ICT IPDA Data Ranges: 20, 40 and 60-Day Lookbacks

A chart can contain more old highs, lows, order blocks, and fair value gaps than a trader can use. Zoom out far enough and there will almost always be another level that appears to justify a bullish or bearish opinion. The practical problem is deciding which part of that history belongs in today’s analysis before the market reveals its next move.

ICT IPDA data ranges provide a time-based way to organise that search. The familiar 20, 40, and 60 trading-day lookbacks define how much completed daily history you inspect. Their value is in making the selection process explicit: which sessions were included, which price references remain relevant, and why a nearer or older objective deserves attention.

This guide turns that idea into a reproducible chart routine. It explains nested windows, daily candle boundaries, rolling updates, and the distinction between a historical range and an entry signal. The price data and trades are hypothetical teaching examples, with complete risk calculations and a losing scenario. Ghana timing guidance helps connect the daily map to an intraday execution window.

1. What are ICT IPDA data ranges?

IPDA stands for Interbank Price Delivery Algorithm in ICT terminology. In the Month 7 lesson on blending data ranges with PD arrays, ICT distinguishes the time component of the lookback from the price component of the arrays. He describes reviewing the previous 20, 40, and 60 trading days and shifting that reference forward as new days arrive. Source: ICT’s data ranges and PD arrays lesson.

A PD array is a price reference used in the ICT framework, such as a fair value gap, order block, breaker, or old swing extreme. A data range is the historical window in which you look for those references. These are different objects. The dates define eligibility for the search, while the candles define the actual prices and the events surrounding them.

For example, a high from seven trading sessions ago lies inside all three lookbacks. A high from twenty-nine sessions ago lies inside the 40-day and 60-day windows but outside the 20-day window. That age information says nothing by itself about whether either high will be taken next. It simply makes the source of the level traceable.

The algorithm language is part of ICT’s explanatory model. It is not evidence that a retail chart reveals the instructions of a single market-wide program. The practical method can be described and tested without claiming access to institutional code: select the observations, identify price references, state a directional hypothesis, and record whether subsequent prices support it.

This article uses a rolling lookback anchored to the start of the analysis session, with the current unfinished daily candle excluded. That is a fixed research convention for the examples. A study anchored to a month opening or another reference date must be labelled separately. Two charts with different anchor dates can produce different ranges without either person making an arithmetic error.

2. Why the lookback matters in institutional order flow analysis

ICT and SMC analysis asks how price may move between liquidity references and areas of inefficient delivery. The lookback supports that task by limiting the history used to frame the idea. Instead of selecting an old high because it fits an opinion, the trader can explain that the high belongs to a defined set of recent sessions and remains relevant under a stated rule.

This helps separate the location of a setup from its trigger. An old daily low can identify an area in which to observe a possible liquidity excursion. A later intraday displacement may provide evidence that delivery has changed. Neither event is interchangeable with the other. Buying solely because a price is the lowest of twenty days skips the part of the analysis that establishes an entry and invalidation.

The method also creates a useful stopping point for research. ICT’s Model 2 trade plan starts with the 20-day range, considers longer windows when the nearer history no longer offers a clear array, and allows standing aside when the broader search still lacks a logical setup. The essential lesson is disciplined selection, not an obligation to find a trade in every window. Source: ICT Model 2 trade plan, around 2:30 to 3:52.

A longer lookback does not automatically create stronger institutional evidence. An older level may be useful because of its price location and the path leading toward it, while a recent level may already have been visited repeatedly. Age and condition should be recorded separately. “Inside sixty days” is an eligibility description, not a quality score.

Remember the limits of the underlying data. A spot FX chart shows prices supplied by its feed, not every stop or order held across the market. The BIS describes fragmented FX execution and dealer internalisation, which limits what one chart can reveal about aggregate positioning. Treat liquidity pools as analytical references rather than a complete visible order inventory. Source: BIS, FX execution algorithms and market functioning.

3. Count trading sessions correctly before marking prices

Use completed daily sessions, not calendar subtraction

Under this article’s convention, session one is the most recent completed daily trading candle before the reference session. Session twenty is the twentieth completed candle when counting backward. The current daily candle is not part of that frozen historical set. Its developing high and low are tracked separately until the next scheduled update.

Twenty trading sessions are often roughly four calendar weeks, but they are not the same as twenty calendar days. Weekends, market holidays, missing bars, and the provider’s daily aggregation can alter the relationship. Count eligible observations on the chosen feed rather than subtracting an approximate number of days and assuming that the result is correct.

For a calendar-only illustration, suppose the reference date is Monday, September 14, 2026, and assume that every preceding Monday through Friday contributes exactly one completed daily session. The newest observation is Friday, September 11. The twentieth is August 17, the fortieth is July 20, and the sixtieth is June 22. This exercise verifies counting logic only; it does not assert that a particular broker supplied an uninterrupted dataset on those dates.

