OPEN-SOURCE SCRIPT
Confluence Engine+ (M1D)
Confluence Engine+
Maps the PD arrays taught by ICT. Instead of plotting a dozen isolated objects and leaving you to weigh them by eye, it reads the confluences present at price — a liquidity raid into a discount PD array, confirmed by a breaker, with SMT, inside a killzone — into an ICT directional bias, the current draw on liquidity, and a corner dashboard. It reports context. It is not a signal generator: it does not fire buy or sell orders and it does not place entries or exits.
What it does
Six modules, each toggleable, built so later stages read the state the earlier ones capture.
1 · Killzones & Sessions. Time-gates the London, NY AM and NY PM killzones and marks the Asia, London and New York session highs and lows as unmitigated levels, plus the 00:00 New York Midnight Open — a core daily reference (below it leans the day bullish, above it bearish). Every time window here — the killzones included — is resolved on the session source timeframe rather than the chart's. A killzone is ninety minutes to three hours, narrower than a single higher-timeframe candle, so judged off the chart a candle merely overlapping one would report as fully inside it. The killzone is read at the candle's close, and the Midnight Open the same way, so it still resolves on charts whose own candles never open at 00:00.
2 · Structure & Dealing Range. Pivot highs and lows define the swing structure the arrays build on. The dealing range — the window whose midpoint separates premium from discount — is taken from a fixed higher-timeframe period: the weekly range on 1-hour-and-up charts, the daily range on anything intraday below that.
3 · Liquidity.
Session highs and lows (Asia, London, New York) plus prior day and prior week highs and lows are drawn as reference liquidity, each anchored to the candle that formed it. Session extremes are measured on a lower timeframe rather than the chart's: a three-hour London window is shorter than a single 4-hour candle, so read off the chart it would collapse to the high of whichever candle happened to contain it. Sourcing them lower keeps session levels correct on any chart, and charts already at or below that timeframe track natively.
A level tracks the right edge while it rests; the instant price touches it, it is mitigated — the line stops extending at that candle and turns dotted — so taken liquidity stays readable as history and can never be mistaken for a live level. Where levels land close together only the most significant is drawn, a weekly level outranking a prior-day level, which outranks a session level, so near-duplicates never stack. Each level then retires once it ages past its lookback window: the "back" settings are a window in days and weeks, so nothing lingers long after the session or period that formed it.
4 · PD Arrays.
Fair Value Gaps (BISI / SIBI), Volume Imbalances, Order Blocks (the candle body before a displacement) and the Order-Block-to-Breaker lifecycle. A FVG registers only when the gap clears a height floor and its middle candle is a genuine displacement candle, so routine three-bar gaps are filtered out. A Volume Imbalance is the FVG's thinner cousin — a two-candle gap between the bodies that a wick still trades through, so the only volume in the gap changed hands in wicks; it carries its own sensitivity floor and lives under the same lifecycle as the FVGs. A new gap also removes any stale opposite-direction gap it overlaps: that range has since been delivered through the other way, so the old one-sided imbalance cannot stand. An Order Block must earn its place with the full ICT sequence: the close that breaks the prior swing — a structure break grants that credit exactly once, so blocks sit at real breaks rather than printing mid-trend — and the leg must leave a Fair Value Gap behind it. The gap is the displacement evidence: a leg that never gaps did not really displace, and its block never registers. Only the order blocks that matter make the chart. A percentage-fill mitigation decider governs each array: once price trades a chosen depth into it from the side it is approached from (default 50%, consequent encroachment), the array is mitigated — either faded and kept (dotted, faint, check mark) or removed, whichever you set. A largest-array-wins declutter keeps overlapping zones from stacking.
5 · SMT Divergence.
A liquidity-sweep read against a correlated symbol — auto-paired (NQ to ES, ES to NQ, YM to ES, GC to SI, and their micros) or a symbol you set. When your chart takes a swing level but the peer holds its aligned level and refuses to follow, the move lacks participation: a SMT is drawn from the swept level to the sweep. Pooled swings age out after a set number of bars, so a divergence is only ever drawn between swings that were still contemporaries — never between two that are days apart.
6 · Dashboard.
A pure confluence read-out of the state each module captures — the ICT bias, and the current draw on liquidity: the nearest unmitigated high above (the resting buyside) and the nearest unmitigated low below (the resting sellside), each named with its distance — where price is being drawn to, read straight from the level engine. When liquidity is taken, the sweep row names the level that was raided — PDH, PDL, a session high or low — rather than a generic flag. Below that, whichever confluences are live (a OB or Breaker tap, a SMT), the killzone, the Midnight Open and the dealing-range position — every row a decision input, nothing that is merely inventory. It reports context; the trade is left to you, and nothing fires.
