ICT Propulsion Block Explained: Entries and Mean Threshold

You mark a bullish order block, watch price leave it, and wait for the original zone to return. Instead, the next pullback stops higher and price moves away again. Chasing that movement creates a poor entry, while placing a limit order at the oldest low may leave you waiting for a retracement that never arrives. ICT propulsion blocks provide a framework for studying this handover between an older reference and a newer candle.

The difficult part is identifying that relationship before the next rally or decline makes it obvious. A large candle moving away from a zone is not, by itself, the propulsion block. The candle you need to examine is the one that interacted with a previously established order block. Its direction, location, completed body, and subsequent departure all matter.

This guide develops a practical study protocol around that relationship. It explains the measurements, separates a formation from an executable order, and works through both directions using hypothetical EUR/USD prices. For traders in Ghana, it also connects the setup to GMT session planning and the execution costs that can dominate a very small stop.

1. What is an ICT propulsion block?

In ICT’s original Month 4 propulsion block lesson, the concept extends order block theory through a relationship between two formations. A new down candle reaches into an older bullish order block and becomes a higher reference for subsequent buying. An up candle reaching into an older bearish order block provides the bearish counterpart. The lesson emphasises responsiveness on a later return and the importance of the newer candle’s body midpoint.

For clarity, this article calls the older formation the parent order block and the newer interacting candle the propulsion candle. These are labels for keeping your chart organised. They do not mean that the two candles must belong to different timeframes. Both can occur on the same five-minute chart. If you use a higher timeframe parent, record that timeframe and wait for its relevant candle to finish before treating its measurements as fixed.

Think of the chart as a sequence of decisions. First, a previous price area becomes relevant through departure and surrounding structure. Second, a later candle revisits that area. Third, a new departure provides information about whether the later interaction is worth studying. Only after those observations exist can you specify an order for a future return. This sequence keeps later knowledge out of the original decision.

What the institutional interpretation adds

Within ICT and SMC thinking, the hypothesis is that an earlier area of buying or selling continues to influence price delivery, while a newer interaction provides a closer reference for participation. That interpretation directs attention to context, displacement, liquidity objectives, and the quality of the response. It does not turn a candle into a record of a bank’s inventory.

An ordinary retail forex chart cannot identify which institution traded, its intention, or how many unfilled orders remain at a price. Write observations separately from interpretations: “the completed candle reached the parent interval” is observable; “a bank defended its remaining position” is an explanation you cannot verify from that candle alone. This distinction makes your trading journal more useful because you can test the observation even when the narrative sounds convincing.

How this guide uses the idea

The timed rules, cost allowances, confirmation requirement, and trade management below form an illustrative research model created for this guide. They are not presented as a complete mechanical system prescribed by ICT. Their purpose is to make decisions reproducible. You should be able to hand the rules and a chart to another learner and obtain broadly the same candidate list, including the setups that fail.

2. Why the relationship matters more than the candle shape

A red candle has little meaning in isolation. During one sequence it may be an ordinary pause in a decline. During another it may retrace into a previously identified bullish reference. The distinction lies in what happened before the candle formed and where its range traded. If you remove the parent from the chart and your explanation remains unchanged, you may be describing a generic pullback rather than a propulsion relationship.

This matters for risk placement. Trading from the original parent can leave a large distance between entry and invalidation after price has already advanced. Using a newer reference can reduce that distance, but only by making a stronger assumption about how little adverse movement the continuation should need. You are exchanging tolerance for precision. A smaller stop does not automatically improve expectancy, because it can also increase the number of otherwise reasonable moves that stop you out first.

The parent also helps distinguish continuation from an attempted reversal. Suppose a bullish candidate forms directly beneath a clear higher timeframe bearish area while nearby buy-side liquidity has already been taken. The newer candle may still fit the geometric description, but the intended destination is less clear. The appropriate response is to downgrade or exclude the candidate under your written rules, not add more labels until the trade appears convincing.

