Atlas IB (NQ) guide

What this indicator is

Atlas IB takes any 30- or 60-minute window of the NQ futures day, measures the range price made during it (the Initial Balance, or IB) and shows what has historically happened next: how often price reached the upper extension target first, the lower extension target first, or neither before the session closed.

The three figures add to 100% before rounding. They are measured frequencies from seventeen years of NQ history, not forecasts. A 62% on the upper rail means that, in the measured history, comparable windows reached the upper target first 62% of the time. It says nothing certain about today. What it gives you is the base rate you would otherwise be guessing at.

This is context, not a signal. There is no buy or sell instruction anywhere in the indicator, by design. It tells you what usually happens, so that when you take a trade you know whether you are trading with the historical base rate or against it.

Setup

Getting it

Atlas IB (NQ) is free on any TradingView plan, including the free one. Add it from its TradingView page ↗. This guide matches the published build, "Atlas IB (NQ) v1.25 [by AlgoTrader NQGC]", which shows as "AtlasIB NQ" in the chart legend.

Symbol

NQ or MNQ futures. The statistics are measured on CME NQ and apply to both, since MNQ tracks the same index. They do not apply to any other instrument. Gold has its own build, measured separately, described on the Atlas IB page.

Timeframe

Any intraday chart timeframe that divides 30 minutes: 1m, 2m, 3m, 5m, 6m, 10m, 15m and 30m all qualify. The figures are clock-time based, so they are identical on all of them; use whichever you like to look at. On any other timeframe the indicator draws nothing, and with Hide on unsupported timeframes on (the default) the table is replaced by a one-line notice rather than showing numbers that do not apply.

First run

Open the settings and choose your window. IB start (NY time) sets the 30-minute boundary the window starts at, and IB duration (min) sets 30 or 60 minutes. Every 30-minute boundary of the futures day is available, 18:00 through 16:30 New York time; the default is the 09:30 cash open with 60 minutes. The 16:30 start works with the 30-minute duration only: a 60-minute window would run past the close, and nothing is drawn. The third input that shapes what you see is Extension target (× IB range), covered under the rails below.

Contract rolls

A contract roll does not stretch the target distances or the run-on guides. On the roll day itself, two optional features can still show the price gap between the expiring and the new contract: the Overnight-gap context and Prior-day-high context adjustments (both off by default), and previous-IB levels projected with Extend set to To current bar.

Reading the chart

The IB box

The box spans the window's high and low. Once the window completes it is tinted towards the favoured side, teal for up and red for down by default, or grey when the indicator abstains. A dotted line marks its midpoint. While the window is still forming, the box is dashed and grey and tagged "IB building…".

A label above the completed box gives the IB range in points, the average range of the last five IBs drawn for this window, and a behaviour tag (see the two regime tags, under the status table).

Beneath the box a compact line restates the three shares, ▲ up first, ▼ down first and ∅ neither, tagged "null %" (the indicator's term for the historical base rate), together with n, the number of comparable historical sessions they rest on. When an optional adjustment is switched on and covers this window, a second line appears showing each adjustment and the shift it applied in percentage points, and the first line then shows the adjusted shares.

The extension rails

Two dotted lines are projected above and below the box at a multiple of the IB range; these are the targets. Each carries a badge with the historical frequency of that side being reached first, and a second line, 🏁 2h · 4h, giving how often that side was reached first within two hours and within four hours of the window closing, rather than all the way to the close.

The multiple is set with Extension target (× IB range): 0.25, 0.5 or 1.0 × the IB range, or Auto. 0.5 is the default, and the setting we recommend on NQ. 0.25 puts the rails nearer, so they are reached more often; 1.0 puts them further away, so they are reached less often. Because the rails sit at a multiple of the IB's own range, a fixed multiple already adapts to session size: a wide IB gets a proportionally distant target, and on NQ roughly the same share of sessions resolve whether the IB was quiet or wide.

Auto varies the multiple per session with the IB's size tag. It buys extra reach on quiet sessions, and what it costs you is that even resolution rate: expect the rails to go unreached more often after a compressed IB, and less often after a wide one. It does not sharpen the probabilities. Choose it if you want the reach, not because it is a better forecast. On Auto the multiple is also widened when the target would otherwise sit under about 15 points. Whatever you choose, check the Target range row: a target under about 15 points sits inside typical friction, and that is not a real target.

Atlas mode and abstain

Every completed window ends in one of two states. When one side clears the Abstain threshold % (62% by default), and on the default basis enough comparable sessions resolved at all, the window is in Atlas mode: the box is tinted towards that side, the rails show in their up and down colours, and the favoured rail's badge carries a magnet 🧲. When neither side clears it, the box, both rails and their badges go grey and the badges add "· abstain". That is the indicator telling you it has nothing useful to say about direction for this window.