Handle Sunday bars and session boundaries deliberately

ICT’s Model 6 preparation lesson explicitly excludes Sundays from its trading-day count. That instruction needs a consistent chart convention: a separate short Sunday candle should not become an extra full trading day in the study. Prefer a feed or aggregation that represents the intended five-session forex week coherently. Do not simply erase valid Sunday evening prices from your dataset. Source: ICT Model 6 buyside trade plan, around 1:20.

Record how your provider labels daily candles and where each session ends. If you choose a daily boundary of 5 PM New York, that is a specific aggregation rule, not Ghana midnight. It corresponds to 10 PM Ghana time during New York standard time and 9 PM during New York daylight time. A GMT-midnight chart can split the same price activity into different daily candles.

Changing the chart display timezone may alter only the labels, rather than rebuild the daily bars. Verify what the platform actually does. A count that seems correct on screen can still differ from another trader’s count if their provider produces six bars in some weeks or uses a different daily boundary.

If data are missing, flag the gap and decide whether the sample remains usable. A genuine session with no trading is different from an outage in your historical feed. Do not count a missing observation as though you knew its high and low, and do not quietly shorten the historical coverage while continuing to call it a sixty-session study.

4. Understand nested lookbacks and rolling updates

The three lookbacks share the same recent endpoint. The last forty sessions contain the last twenty, and the last sixty contain both shorter windows. They are not three independent twenty-day blocks. You may colour the older additions separately for readability, but a 40-day high must be calculated from all forty included sessions.

Those inequalities are useful error checks. If your forty-day high is below your twenty-day high, the windows were not counted consistently or the calculation omitted observations. Likewise, an older twenty-day segment can have a higher low than the recent segment, but that does not make it the low of the entire forty-day range.

At the next daily update, each rolling window gains the newly completed session and loses its oldest included session. For the twenty-day window, yesterday’s session twenty becomes session twenty-one and falls outside that shorter search. It can remain inside the forty-day and sixty-day searches. The level’s historical existence does not disappear when its age changes category.

A rolling high can therefore move downward even if current price never trades lower. The old maximum may simply leave the window. Similarly, the midpoint can change because an old extreme is removed. That is a change in the dataset, not proof that price has just created a new displacement or that institutions have changed direction at that exact moment.

For intraday execution, this guide freezes the pre-session map and schedules the next range calculation after another daily session completes. New intraday highs, lows, and reactions are still observed. They can invalidate a trade, but they do not silently rewrite the original reference map while the trade is being evaluated. Save a new version when the next scheduled update occurs.

5. Map the highs, lows, and arrays on a complete example

Consider sixty invented EUR/USD daily candles. They are a measurement exercise, not prices from the calendar example above. The most recent completed close is 1.08605. The last twenty sessions have a high of 1.08837 and a low of 1.08213. Extending the same endpoint to forty sessions adds a higher high at 1.09163 but no lower low. The sixty-session window adds a high at 1.09471 and a low at 1.07685.

The midpoint formula is (highest high + lowest low) / 2. For twenty sessions, (1.08837 + 1.08213) / 2 = 1.08525. For forty sessions, (1.09163 + 1.08213) / 2 = 1.08688. For sixty sessions, (1.09471 + 1.07685) / 2 = 1.08578. These are range-location measurements, not separate trade examples.

The same reference price is in premium relative to the twenty-day range, discount relative to the forty-day range, and premium relative to the sixty-day range. There is no contradiction because the denominators differ. An unqualified statement that EUR/USD is “in discount” leaves out the range that makes the statement meaningful. State the selected window and anchors whenever you use that language.

The twenty-day low at 1.08213 is also the forty-day low. That does not create two independent pools of evidence. It is one price extreme that belongs to two overlapping datasets. Counting the duplicated label as additional confirmation would overstate what the chart contains. The older window adds useful information only where it actually introduces different references or context.

Now examine arrays inside the selected history. Record each candidate’s formation session, price interval, subsequent visits, and relationship to the intended draw. Do not mark every candle as an order block or assume every gap remains unvisited. The premium and discount guide explains range location, while the internal and external liquidity guide helps distinguish an objective inside a range from one beyond its boundary.

6. Select a useful reference before planning the entry

Begin with the nearest window that provides a clear, relevant price narrative. If the recent history has already been extensively revisited or offers no suitable destination, inspect what the longer window adds. This is a search hierarchy, not a requirement that price must visit the twenty-day extreme, then the forty-day extreme, and finally the sixty-day extreme in order.