A fair value gap may accompany the departure. It measures a different relationship between three candles. Record it as additional context if your model uses it, but do not rename the gap a propulsion block or assume that two labels create two independent pieces of evidence. They may describe different features of the same movement.

3. How to identify a propulsion block on a chart

Start with a clean chart and an explicit directional hypothesis. For this guide’s examples, the parent is an already completed opposite-colour candle followed by a departure through a previously marked internal swing. We draw the parent’s entire high-to-low range so the contact test is unambiguous. Other order block drawing conventions exist, but switching conventions after seeing a result destroys the value of your comparison.

1. Fix the parent first. Record its timestamp, timeframe, open, high, low, and close. Also record the swing and completed departure used to qualify it. A box added after a profitable continuation is an illustration, not evidence of a decision made in advance.
2. Require actual contact. For a bullish candidate, a later down candle must trade into the parent interval. For a bearish candidate, a later up candle must do so. The overlap can occur through a wick. A candle that stops short by a visible distance has not met this particular contact rule.
3. Wait for the newer candle to close. Until then, its final colour, body midpoint, and range can change. Save the completed values once it qualifies. Avoid drawing a midpoint from an unfinished candle and later judging the trade against a different midpoint.
4. Observe departure. Our study model requires a subsequent completed candle to close beyond the propulsion candle’s high for a bullish setup, or below its low for a bearish setup. This is a deliberately stated confirmation filter, not a universal definition of every ICT entry.
5. Plan a later return. Once confirmation exists, specify entry, stop, target, costs, and expiry. If price has already touched the proposed entry before the order could have been placed, do not award yourself that earlier fill.
6. Freeze the references. Save an image or written record before advancing the replay. Keep the parent and propulsion candle unchanged, even if a different candle would make the eventual result look better.

Use one observation scale before combining timeframes

A fifteen-minute candle can contain several five-minute candles with different colours. A contact visible inside it may disappear when you treat only the fifteen-minute body as the reference. Neither representation is automatically wrong, but they describe different rules. Begin with one execution timeframe and one parent convention. Add a second timeframe only when you can explain what information it supplies and when that information became available.

For a continuation model, an internal structure break can help describe departure. Avoid treating every tiny wick break as equivalent to a completed close through a swing you had already selected. A useful journal entry states the exact broken level and close. “Strong institutional move” alone is too subjective to audit later.

Multiple overlapping candidates need a selection rule. One manageable starting point is to take only the first qualifying propulsion candle after the parent becomes valid, then allow only its first eligible return after confirmation. If it fails or expires, end that attempt. This reduces the temptation to keep selecting new candles until one finally produces a winning screenshot.

4. Calculate mean threshold from the body, not the wick range

The measurement used here is straightforward: mean threshold = (open + close) ÷ 2. Because a candle’s body lies between those two values, its colour does not change the arithmetic. Wicks remain important for contact with the parent and for alternative risk definitions, but they are excluded from this body midpoint calculation.

Our bullish example uses a down candle with open 1.08698, high 1.08706, low 1.08621, and close 1.08650. Its body spans 4.8 pips, so its midpoint is (1.08698 + 1.08650) ÷ 2 = 1.08674. Its high-to-low midpoint is instead (1.08706 + 1.08621) ÷ 2 = 1.086635, approximately 1.08664 at five decimal places. Those references differ by about one pip, which is significant when the proposed stop is only a few pips away.

A long lower wick makes this error especially easy. It stretches the full candle range while leaving the body calculation unchanged. If your charting tool automatically places a fifty-percent line inside a high-to-low rectangle, that line is not the body’s mean threshold. Either draw a separate body rectangle or enter the midpoint numerically.

Keep three decisions separate: the midpoint you measure, the amount of penetration your model tolerates, and the executable stop order you place. Our examples use a small stated buffer beyond mean threshold. That buffer is a modelling choice, not a claim that ICT requires the same number of pips on every instrument. Define it before entry and include it in position sizing. If your platform rejects the distance or spreads make it impractical, skip the attempt or test a different complete rule set.