Abstain sessions are common, and they are not a failure: a coin-flip base rate honestly reported is more useful than a manufactured lean. Treat the grey as risk information rather than silence. Knowing before you commit that the history offers no lean is worth having, and standing aside is a legitimate plan.

Which number the threshold is applied to. The main figure on each badge is the three-way share: the share of all comparable sessions in which that rail was reached first. On the two Direction settings of Atlas mode basis, including the default, the threshold is applied to a different figure, printed after it on the leading side's badge as "dir": that side's share of only those sessions in which either rail was reached. A window whose shares are ▲ 65%, ▼ 16% and ∅ 19% prints "▲ 65% · dir 80%", so at a 75% threshold the Direction settings name the up side while Outcome (3-way) abstains. The 🧲 ATLAS MODE row in the status table repeats the rule in force and the value this window reached against it.

Two kinds of abstain, and the (scalps) tag. Abstain means direction is unclear, and that happens for two different reasons which matter differently at the desk. Either the two rails are close to a coin flip, in which case a rail may well still be reached and the session can run hard; or too few comparable sessions reached either rail at all, in which case the range itself tends to hold. The badges print (scalps) for the second case. When you see it, comparable sessions mostly stayed inside both rails, so a directional run is less likely and shorter range-bound trades may suit better than waiting for a break. Without the tag, read the grey as honest uncertainty about direction, not as a forecast of a quiet session. The tag is set by Min decided share % (floor basis), so it appears whenever that share is low, on any basis and whether or not a side is named.

Outcome marks. When a race resolves, a mark prints at that bar, beyond the rail that was reached: ✅ the higher-percentage side was reached first, ❌ the other side got there first, ◇ both rails were touched within the same chart bar, so the order cannot be told. Marks print on every session that resolves, abstain sessions included, and a 🧲 in front of the mark means the window was in Atlas mode. No mark prints when neither rail is reached by the close. The marks are on by default, and they are the chart scoring the figures in front of you, window after window: the visible counterpart of the tally rows in the table.

Run-on guides (optional, off by default)

Switched on with Show MFE excursion lines, these are dotted guides beyond each target showing the median distance price continued past that target within 15, 30 and 60 minutes of reaching it, labelled "15m MFE", "30m MFE" and "60m MFE". A 30-minute IB draws the 15m and 30m layers; a 60-minute IB draws all three.

They are off by default for a practical reason worth knowing: the guides sit well beyond the rails, and TradingView's automatic price scaling makes room for everything the indicator draws. With them on, the scale stretches to include them and the IB box renders smaller. If you turn them on and the box looks squashed, that is the scale making room for the guides, not the box changing. Right-click the price scale and choose "Scale price chart only" if you would rather the scale ignored them.

Read these figures carefully, because they are easy to misread. Continuation past a target behaves like drift, not momentum. Reaching the extension target does not ignite a move. The run-on figures are useful for shaping a second target or a trail; they are not evidence that continuation is likely. Read them as "how far price typically drifts", not "where it is going". One asymmetry worth knowing: downside run-ons are generally larger than upside ones.

Projected levels (optional, off by default)

A completed IB is not only a box. Its edges, midpoint and quarters stay useful as reference prices after it closes, and under IB levels: projection you can carry the most recent completed window's geometry forward:

• Extend IB EQ (50%) projects the midpoint.
• Extend IB quadrants (high / low / 25% / 75%) projects the high, the low, and the 25% and 75% levels.

Extend sets how far they run. To next IB window runs them one window-length past the IB close, so a 30-minute IB projects 30 more minutes and a 60-minute IB an hour, and the levels stop while they are still about the window that produced them. To current bar keeps them travelling with the session, ending a few bars ahead of the latest bar (Bars beyond the last bar (current-bar mode), 3 by default); use it if you want them present all day as reference.

Only the most recent completed window is projected. That is deliberate: a chart carrying every past window's quarters is unreadable, and the levels stop being about the session you are trading. Colour, width and line style are configurable, defaulting to a quiet grey with solid high, low and midpoint and dashed quarters, so they sit behind the price action rather than competing with it.

The Event Horizon (optional, off by default)

Project IB Event Horizon draws one level: the halfway mark between the favoured side's IB edge and that side's target. It follows the same Extend setting as the projected levels and is dashed by default.