Keep a separate status for every candidate. A level may be unvisited, already traded through, currently being revisited, or invalid under the specific setup rule. “Swept” records that price passed a reference; it does not prove that every order at that price was removed forever. A previously used level can matter in another model, but reusing it requires a new explanation rather than pretending it is untouched.

If price trades below the twenty-day low, the next useful reference may come from the sixty-day history because the forty-day window adds no lower level. That is exactly the kind of situation a nested calculation makes visible. There is no obligation to invent a forty-day target or force an intermediate stop in the narrative.

Also match the destination to the holding period. A sixty-day range boundary several hundred pips away can help describe broader context without being a reasonable objective for a short London trade. An intraday array or local liquidity pool may provide a more appropriate exit. The historical window identifies the setting; the execution model determines how much of any movement is actually being attempted.

1. Fix the reference session and feed. Record the last completed daily candle and the counting convention.
2. Calculate the nested windows. Check their high-low relationships and save the dates or session indices.
3. Choose the relevant price reference. State its age, interval, condition, and role as a possible entry context or destination.
4. Wait for the intended intraday sequence. The worked protocol below requires a liquidity excursion, a completed close through a local swing, and a completed three-candle FVG.
5. Set the entry, stop, and target independently. Calculate the actual risk and reward rather than using the entire daily range as an assumed payoff.
6. Define expiry and invalidation. Cancel an unfilled order if the objective is reached first, the premise fails, or the chosen observation window ends.

7. Complete bullish EUR/USD example at a twenty-day low

The historical map identifies where to pay attention

Use the hypothetical daily map above in a separate imagined winter London session. The previous completed close was 1.08605, and price has since weakened toward the frozen twenty-day low at 1.08213. All intraday times below are five-minute candle closing times in Ghana GMT. This is a constructed sequence, not a report of a real session or a forecast from the September counting exercise.

Before the entry sequence, an intraday buy-side reference has been marked at 1.08543. The intended long would seek a portion of a possible recovery toward that reference. The twenty-day high at 1.08837 remains part of the wider map, but the trade does not assume that the whole historical range will be crossed during the morning.

An intraday high forms at 1.08297 at 7:45 AM. During the bar closing at 8:00, price falls to 1.08197, passing the twenty-day low of 1.08213 by 1.6 pips. That excursion is a reason to observe the response, not permission to buy immediately. Price could continue lower toward an older reference if the attempted reversal fails.

The entry depends on later price delivery

The 8:05 candle has a high of 1.08259. The displacement candle closing at 8:10 finishes at 1.08321, above the local high of 1.08297. The 8:15 candle has a low of 1.08279, completing a bullish FVG from 1.08259 to 1.08279. The midpoint is (1.08259 + 1.08279) / 2 = 1.08269.

Only after that third candle completes does the protocol authorise a later buy limit. During the candle closing at 8:20, assume the broker’s ask reaches 1.08269 and fills the order. The displayed bid low reaches 1.08257. In a real study, the ask evidence would need to support the fill rather than being inferred automatically from the bid candle.

The stop is 1.08183, which is 1.4 pips beneath the sweep low at 1.08197. The target is 1.08527, positioned 1.6 pips before the intraday buy-side reference at 1.08543. In the illustrated path, price reaches the target during the candle closing at 9:05 without touching the stop. No partial exits, trailing stop, or breakeven move is included.

For an illustrative USD account with $2,000 equity and a chosen 0.5% risk budget, the maximum planned loss is $10. Assume a standard EUR/USD lot represents 100,000 euros and has an approximate $10 pip value in the USD account. Raw position size = $10 / (9.4 × $10) = 0.106383 lots. With a 0.01-lot increment, round down to 0.10 lots.

At that size, the estimated stopped loss is 9.4 × $10 × 0.10 = $9.40, and the estimated net target gain is 25.0 × $10 × 0.10 = $25.00. Their ratio remains approximately 2.66. These contract, cost, and risk assumptions are teaching inputs. Check the actual instrument specification and account conversion; a gap or unusually poor execution can exceed the allowance.

The twenty-day low supplied context, while the local structure and FVG supplied the entry rule. The daily low’s age did not determine the stop size, and the sixty-day boundary did not determine the target. If no intraday confirmation had appeared, the historical map would still have been valid research with no authorised long entry.

8. A bearish example that loses despite the historical context

Consider another invented winter New York setup with its own pre-session map. The prior twenty completed sessions have a high at 1.09643 and a low at 1.08827. Price approaches the high, while a nearer intraday sell-side reference at 1.09307 offers a possible downside destination. The range is background context; the plan is not to sell simply because price is near its upper boundary.

At 12:05 PM Ghana time, price reaches 1.09661, exceeding the frozen twenty-day high by 1.8 pips. A subsequent completed close at 1.09551 breaks an intraday low at 1.09563. The three candles closing at 12:10, 12:15, and 12:20 form a bearish FVG: the first candle’s low is 1.09599 and the third candle’s high is 1.09575.