Visual grammar
A live PD-array zone is solid, its label centred on the zone and travelling with it as it extends toward price. Once a zone is mitigated to the chosen fill depth it turns dotted, fades, and its name folds into the zone itself — a faint "✓ name" carried inside the frozen box, quiet by design, so worked zones read as history at a glance without any label cluttering the chart. A liquidity level extends while it rests and freezes into a dotted line the moment it is taken — full-strength black by default, with an optional dim. Purple marks bullish arrays, magenta bearish; liquidity and levels are neutral. Nearby level labels merge so the chart stays readable, and objects project a few bars past the last candle so labels sit in clear space, never on price.
Method & repainting
Every detection path — liquidity capture, PD-array formation, mitigation and SMT — evaluates only on closed bars, so nothing is drawn, moved or removed on the strength of an unfinished candle. Once a level, zone or SMT line is on the chart it stays where it was placed. Completed period levels fix at the rollover, and session extremes are read from the closed intrabars of the session source timeframe.
Two things update live, by design. The dashboard reads current price, so the bias and draw rows move during the forming candle and settle at its close; and the right-edge labels re-merge as levels are added or taken. Neither creates or moves a drawn object.
Swing-based features
— The order-block structure break and SMT — depend on pivots, which confirm a set number of bars after the swing itself forms. That is a fixed delay, not a revision: a pivot never moves once printed. Session levels also rely on lower-timeframe data, which is available for a limited span of recent history, so they thin out far back on the chart.
Settings
Session timezone, session windows and the session source timeframe, per-array sensitivities and the mitigation decider, the SMT peer and liquidity memory, and full dashboard controls are all exposed as inputs.
This is a decision-support tool for discretionary ICT trading. It is not financial advice, and no market's past behaviour is indicative of future results.
Maps the PD arrays taught by ICT. Instead of plotting a dozen isolated objects and leaving you to weigh them by eye, it reads the confluences present at price — a liquidity raid into a discount PD array, confirmed by a breaker, with SMT, inside a killzone — into an ICT directional bias, the current draw on liquidity, and a corner dashboard. It reports context. It is not a signal generator: it does not fire buy or sell orders and it does not place entries or exits.
What it does
Six modules, each toggleable, built so later stages read the state the earlier ones capture.
1 · Killzones & Sessions. Time-gates the London, NY AM and NY PM killzones and marks the Asia, London and New York session highs and lows as unmitigated levels, plus the 00:00 New York Midnight Open — a core daily reference (below it leans the day bullish, above it bearish). Every time window here — the killzones included — is resolved on the session source timeframe rather than the chart's. A killzone is ninety minutes to three hours, narrower than a single higher-timeframe candle, so judged off the chart a candle merely overlapping one would report as fully inside it. The killzone is read at the candle's close, and the Midnight Open the same way, so it still resolves on charts whose own candles never open at 00:00.
2 · Structure & Dealing Range. Pivot highs and lows define the swing structure the arrays build on. The dealing range — the window whose midpoint separates premium from discount — is taken from a fixed higher-timeframe period: the weekly range on 1-hour-and-up charts, the daily range on anything intraday below that.
3 · Liquidity.
Session highs and lows (Asia, London, New York) plus prior day and prior week highs and lows are drawn as reference liquidity, each anchored to the candle that formed it. Session extremes are measured on a lower timeframe rather than the chart's: a three-hour London window is shorter than a single 4-hour candle, so read off the chart it would collapse to the high of whichever candle happened to contain it. Sourcing them lower keeps session levels correct on any chart, and charts already at or below that timeframe track natively.
A level tracks the right edge while it rests; the instant price touches it, it is mitigated — the line stops extending at that candle and turns dotted — so taken liquidity stays readable as history and can never be mistaken for a live level. Where levels land close together only the most significant is drawn, a weekly level outranking a prior-day level, which outranks a session level, so near-duplicates never stack. Each level then retires once it ages past its lookback window: the "back" settings are a window in days and weeks, so nothing lingers long after the session or period that formed it.
4 · PD Arrays.