It answers a question the rails alone do not. The target tells you where the move ends; the Event Horizon tells you where it is half done. Price that has crossed it has covered half the run to the favoured rail. Price that keeps stalling beneath it has not, and the run to the target still has as far to go as the run already made. Use it as a checkpoint for reassessing, not as a place to act.

On abstain sessions no Event Horizon is drawn. There is no favoured side, so there is no run to be halfway through, and drawing one would imply a lean the figures do not support.

The live ladder (while the window forms)

Before the window closes, the indicator shows something different. At each checkpoint it reports the historical frequency that the current window high or low survives as the final IB extreme. The figures sit in labels at the running high (▲) and the running low (▼), and in the table's IB forming row. Use them to judge whether the range is already set or still likely to extend. A ~ beside a figure marks a reading to treat with caution.

A ⚡ appears when a new extreme prints while the stated survival frequency was already high, meaning something historically uncommon just happened. It is a flag for awareness, not a signal and not a direction call, and it exists because knowing "that was unusual" is useful when you are watching a window form. The ladder is switched by Show forming-IB hazard (row + extreme labels), on by default.

Reading the status table

The table sits top right by default. Rows from the top, as the full table shows them:

The last line is a standing reminder that every percentage is a historical frequency from 2009 to 2026, not a prediction. A few things are worth dwelling on.

The table shows only the rows that currently mean something. IB forming appears while the window is building and leaves when it completes; Adjustments appears only if you have one switched on. A row that is absent is not a row that failed. It has nothing to say about this window, and leaving it out keeps the table short enough to read at a glance.

A tick marks the outcome once the race resolves. When a target is reached first, a ✓ appears against that row; a same-bar tie puts ◇ against both direction rows; and if the session closes with neither reached, the ✓ appears against No target hit once the session has ended. It is the same scoring the marks on the chart use, in the place you are already looking.

Compact table (mobile) keeps the glance rows: IB window, IB range, Target range, a single row for the higher-percentage side, 🧲 ATLAS MODE, Atlas [W/L] and Window strength. That single row also carries "· abstain" and "(scalps)" when they apply, and its tick reads ✓ when the higher-percentage side came first, ✗ when it did not, ◇ for a same-bar tie and ∅ when neither was reached.

No target hit is often the biggest of the three figures, especially at later starts and wider multiples. This is not a defect. Many sessions simply do not travel that far. If you are planning a trade off a target that historically goes unreached most of the time, the table is telling you so before you take it.

The hit-rate rows are the honesty check. They score the figures against what actually happened on your own loaded history. The bracketed z says how far the realised rate sits from the expected one in units of ordinary sampling noise: near zero means the tables are behaving; much beyond ±2, look closer. If the tables are well calibrated, realised and expected track each other. You are invited to watch them, and to tell us if they drift. Between August and September 2026 we also published, week by week, what Atlas IB showed at the 09:30 open and what price did next, misses included: see the NQ IB record.

The two regime tags can disagree, and that is information

The label above the box compares this IB with the last five sessions and speaks in behaviour words: 💰 ACCUMULATION, ⚖️ NORMAL, 🚀 EXPANSION. The table's IB range row compares it with seventeen years and speaks in distribution words: 💰 COMPRESSED, ⚖️ NORMAL, 🚀 EXPANDED. The vocabularies differ on purpose. They are different measurements, and identical names would invite reading one as the other. When they disagree you are looking at an IB that is large for this week but ordinary by long-run standards, or the reverse. That disagreement is real information about context, not a bug. If you switch the extension multiple to Auto, it varies with the table tag.

The scenarios table

Optional and off by default. Switched on with Show scenarios table (placed bottom right unless you change Scenarios position), a second table restates the same figures as conditional statements. One row per target gives its price, how often it was reached first and how far price has typically pushed on past it, in the form "▲ [target price] first | X% · then +N pts/60m". The run-on horizon matches the window: 30 minutes for a 30-minute IB, 60 for a 60-minute IB. The Neither row gives its share by the close.

As the session resolves, the rows latch: the target reached first reads HIT with the New York time it happened and a ✓, and the other rows grey out; a same-bar tie reads "◇ both in one bar"; a session that closes with neither reads "NEITHER ✓". Hovering over the table shows the 2h and 4h figures for both sides.

Nothing new is computed here; it is the same measured figures in a format some people find easier to act on. Conditional figures attach only to a target being reached first. There is deliberately no row that starts from any other event, because the table shows no statistic that was not measured.

Choosing a window on NQ

This section exists because the obvious way to pick a window, sorting by the highest percentage, is a trap, and NQ's version of the trap is not the one you might expect.