5. How to trade it with rules you can actually test

For a bullish setup in this study, the qualifying departure must close above the propulsion candle’s high. Only afterward can a buy limit be placed at that high for a future return. The bearish mirror uses a completed close below the propulsion candle’s low, followed by a sell limit at that low. The limits are placed while the market is on the correct side of the entry; an already passed level is not a valid historical fill.

Before placing the order

Identify the intended liquidity destination before calculating reward. A nearby previous swing high can help frame a bullish objective, while a previous low can frame a bearish one. Position the hypothetical take-profit slightly ahead of that reference if your model requires an allowance. Do not select a distant target solely because it produces an attractive ratio. If meaningful opposing structure lies between entry and target, record that obstacle and apply your exclusion rule.

Our research model requires at least 2R after the stated cost allowance, permits only one attempt per selected formation, and cancels an unfilled order after three completed execution candles or at the session cutoff, whichever comes first. It also cancels if the target is reached before entry. If price has crossed mean threshold against the planned direction after confirmation but before you can place the order, skip the attempt. The midpoint sensitivity test begins after departure, not during the original candle’s formation. These are starting assumptions for testing, not evidence that the settings are profitable.

After a fill

The examples use one protective stop and one full-position target. They do not add partial exits, automatic break-even moves, or discretionary stop widening. This keeps the arithmetic transparent. The stop governs the defined invalidation allowance after entry, and any adverse slippage is recorded separately. If neither exit has occurred by the session cutoff, close at the available executable price and record the actual result, including costs.

Someone using a lower timeframe confirmation entry is testing a different implementation. Confirmation can avoid some weak returns, but it can also place entry farther from the protective level or miss a move entirely. Compare that version with the limit model in separate samples. The site’s guide to confirmation versus aggressive entries explains that trade-off in more detail.

Define what “responsive” means

Do not describe every eventual winner as an immediate response. Record how many candles elapsed after entry before price first moved one initial risk unit in your favour. Also record the deepest adverse movement during that interval. You can later compare quick and slow responses without pretending that an arbitrary time limit is part of the concept’s definition. Any new time-based exit should be evaluated as a new rule, including trades where it cuts off a later winner.

News can make an otherwise small-distance model untradeable. Mark scheduled releases for both currencies, define a no-entry window before the session, and keep it consistent in replay. The relevant question is whether your orders and cost estimates remain plausible during that window. A textbook candle sequence does not make an unstable spread or an uncertain fill disappear.

6. Complete bullish EUR/USD example with explicit risk calculations

This is a constructed teaching sequence, not a historical trade, current signal, or performance claim. All stated candle times are closing times on a five-minute chart during a hypothetical winter London session in Ghana GMT. The hourly context is assumed bullish, with an already identified buy-side liquidity reference at 1.08883. The selected take-profit is 1.08866, 1.7 pips ahead of that reference.

Build the parent and newer reference in order

The parent candle closes at 7:35 AM with open 1.08634, high 1.08642, low 1.08582, and close 1.08596. Its full range is the recorded parent interval. By 7:45 AM, a departure closes at 1.08723, above an internal high previously marked at 1.08688. The parent is now eligible under the study rules, before the later contact occurs.

At 8:00 AM, the newer down candle closes with the open, high, low, and close used in Section 4: 1.08698, 1.08706, 1.08621, and 1.08650. Its low has entered the parent interval, while its body remains above the parent’s high. Its body midpoint is fixed at 1.08674. The extended lower wick does not move that midpoint down to the parent.

The 8:05 AM candle then closes at 1.08736, above the newer candle’s 1.08706 high. After that close, the plan places a buy limit at 1.08706, a protective stop at 1.08666, and a take-profit at 1.08866. The stop is 0.8 pip below mean threshold. At 8:10 AM, the hypothetical next candle trades down to 1.08693, allowing the planned limit fill in the simplified reference-price model, before recovering. Later candles reach the target first.

