OPEN-SOURCE SCRIPT
已更新 Ehlers Adaptive Trend Filter + Consensus Engine v13
Ehlers Adaptive Trend Filter + Consensus Engine
WHAT IT DOES
This screener reads a universe of up to 40 tickers through three independent Ehlers filter pairs running in parallel on every name, then groups the names by how many of the three filters agree the trend is up. The grouping is the "phase"; the intensity and direction of each name's move is the "temperature". It describes the structural state of each name. It does not tell you what to trade.
Because the three filters react at different speeds, the number that agree is a measure of how broadly a trend is confirmed across timescales: a move that only the fastest filter sees is early or fragile; a move all three see is broadly confirmed.
THE THREE FILTERS
Fast (2-pole SuperSmoother): the most reactive. It turns first and catches short-cycle moves, at the cost of more false turns.
Core (3-pole + 2-pole): the balanced middle. Slower to turn, steadier once it does. In most conditions it is the strongest single read.
Adaptive (MAMA/FAMA): retunes itself to the dominant detected cycle rather than using a fixed period. It often turns before the other two at a genuine cycle change, which is why an Adaptive-only signal is treated as an early, unconfirmed one.
THE FOUR PHASES
Heated: all three filters agree the trend is up. The fullest agreement the engine measures.
Liquid: two of three agree; one dissents. The dissenter is where the interest lies.
Thawing: one of three agrees. The earliest single-timescale turn, not yet confirmed.
Frozen: none agree. No bullish consensus.
Within each phase a per-row temperature word (Boiling, Hot, Simmering, Warming, Cooling, Cold, Freezing) describes how far the name sits from its own normal range and whether that separation is rising or falling. Direction is shown by an arrow, so "Boiling" means an extreme move regardless of sign.
READING A ROW
A Heated-phase row might read:
NVDA (up) 14b C:1.8 F(up)+12.4 Co(up)+9.1 A(up)+15.2 [+6.3%] -> Heated . Boiling (up)
That is: NVDA, 63-bar direction up, 14 bars in its current phase, Calmar 1.8; all three filters bullish with their separations from baseline in percent (Fast +12.4, Core +9.1, Adaptive +15.2); stop distance to the Core baseline +6.3 percent; and the state tag, Heated . Boiling (up) - all three aligned, an extreme-magnitude move, rising.
The engine also flags "clusters": several names dissenting the same way at once (Adaptive turning bearish under apparent strength, or bullish under apparent weakness), which is structural pressure forming across the universe rather than in a single name.
HOW TO READ IT, NOT WHAT TO DO
This tool is descriptive. It reports what the filters show: which names are in which phase, how hot or cold each move is, and where broad pressure is building. It does not generate buy or sell instructions, position sizes, or price targets, and nothing in it should be read as financial advice. What you do with the information is your own decision.
CREDITS
Method: the MAMA/FAMA (MESA Adaptive Moving Average) algorithm and the SuperSmoother filters are John Ehlers' work (MAMA paper: mesasoftware.com/papers/MAMA.pdf).
Implementation: the MAMA/FAMA Pine derives from Ehlers MESA Adaptive Moving Average by LazyBear, an open-source script, refactored here from series form into a per-ticker state architecture. Credit and thanks to LazyBear for the original.
Original in this script: the multi-ticker consensus engine, the phase and temperature classification, and the cluster detection are my own work. Released open-source under MPL-2.0.
WHAT IT DOES
This screener reads a universe of up to 40 tickers through three independent Ehlers filter pairs running in parallel on every name, then groups the names by how many of the three filters agree the trend is up. The grouping is the "phase"; the intensity and direction of each name's move is the "temperature". It describes the structural state of each name. It does not tell you what to trade.
Because the three filters react at different speeds, the number that agree is a measure of how broadly a trend is confirmed across timescales: a move that only the fastest filter sees is early or fragile; a move all three see is broadly confirmed.
THE THREE FILTERS
Fast (2-pole SuperSmoother): the most reactive. It turns first and catches short-cycle moves, at the cost of more false turns.
Core (3-pole + 2-pole): the balanced middle. Slower to turn, steadier once it does. In most conditions it is the strongest single read.