Every window on the NQ day was scored on how often its favoured rail was reached first, measured at the 0.5 multiple. That score, and the window's rank among the 91 windows, is the Window strength row.

Read that row with the No target hit share and the Target range row beside it, always. The highest-ranked windows are in the late afternoon, 15:00 to 16:30, and some of them print above 75%. What the score does not say is that those windows resolve only a small share of their sessions: there is very little session left, so the only races that finish at all are the ones where price was already close to a rail. A large part of that percentage describes which races got to finish, not which way they went. Costs are rarely the disqualifier on NQ: almost every window carries a target above the 15-point friction floor at some multiple. The question is not whether a window can pay for itself, but whether it resolves often enough for its lean to matter.

The New York morning, roughly 09:00 to 12:30, is where windows resolve in most sessions and carry bigger targets than the rest of the day, with favoured-rail rates mostly between about 65 and 70%.

• Overnight starts are the opposite trade-off. They resolve almost every session, but the lean is weaker, roughly 58 to 64%, and the targets, while above the floor, are smaller than the morning's, so costs weigh proportionally heavier on them.
• In the morning the 30-minute windows hold their own. 09:30 +30m sits almost level with the hour, resolving more often at smaller targets. From late morning through the afternoon the 60-minute window scores higher; overnight the two are level or the 30-minute is slightly ahead.

Two windows, 16:00 +60m and 16:30 +30m, end at the current 17:00 close, so no recent session has any time left to resolve after them. Their figures come from years when the session ran later and are historical only, and the Window strength tooltip says so. On your chart these two windows are still scored against the following session, so ignore their outcome marks and their entries in the running tallies.

None of this makes the favoured side an edge. A favoured-rail rate is a base rate. It is high largely where the favoured rail is the easier one to reach, and it says nothing about whether a trade towards it pays after costs. Window strength helps you choose a window whose races usually finish and whose figures are well supported; it does not choose trades. The Atlas-mode magnet on its default basis applies a similar screen to the session in front of you: a clear lean among the sessions that resolved, and enough sessions resolving.

Optional contextual adjustments

Three optional adjustments ship with the indicator under Contextual adjustments: off by default. Each applies only to IBs within the hours named in its setting:

• Overnight-gap context (IBs 23:30–07:30) refines the overnight figures by comparing the window's open with the previous session's close.
• Prior-day-high context (IBs 21:30–07:30) refines the overnight figures using whether the previous day's high was already in play when the window closed.
• Midday context (IBs 11:00–12:30) refines the midday figures using how the window sat against the hour before it.

An active adjustment moves the split between the two rails by a measured amount; the No target hit share is unchanged. The line under the IB box shows each active adjustment and the shift it applied, in percentage points, and the Adjustments row reads "active". When you have adjustments switched on but none of them covers this window, the row reads "on · n/a this window" and the base figures are shown unadjusted, rather than the setting silently doing nothing.

Each one cleared a deliberately high bar in our own research before it was included. They are off by default because the base figures are the product and these are refinements: look at the unadjusted figures first.

How the figures were made

Seventeen years of NQ at one-minute resolution, 2009 to 2026, cross-checked against tick data. Every 30-minute start time was measured at each duration the session allows, producing outcome figures for every supported window, plus the forming-window figures, the 2h and 4h figures and the run-on figures.

Everything the indicator displays is read from fixed tables built from that history. Nothing is re-estimated on your chart: the indicator looks up what happened, in comparable situations, in the measured history.

What matters more than the size of the dataset is what the research rejected. Getting to these figures took a long and deliberately sceptical piece of proprietary research. A large number of candidate influences on how a window resolves were considered, and each had to show that it was more than chance. Most did not, and several that looked convincing early fell away under closer checking. What you are using is what was left standing once the failures were removed, which is why the indicator has a short list of options rather than a long one: the short list is the finding. Post-target continuation belongs in that story. It resembles momentum right up until it is measured properly, at which point it resolves to drift, and it is labelled as such wherever it appears.

Before release, the figures on the chart are checked against the research data they came from.

Honest limitations

• These are frequencies, not predictions. No individual session is bound by them.
• A lean is not an edge. The favoured side is the one comparable sessions reached first more often. That frequency largely reflects how far price had to travel to each target, and it does not by itself make a trade towards it pay after costs.
• The three-outcome split is measured to the session close. Over shorter horizons the split is materially different, which is why the separate 2h and 4h figures sit on the rail badges. Do not reuse the by-close figures for a short-horizon plan.
• Sample size varies. Some windows and situations are far better supported than others. The Sample (bin) row tells you which, and you should weight your confidence accordingly.
• Measured on the history it describes. The tables were built from the same years they report on. Consistency from one era to the next is reassuring, but it is not the same as evidence from years the tables have not seen.
• NQ and MNQ only.
• Targets inside the friction band are not real targets. Below roughly 15 points, costs dominate. The indicator warns you; please believe it.
• Ties happen, where both targets are touched within the same chart bar. They are rare on a 1-minute chart, are marked ◇, and are not counted for either side.
• The tallies are small samples. The hit-rate rows count only the sessions your chart has loaded, so they move as history loads and say little over a handful of sessions.