The midpoint is (1.09599 + 1.09575) / 2 = 1.09587. At 12:30, assume a later bid retracement fills a sell limit at that price. The stop is 1.09675, 1.4 pips above the raid high. The target is 1.09323, placed 1.6 pips before the selected intraday low at 1.09307.

Price initially falls to 1.09511, then reverses upward. During the bar closing at 1:00 PM, the ask reaches 1.09681 and crosses the stop. Assume baseline stop execution at 1.09675, with the separate additional cost allowance applied below. No partial profit or stop adjustment was authorised, so the initial favourable movement does not change the stopped result.

Using the same $10 illustrative budget, raw size is $10 / (9.6 × $10) = 0.104167 lots, again rounded down to 0.10. This confirms that the stopped loss fits the planned budget under the assumptions. It does not guarantee that an actual stop order will fill at the intended price during every market condition.

The correct review preserves both facts: the twenty-day high was identified before the trade, and the resulting short lost. Calling the high “institutional” or placing it inside a longer lookback cannot cancel that outcome. A later move lower would require a separately qualified trade rather than a revision of the original result.

9. Ghana timing: connect the daily map to the session

Ghana stays on GMT throughout the year. New York changes between standard time and daylight saving time, so a New York-anchored ICT window shifts by one hour on a Ghana clock. The familiar London Kill Zone of 7 to 10 AM and New York Kill Zone of noon to 3 PM Ghana time are winter conversions for the definitions shown below. Ghana time reference; New York time reference.

London and New York do not change clocks on the same dates. If your session rule is anchored to New York, convert that reference for the actual date. A routine anchored to the local London opening is a different rule during some transition weeks. Keep the anchor explicit instead of treating the word “London” as a complete timezone instruction.

The midnight open is also different from the daily candle boundary used for the lookback count. It is a price reference at a specified New York time. A trader can use that opening price in intraday analysis while counting daily sessions under another documented aggregation. Mixing the two clocks silently can change which observations are considered complete.

For a Ghana-based trader with limited chart time, prepare the daily map once at a consistent point and keep a short list of the levels closest to the intended session. Update their intraday status as price develops. A phone chart does not need sixty labelled levels to preserve sixty sessions of context. The recorded dates and prices can remain in the journal while the execution view shows only what is relevant.

10. Avoid hindsight when reviewing IPDA data ranges

Save the map before the trade. A historical screenshot created at the end of the day can accidentally include the very low or high that the analyst claims to have anticipated. Keep the reference timestamp, eligible session list, calculated extremes, selected arrays, and intended update time. Those records make it possible to reconstruct what was actually available at the decision point.

Separate a source level’s age from its state. The journal should say both “formed sixteen sessions ago” and “first traded below during today’s London window,” for example. If you record only that it was a twenty-day low, you lose the information needed to distinguish an untouched objective from a level already crossed several times.

Do not treat the twentieth, fortieth, or sixtieth session as a scheduled reversal date. These numbers define a historical search under this framework. They are not a countdown that forces a new high, low, or trend change when the count reaches a boundary. A rolling update can alter the displayed range without creating a tradable price event.

Test the value of the context filter separately from the entry model. One study might hold the same intraday confirmation, costs, and exit rules constant while recording whether each setup occurred at a reference from the defined lookbacks. Compare complete samples, not only the cases where an old daily level happened to produce a dramatic move.

Track eligible sessions, qualified setups, filled orders, unfilled orders, cancellations, and realised net R. A day without a setup belongs in the opportunity record. A target that is reached after the stop does not count as a winning trade. If both target and stop appear inside one historical candle, use finer quote data or a conservative ambiguity rule chosen before testing.

In the usual bid/ask convention, buys open at the ask and sells at the bid, while exits use the opposite transaction. A short stop can therefore trigger on the ask even if a bid-only chart appears not to touch it. The same issue affects whether a limit entry was filled. Use broker execution data where available, and avoid counting spread twice when actual executable prices already include it. Source: MetaTrader 5 trading concepts.

Retain a later sample that was not used to design the rules. If the definition of a usable array, the session boundary, or the order expiry changes, start a new version of the study. An honest finding that a broader lookback added little value is more useful than an expanding collection of exceptions. The forex trading journal guide provides a place to organise those records.

11. Frequently asked questions about IPDA lookbacks

To connect the daily lookback with a complete ICT study routine, review fair value gap boundaries, learn how to qualify order blocks, organise execution with the London Kill Zone guide, distinguish local confirmation through market structure shifts, and place the opening excursion in context with the ICT Judas Swing. Let the lookback make the historical search consistent, then require the trade itself to justify its entry, risk, and destination.