a key level
The method's own sources point to the lack of a key level as the main cause of failure: an isolated CRT isn't enough. The accepted levels are previous highs and lows, order blocks and fair value gaps.
4. Check the sweep
Candle 2 must sweep one side only and close inside. Rule out two cases:
- Inside candle: it sweeps nothing. It isn't a CRT.
- Engulfing candle that sweeps both sides: scripts exclude it or treat it separately as a "double purge".
5. Look for the trigger on the lower timeframe
CRT scripts list three entry models:
- Model #1: a full candle closes beyond the previous one and you enter at that close. It's the fast version.
- CISD: the open level of the last directional leg breaks and you enter on the break.
- MSS + FVG: a change of structure with displacement and an entry in the fair value gap it leaves.
6. Define stop, targets and invalidation
The stop goes beyond the swept extreme or, in the tighter version, at the swing that creates the MSS. Target 1 is the range's 50% level and target 2 is the opposite extreme. If price takes out the extreme of the sweep, the CRT is invalidated.
CRT and Turtle Soup Systems With Rules
These are the systems that have written rules in the sources. None of them is "the official method".
| System | Entry | Stop | Target | Time filter | Management |
|---|---|---|---|---|---|
| Turtle Soup (Connors and Raschke, 1995) | New 20-day low with the previous 20-day low at 4 or more sessions back; buy stop 5-10 ticks above the previous low | One tick below today's low | No fixed target | Daily chart | Trailing stop as it gains |
| Turtle Soup Plus One | Day 1 closes at or below the 20-day low; on day 2, buy stop at that low | Below the extreme | No fixed target | Daily chart | Partials at 2-6 bars and trailing; canceled if not filled on day 2 |
| Reference CRT (synthesis) | 4h candle (2-6-10 on CME) swept and closed inside, at a key level, with MSS or FVG on 5-15 min | Swept extreme or MSS swing | Range midpoint (50%) and opposite extreme | Killzone | Invalidated if price takes out the swept extreme |
| Mechanical CRT on 1h (published backtest) | After candle 2 closes, stop order one tick beyond its extreme | Sized for 2:1 | Opposite extreme of candle 1 | 9:30 AM–1:00 PM ET | Maximum 3 entries a day; the setup expires after 3 h |
The Turtle Soup rules are those from Street Smarts; the mechanical 1h version comes from a backtest published by an automation tool vendor, unaudited. Sells are the mirror image of buys.
There are discrepancies between sources: the base version asks for the previous extreme to be 4 sessions old or more and the Plus One talks about 3 days. The logic of Plus One is that many people enter on the breakout day at the close, and that traps more traders. The "reference CRT" is a synthesis of what scripts repeat, not the rule of any specific author.
A warning about the widely shared "4H CRT 1AM/5AM" strategy: it marks the range of the 1 AM candle and waits for a 15-minute MSS if the 5 AM candle sweeps. But it doesn't define a stop, a target or a time zone. Without those three things, it's not a system.
Common CRT Mistakes
- Mixing up 1-5-9 and 2-6-10. On NQ, ES, CL or GC, the CME 4h candle doesn't match the forex one. If you mix the two, you mark different ranges.
- Trading without a key level. It's the failure the method's own sources point to most.
- Taking an engulfing candle for a CRT. If the candle sweeps both sides, there is no clear "false" side.
- Choosing a candle 1 with a big body. In the only large study (published on TradingView, no external review), a "conviction" candle 1 lowers the hit rate by 0.6 to 4.2 percentage points, and a doji-type candle 1 raises it by 0.9 to 2.7.
- Entering before candle 2 closes. Until it closes, the sweep can turn into a breakout. A CRT that hasn't closed isn't a CRT.
- Placing the stop at the 4h extreme without calculating it in dollars. On NQ it can be a number that doesn't fit your account (more on that below).
What to Combine CRT With
CRT asks for a level and a trigger. Those pieces have their own guides:
- Levels and times: the previous day's high and low, the Asia range (8:00 PM–midnight ET) and the killzones. You have it in ICT killzones and in the session levels indicator.
- Trigger: MSS and order blocks are explained in order block and market structure. The entry gap, in fair value gap.
- Cross-asset confirmation: a 4h candle that sweeps on NQ but not on ES is the cross-asset version of the same argument. It's an SMT divergence.
A counterintuitive data point: in that same study, adding multi-timeframe confluence lowered the win rate to 28-32.5%, even though it reduced risk. The best-performing filter was a 9 EMA in the direction of the trend (+2.9 to +6.6 percentage points), at the cost of discarding 60-65% of trades. "More confluences, better" didn't hold in the only large test that exists.