Using it in practice

Some patterns that make sense given what the figures actually support:

• As a veto rather than a trigger. If you were about to go long into a window where the historical split leans hard the other way, or towards a target that comparable sessions mostly failed to reach, that is worth knowing before you commit.
• Respect the abstain. A grey rail means the history does not favour a side. Trading it anyway is fine, but do it knowing the trade rests on your own read alone.
• Let the Target range row decide whether the trade is worth taking. If it shows a distance that barely clears friction, reconsider the trade regardless of the percentages.
• Watch the realised-versus-expected rows over time. They are your own running audit of whether the tables are behaving on your data.
• Do not stack the run-on figures into a continuation thesis. They measure drift. If you use them, use them for target and trail geometry.

What a week with it looks like

A typical week shows you both states. On some mornings one side clears the threshold: a rail carries the magnet, there is a target at a measured distance, and the 2h and 4h figures say how quickly comparable windows tended to resolve. Whatever your entry logic is, you know which way the base rate leans and how far away the target is. On the other mornings the indicator abstains, and its contribution is the trade it gives you a reason not to force. No lean means any position rests on your own read, and standing aside is a legitimate plan; the grey rail says so plainly.

The quieter benefit is what this does to the decision itself. A few minutes after the window closes you hold the same few facts every day, lean or abstain, target distance and the share of sessions that reached neither, so the plan exists before the market has finished deciding. Less improvising mid-bar. Fewer trades taken just to be involved. And the outcome marks and tally rows keep score in the open, so you can see whether the figures are behaving.

Using it alongside Atlas Sessions

The two indicators answer different questions about the same day, and the difference is in what they measure from.

Atlas Sessions measures against the previous session, using fixed reference points. Does this session take the previous session's high, its low, both, or neither? Those levels exist before the session opens and do not move.

Atlas IB measures against the current window. Price builds its own range in the chosen window, and the targets are derived from that range. Nothing is known until the window completes.

So Sessions gives you the session frame, whether this is likely to be a session that expands and takes out levels or one that stays inside, and Atlas IB gives you the execution frame within it: specific targets, distances in points, and how often they are reached. The natural pairing is the cash open: a 09:30 +60m IB sits inside the New York morning session, so Sessions describes the session the IB window is forming in.

The important caveat runs against instinct. When the two agree, that is not two independent confirmations. Both ask which level price reaches first on the same instrument, and both are driven largely by how far price has to travel, so agreement is largely the same fact stated twice. Read agreement as coherence, not as doubled evidence, and do not size up because both agree.

Disagreement is the more informative case. If Sessions indicates a session likely to stay inside its prior range while Atlas IB projects a wide extension target, the two are telling you the target is probably too far for the day you are in, and the IB No target hit figure will usually already be saying so. If Sessions leans clearly to one side while the IB is in abstain, you have a session-level lean with no window-level structure to express it through. Both are worth knowing before you act.

The practical order: read Sessions first for the session's character, then Atlas IB for the geometry inside it. Sessions rarely changes what you do at the window level; it changes how far you expect the day to travel at all.

Tracking it yourself

The hit-rate rows score the figures on the history loaded on your chart. We also keep a public record of what Atlas IB showed at the 09:30 open and what price did next, misses and abstentions included: week by week from August to September 2026, and month by month from October 2026. See the NQ IB record.

Settings reference

Inputs as they appear in the settings dialog, group by group, with their defaults.

Atlas mode basis deserves a note. The default, Direction + decided floor, asks two things of a window before the magnet appears: the leading side's "dir" figure must clear the threshold, and at least the Min decided share of comparable sessions must have reached a rail at all. This keeps Atlas mode available across the whole day at comparable quality. Direction only drops the second condition, so you also get calls late in the afternoon. Outcome (3-way) applies the threshold to the printed three-way share, which includes neither, so in practice it fires mainly on the overnight and London windows, where the neither share is small.

Dark and light themes are supported throughout.

FAQ

Support

• support@algotradernqgc.com
• @AlgoTraderNQGC

Questions about interpreting a specific session are welcome; send a screenshot.