Include costs before deciding position size

For teaching, assume a total round-trip friction allowance of 0.8 pip. This is an illustrative allowance for spread, commissions expressed in pips, and expected execution friction, not a quoted broker price. Apply it once to each outcome: the losing outcome costs 4.0 + 0.8 = 4.8 pips, while the winning outcome yields 16.0 − 0.8 = 15.2 pips. The resulting ratio is 15.2 ÷ 4.8 = 3.17R, or approximately 1:3.17.

On a hypothetical US$2,500 account, a chosen risk budget of 0.4% equals US$10. For a standard EUR/USD contract of 100,000 euros in a US dollar account, pip value is US$10 per standard lot. The cost-inclusive size is US$10 ÷ (4.8 × US$10) = 0.2083 standard lots. If the broker permits increments of 0.01 lot, rounding down to 0.20 lot gives an estimated loss of 4.8 × US$2 = US$9.60 and a target-first gain of 15.2 × US$2 = US$30.40. Slippage beyond the allowance can increase the loss.

These calculations use reference prices plus a separate friction allowance. In actual bid-and-ask execution, a long opens at the ask and closes at the bid. A chart touch alone may therefore be insufficient to establish a limit fill. When your test uses executable bid-and-ask prices, account for the spread through those prices and add only remaining costs. Do not deduct the spread twice.

Why moving the stop changes the entire setup

Consider a separate, preplanned version with the same entry and target but a stop below the propulsion wick at 1.08609. Its price risk is (1.08706 − 1.08609) ÷ 0.00010 = 9.7 pips. Reward remains 16.0 pips, so the ratio before costs becomes 16.0 ÷ 9.7 = 1.65R. With the same allowance it becomes (16.0 − 0.8) ÷ (9.7 + 0.8) = 15.2 ÷ 10.5 = 1.45R.

That version fails this study’s minimum 2R after costs, so it is excluded. It could belong to a different tested model, but it cannot be substituted after the smaller stop comes under pressure. Wider invalidation requires a fresh sizing calculation and a fresh eligibility decision before entry.

7. Bearish example: reverse the sequence, not the arithmetic

For a separate hypothetical winter New York session, assume a bearish context and an already identified sell-side liquidity low at 1.09113. The parent is an up candle closing at 12:20 PM GMT with open 1.09344, high 1.09416, low 1.09336, and close 1.09398. Subsequent departure closes at 1.09253 by 12:30 PM, below the previously marked internal low of 1.09312.

The 12:45 PM propulsion candidate is an up candle with open 1.09278, high 1.09364, low 1.09269, and close 1.09324. Its upper wick overlaps the parent interval from 1.09336 to 1.09416. Its body midpoint is (1.09278 + 1.09324) ÷ 2 = 1.09301. After the 12:50 PM candle closes below its low at 1.09238, a sell limit is placed at 1.09269 for a later retracement.

The protective stop is 1.09309, 0.8 pip above the midpoint, and the target is 1.09129, 1.6 pips ahead of the liquidity low. A later retracement to 1.09283 reaches the entry without reaching the stop in the simplified reference-price sequence. The illustrated continuation then reaches the target. As with the long, actual fill eligibility must be checked against the broker’s executable quotes.

Price risk is (1.09309 − 1.09269) ÷ 0.00010 = 4.0 pips. Price reward is (1.09269 − 1.09129) ÷ 0.00010 = 14.0 pips. Therefore the risk-to-reward ratio before costs is 1:3.5, because 14.0 ÷ 4.0 = 3.5R. With a 0.8-pip round-trip allowance, the loss is 4.8 pips, the profit is 13.2 pips, and the cost-adjusted ratio is 13.2 ÷ 4.8 = 2.75R, or 1:2.75.