What the Evidence Says About CRT
There isn't much, and it isn't favorable.
- CRT without filters: a study published on TradingView, without external review (gold 2009-2019 and EUR/USD 2000-2021, from 1 minute to daily) measures how often price reaches the opposite extreme before the stop. It comes out between 42.8% and 49.0%. The author describes it as "very similar to flipping a coin". It doesn't analyze the reward/risk ratio or costs.
- Original Turtle Soup: tested on 42 US futures between 1980 and 2011, with and without costs, it gets a grade of "D" (poor) (Oxford Capital Strategies). Some of the people who spread it acknowledge that the edge has degraded and depends on the market regime.
An honest caveat: since the sweep wick is usually shorter than the range, the reward/risk ratio tends to be greater than 1. That's why a "coin-flip" win rate doesn't by itself imply zero expectancy. But nobody has measured it net of costs with a large sample on futures. What the people spreading the method claim about its effectiveness isn't proven.
How CRT Fits a Prop Firm Account
On a funded account, the problem with CRT isn't the pattern. It's the size of the stop.
Calculate the stop in dollars before entering. With the stop at the swept extreme of a 4h candle, the distance can be large. An example with a 60-point stop on NQ and a 15-point stop on ES:
| Contract | Point value | Stop risk |
|---|---|---|
| NQ | $20 | 60 points = $1,200 |
| MNQ | $2 | 60 points = $120 |
| ES | $50 | 15 points = $750 |
| MES | $5 | 15 points = $75 |
With a daily loss limit of, say, $1,000, a single NQ stop no longer fits, and an ES stop eats 75% of the day. In a backtest published by an automation tool vendor (unaudited), even with 1h candles and one contract, the maximum drawdown on NQ over two months was $5,722.50. That exceeds the typical limit of a 50K account.
Duration and news. In the study published on TradingView, 86-92% of setups resolve within three candles. On a 4h chart, that's up to about 12 hours of trading: it can run through the 8:30 AM ET data releases or the daily close. Many prop firms restrict trading around news such as CPI or NFP; check the rules of yours.
Consistency. A system with a win rate near 50% and a distant target concentrates profit in a few good days. That clashes with consistency rules that limit the weight of the best day.
A sensible adaptation (this is our own judgment, not from any source):
- Use the 4h candle only as bias.
- Put the stop at the 5-15 minute MSS swing, not at the 4h extreme.
- Trade micros (MNQ, MES) until the stop fits with room to spare.
- Avoid the 6–10 AM ET candle, which contains the 8:30 AM release.
- Be flat before your firm's daily close (between 3:10 and 3:59 PM CT, depending on the firm).
To compare daily loss limits across firms, use our prop firm comparison tool.
Free NinjaTrader Indicator
We built a CRT indicator for NinjaTrader 8 that draws the 4h candles aligned to CME (2-6-10) or to 1-5-9, marks each sweep with its 50% and tracks whether the target is hit or invalidated. It's free, built by El Trader Financiado for the community, and available in English and Spanish. Download it and see its settings on the CRT indicator for NinjaTrader page.
CRT Trading FAQ
What is CRT in trading?
It's Candle Range Theory: a reference candle sets a range, the next one sweeps one extreme and closes back inside, and the targets are the range's 50% level and then the opposite extreme. It's attributed to @Romeotpt.
Which 4-hour candles should I use for CRT?
In forex, traders use the 1 AM, 5 AM and 9 AM ET candles. On CME futures, the native 4h candles open at 2 AM, 6 AM and 10 AM ET. Pick one grid and stick to it.
Is CRT the same as Turtle Soup?
Not exactly. The 1995 Turtle Soup is a mechanical rule on 20-day extremes on a daily chart. CRT applies the same sweep-and-reversal idea to the range of one specific candle and adds lower-timeframe confirmation.
Does the CRT strategy work?
There's no proof that it works on its own. Without filters, the pattern reaches the opposite extreme before the stop in 42.8-49.0% of cases in the largest study, and the original Turtle Soup gets a "D" on 42 futures.
Which timeframe do you enter on?
The usual setup is CRT on 4h and entry on 5 or 15 minutes, with an MSS, a CISD or an FVG as the trigger.
What is a double purge?
A candle that sweeps both the high and the low of the previous one and still closes inside. Most scripts don't count it as a CRT.