Fair Value Gaps (BISI / SIBI), Volume Imbalances, Order Blocks (the candle body before a displacement) and the Order-Block-to-Breaker lifecycle. A FVG registers only when the gap clears a height floor and its middle candle is a genuine displacement candle, so routine three-bar gaps are filtered out. A Volume Imbalance is the FVG's thinner cousin — a two-candle gap between the bodies that a wick still trades through, so the only volume in the gap changed hands in wicks; it carries its own sensitivity floor and lives under the same lifecycle as the FVGs. A new gap also removes any stale opposite-direction gap it overlaps: that range has since been delivered through the other way, so the old one-sided imbalance cannot stand. An Order Block must earn its place with the full ICT sequence: the close that breaks the prior swing — a structure break grants that credit exactly once, so blocks sit at real breaks rather than printing mid-trend — and the leg must leave a Fair Value Gap behind it. The gap is the displacement evidence: a leg that never gaps did not really displace, and its block never registers. Only the order blocks that matter make the chart. A percentage-fill mitigation decider governs each array: once price trades a chosen depth into it from the side it is approached from (default 50%, consequent encroachment), the array is mitigated — either faded and kept (dotted, faint, check mark) or removed, whichever you set. A largest-array-wins declutter keeps overlapping zones from stacking.
5 · SMT Divergence.
A liquidity-sweep read against a correlated symbol — auto-paired (NQ to ES, ES to NQ, YM to ES, GC to SI, and their micros) or a symbol you set. When your chart takes a swing level but the peer holds its aligned level and refuses to follow, the move lacks participation: a SMT is drawn from the swept level to the sweep. Pooled swings age out after a set number of bars, so a divergence is only ever drawn between swings that were still contemporaries — never between two that are days apart.
6 · Dashboard.
A pure confluence read-out of the state each module captures — the ICT bias, and the current draw on liquidity: the nearest unmitigated high above (the resting buyside) and the nearest unmitigated low below (the resting sellside), each named with its distance — where price is being drawn to, read straight from the level engine. When liquidity is taken, the sweep row names the level that was raided — PDH, PDL, a session high or low — rather than a generic flag. Below that, whichever confluences are live (a OB or Breaker tap, a SMT), the killzone, the Midnight Open and the dealing-range position — every row a decision input, nothing that is merely inventory. It reports context; the trade is left to you, and nothing fires.
Visual grammar
A live PD-array zone is solid, its label centred on the zone and travelling with it as it extends toward price. Once a zone is mitigated to the chosen fill depth it turns dotted, fades, and its name folds into the zone itself — a faint "✓ name" carried inside the frozen box, quiet by design, so worked zones read as history at a glance without any label cluttering the chart. A liquidity level extends while it rests and freezes into a dotted line the moment it is taken — full-strength black by default, with an optional dim. Purple marks bullish arrays, magenta bearish; liquidity and levels are neutral. Nearby level labels merge so the chart stays readable, and objects project a few bars past the last candle so labels sit in clear space, never on price.
Method & repainting
Every detection path — liquidity capture, PD-array formation, mitigation and SMT — evaluates only on closed bars, so nothing is drawn, moved or removed on the strength of an unfinished candle. Once a level, zone or SMT line is on the chart it stays where it was placed. Completed period levels fix at the rollover, and session extremes are read from the closed intrabars of the session source timeframe.
Two things update live, by design. The dashboard reads current price, so the bias and draw rows move during the forming candle and settle at its close; and the right-edge labels re-merge as levels are added or taken. Neither creates or moves a drawn object.
Swing-based features
— The order-block structure break and SMT — depend on pivots, which confirm a set number of bars after the swing itself forms. That is a fixed delay, not a revision: a pivot never moves once printed. Session levels also rely on lower-timeframe data, which is available for a limited span of recent history, so they thin out far back on the chart.
Settings
Session timezone, session windows and the session source timeframe, per-array sensitivities and the mitigation decider, the SMT peer and liquidity memory, and full dashboard controls are all exposed as inputs.
This is a decision-support tool for discretionary ICT trading. It is not financial advice, and no market's past behaviour is indicative of future results.
開源腳本
秉持TradingView一貫精神，這個腳本的創作者將其設為開源，以便交易者檢視並驗證其功能。向作者致敬！您可以免費使用此腳本，但請注意，重新發佈代碼需遵守我們的社群規範。
免責聲明
這些資訊和出版物並非旨在提供，也不構成TradingView提供或認可的任何形式的財務、投資、交易或其他類型的建議或推薦。請閱讀使用條款以了解更多資訊。
免責聲明
這些資訊和出版物並非旨在提供，也不構成TradingView提供或認可的任何形式的財務、投資、交易或其他類型的建議或推薦。請閱讀使用條款以了解更多資訊。