A short position closes at the ask. A bid-based chart can consequently appear to remain below a short stop while the ask reaches it. This is one reason to preserve execution records instead of judging every loss from a screenshot. The model’s invalidation calculation and the broker’s trigger convention must be reconciled before you interpret a stop-out as evidence that the pattern was drawn incorrectly.

8. Ghana session timing and practical preparation

Ghana stays on GMT throughout the year, while New York changes between UTC−5 and UTC−4. The familiar London Kill Zone of 7 to 10 AM Ghana time and New York Kill Zone of noon to 3 PM Ghana time are the winter conversions of the New York clock windows used below. They are not fixed year-round conversions. The underlying offsets are documented by Ghana’s time zone reference and New York’s clock-change reference.

Set a chart clock and journal clock you can reconcile. Broker server time may differ from both Ghana and New York, and London and New York do not always change clocks on the same date. This table converts the stated New York anchors; it does not assert that London’s local business opening remains at a constant distance from them.

A useful preparation routine is to check the day’s scheduled releases, mark existing higher timeframe references, record the next plausible liquidity objective, and note the platform’s current spread. Then look for the sequence during your chosen window. A propulsion candidate outside that window is a separate observation, not permission to extend the session indefinitely. The London Kill Zone guide provides broader session preparation, while this article concentrates on the candle relationship.

9. Failure cases, realistic testing, and the cost of small stops

The most revealing failure is a correctly identified setup that still loses. The parent exists, contact occurs, confirmation finishes, and the subsequent entry reaches its stop first. Keep it in the sample. Removing it because the response was not sufficiently “institutional” turns the concept into a label applied only after success.

Other outcomes belong in different categories: no contact, no completed confirmation, target reached before an eligible entry, expired order, execution uncertainty, and filled trade. Combining all of them into one win-rate denominator makes the result difficult to interpret. Report how many candidates produced orders and how many orders actually filled. A model with attractive outcomes but rare eligible fills can be very different from what a collection of screenshots suggests.

Replay without giving yourself future knowledge

Advance one candle at a time. At each decision, record the information currently available, including whether the parent and execution candle are completed. Use fixed target and stop rules. If a candle’s range contains both stop and target and you cannot determine which was reached first, classify the outcome as ambiguous. Resolve it with lower timeframe or tick data when available; otherwise use a conservative assumption and report it.

Record maximum adverse excursion, maximum favourable excursion, fill delay, time to first favourable risk unit, spread at entry, realised slippage, and whether a scheduled event was nearby. Keep losing screenshots alongside winning ones. The trading journal guide can help organise those records, but the essential discipline is preserving the original decision before the result appears.

Measure sensitivity to costs and rule changes

A useful stress test increases the cost allowance while keeping the same candidates and price outcomes. Another tests entry one executable increment worse. These exercises show whether the apparent advantage depends on fills that your account rarely receives. Do not rescue a disappointing result by simultaneously changing the parent convention, target distance, stop buffer, and session window. You would no longer know which change produced the difference.

Estimate expectancy using realised outcomes in consistent risk units: average expectancy = win probability × average winning R − loss probability × average losing R, with any other outcome categories included appropriately. A planned 4R target is not an average winning trade, especially if some positions close at the session cutoff. Use actual net results, count the losses, and reserve later unseen sessions for checking whether a rule survives outside the development sample.

Small stops also interact with leverage, minimum position size, and margin requirements. The CFTC’s forex customer advisory explains that leverage can amplify losses and that trading conditions depend on the dealer. Its US regulatory references do not establish a broker’s authorisation in Ghana. For this setup, the practical lesson is to verify your own contract specification and execution terms, then practise the complete process on historical replay or demo before considering capital exposure.

10. Frequently asked questions

Build your understanding in sequence: review ICT order blocks, examine the role of fair value gaps, practise identifying market structure shifts, organise your London Kill Zone preparation, and compare the session-open behaviour described in the ICT Judas Swing guide. Then return to propulsion blocks with one written rule set and a journal that includes every eligible outcome.