Candle Range Theory (CRT)

Candle Range Theory (CRT) reads a single closed candle as a range rather than as a directional signal. Its high and low become the boundaries of a container, and what follows is interpreted as the market working that container: first probing one boundary to reach the orders resting beyond it, then delivering price toward the other. CRT comes out of the ICT and Smart Money Concepts tradition and borrows its vocabulary, but its central claim is narrow enough to state in one line — ranges tend to be swept before they are expanded.

## How the model works

The canonical form is a three-candle sequence. The first candle establishes the range. The second trades outside one of its extremes, usually with a wick, and closes back inside; this is what practitioners call the manipulation, and the close back inside is what separates it from a genuine break. The third is expected to travel toward the opposite extreme of the first, which is where the objective normally sits. Read this way, CRT is the accumulation–manipulation–distribution idea applied to a fixed, objective container instead of to a subjectively drawn zone.

The second element is fractality. A closed higher-timeframe candle is, on a lower timeframe, a finished consolidation with internal structure. CRT is therefore worked in timeframe pairs: the range is defined above (weekly, daily, 4-hour, 1-hour or a session), and the entry is located below once the sweep has occurred — a market structure shift, a change in how price is being delivered, a fair value gap left by the reversal, or a retracement into the midpoint of the manipulation wick. The higher timeframe supplies levels and direction; the lower one supplies the trigger and the stop.

Time is the third element and the most frequently ignored. A candle only exists relative to a clock, so the model depends on where the session opens: a daily range anchored to a 17:00 New York close is not the same object as one anchored to 00:00 UTC. Many versions anchor ranges to specific opens — the daily open, the London or New York session, the Asian range — and only look for the sweep inside a defined window.

## Main variants

Across the versions catalogued here, a few families recur. Pure three-candle CRT takes any qualifying sequence in either direction. Trend-filtered or tendential CRT keeps only the sweeps aligned with higher-timeframe structure or with an assumed draw on liquidity. Confirmation-based CRT does not act on the sweep itself and waits for a lower-timeframe event, such as a change in state of delivery or a displacement leg. Confluence-stacked CRT adds order blocks, fair value gaps, breakers, or premium and discount positioning inside the range. Session and scalping variants shrink the whole construction, using a 15-minute or 5-minute candle as the range with entries on the 1-minute chart. Some versions are indicator-assisted, with ranges and triggers plotted automatically; others stay explicitly discretionary.

## What typically differentiates implementations

Two versions can share the name and behave very differently. The differences that matter most: which candle is allowed to be the range candle and how long it stays valid; what counts as a valid sweep (wick only, a close beyond, a minimum penetration, a return within the same candle); the timeframe ratio between range and entry; whether entry is at market, on a limit at a level, or only after confirmation; where the stop sits (beyond the sweep wick, beyond the range extreme, or structural); and what the target is (opposite extreme, range midpoint, fixed multiple, or trailing exit). Bias filters and trading windows vary just as much.

## Common mistakes

The most frequent error is retrospective selection: on a finished chart almost every move contains a candle whose extreme was exceeded and then reclaimed, so the pattern looks ubiquitous until it has to be defined in advance. Closely related is dropping the close-back-inside condition, which turns a real breakout into an apparent manipulation. Others: pairing timeframes badly, so the entry chart is either too coarse to offer structure or too fine to be tradeable; placing stops immediately beyond a wick the market has already shown it can reach; assuming the opposite extreme is always delivered and managing nothing along the way; stacking confluences that are not independent, which mainly shrinks the sample; and ignoring the data feed timezone, which silently redefines every range.

## How to evaluate and backtest a version

Start by forcing the rules into writing — range definition, sweep validity, trigger, invalidation, target and time window — and mark whatever cannot be written down as discretionary instead of pretending it is mechanical. Fix the session offset, then re-test with a different one; a version that survives only at one daily open is anchored to an artifact. Because most CRT variants are intraday with tight stops, costs are not a rounding error: include spread, commission and slippage, and when a bar touches both stop and target assume the stop first unless you have tick data. Check sensitivity across sweep threshold, entry timeframe and target, and compare each added confluence against the simpler version without it. Test out of sample, across instruments and across regimes, inspect the distribution of results by session and day for concentration in a few events, and treat a small number of trades as a small number of trades however the equity curve looks.

Strategies in this concept (51)

Frequently asked questions

How is CRT different from a generic false breakout or liquidity sweep?

The mechanics overlap, but CRT is stricter about where the range comes from. Instead of a zone drawn by hand, the container is a single closed candle on a defined timeframe, anchored to a defined session clock, with the opposite extreme of that same candle acting as the default objective. That makes the setup fully specifiable in advance, which is exactly what a hand-drawn zone is not.

Is CRT just a rebrand of ICT's accumulation–manipulation–distribution?

It is the same underlying story with a fixed container. AMD or Power of Three describes a phase sequence without saying which price levels bound it; CRT pins those bounds to one candle's high and low. In practice most CRT material assumes the ICT toolkit alongside it — order blocks, fair value gaps, structure shifts — so the two are usually taught together rather than as alternatives.

Which timeframe should the range candle be on?

There is no canonical answer, and the choice is one of the main things that separates versions. What matters is the ratio between the range timeframe and the entry timeframe: too close and the lower chart shows no usable structure inside the sweep, too far and the trigger drowns in noise. Whichever pair you pick, test at least one neighbouring pair before concluding the model works.

Do I need an indicator to trade CRT?

No. The pattern is readable manually, and many versions are taught that way. Indicators mark ranges, sweeps and triggers automatically, which mostly buys consistency and speed rather than new information. If you use one, confirm it evaluates only closed candles — a tool that flags a range or a sweep before the candle closes will make historical charts look far cleaner than live trading is.

Why do two CRT versions produce different results on the same chart?

Because the name covers a family, not a single rule set. Sweep validity, range expiry, entry trigger, stop placement, target and session window are all free parameters, and each of them changes which sequences qualify and what happens after entry. Before comparing two versions, write both out rule by rule; most apparent disagreements turn out to be different definitions rather than different outcomes.

What should I check first when backtesting a CRT version?

The clock and the costs. Re-run the same rules with a different daily-open or session offset — if the result collapses, it was anchored to a timestamp rather than to behaviour. Then add spread, commission and slippage, and resolve bars that touch stop and target on the same candle in the pessimistic direction. Those two checks eliminate a large share of CRT backtests before any parameter tuning starts.

All strategy concepts · Explore all strategies