Adaptive (MAMA/FAMA): retunes itself to the dominant detected cycle rather than using a fixed period. It often turns before the other two at a genuine cycle change, which is why an Adaptive-only signal is treated as an early, unconfirmed one.
THE FOUR PHASES
Heated: all three filters agree the trend is up. The fullest agreement the engine measures.
Liquid: two of three agree; one dissents. The dissenter is where the interest lies.
Thawing: one of three agrees. The earliest single-timescale turn, not yet confirmed.
Frozen: none agree. No bullish consensus.
Within each phase a per-row temperature word (Boiling, Hot, Simmering, Warming, Cooling, Cold, Freezing) describes how far the name sits from its own normal range and whether that separation is rising or falling. Direction is shown by an arrow, so "Boiling" means an extreme move regardless of sign.
READING A ROW
A Heated-phase row might read:
NVDA (up) 14b C:1.8 F(up)+12.4 Co(up)+9.1 A(up)+15.2 [+6.3%] -> Heated . Boiling (up)
That is: NVDA, 63-bar direction up, 14 bars in its current phase, Calmar 1.8; all three filters bullish with their separations from baseline in percent (Fast +12.4, Core +9.1, Adaptive +15.2); stop distance to the Core baseline +6.3 percent; and the state tag, Heated . Boiling (up) - all three aligned, an extreme-magnitude move, rising.
The engine also flags "clusters": several names dissenting the same way at once (Adaptive turning bearish under apparent strength, or bullish under apparent weakness), which is structural pressure forming across the universe rather than in a single name.
HOW TO READ IT, NOT WHAT TO DO
This tool is descriptive. It reports what the filters show: which names are in which phase, how hot or cold each move is, and where broad pressure is building. It does not generate buy or sell instructions, position sizes, or price targets, and nothing in it should be read as financial advice. What you do with the information is your own decision.
CREDITS
Method: the MAMA/FAMA (MESA Adaptive Moving Average) algorithm and the SuperSmoother filters are John Ehlers' work (MAMA paper: mesasoftware.com/papers/MAMA.pdf).
Implementation: the MAMA/FAMA Pine derives from Ehlers MESA Adaptive Moving Average by LazyBear, an open-source script, refactored here from series form into a per-ticker state architecture. Credit and thanks to LazyBear for the original.
Original in this script: the multi-ticker consensus engine, the phase and temperature classification, and the cluster detection are my own work. Released open-source under MPL-2.0.
版本注释
Changed some of the terminology words.版本注释
Two changes in this update.THEME. Refreshed the colour palette to a cooler, higher-contrast scheme (deep dark base with emerald, rose, cyan and amber accents). Colour continues to encode direction and phase; the per-row temperature word still carries intensity. No behaviour change - the same states render in clearer colours.
REMOVED THE CHART-TICKER REVERSAL OVERLAY. Took out the separate 0-100 reversal-score background and its top-panel row. It duplicated what the phase, temperature and cluster readings already describe, and a single directional score sat awkwardly against the rest of the panel, which describes structural state rather than scoring a call. The screener, the consensus engine, and all phase/temperature/cluster logic are unchanged.
This tool describes structural state. It does not issue buy or sell instructions, position sizes, or price targets, and nothing in it is financial advice.
版本注释
What's new in this update?This update makes the fast frame (4H) and the daily read check each other on their own. You no longer need to open two charts at once. It also makes the top read lines plainer.
The Resolved Read — new line at the top of the screener.
A new line compares this chart's fast read against the daily read for you, and states the result as one plain fact:
CONFIRMED UP — fast frame turning up and the daily is above its baseline. Aligned.
TURNING, NOT CONFIRMED — fast frame turning up, but the daily is still below its baseline. The structure has not turned yet.
PULLBACK — fast frame dipping, but the daily is still above baseline. A dip within the trend, not the end of it.
CONFIRMED DOWN — fast frame falling and the daily is below baseline. Aligned to the downside.
The daily is always the fixed reference. The result is always "this frame against the daily," so it does not change depending on where you start. On a daily chart it uses the daily's own values. On a 4H chart it fetches the true daily. It costs one security call.
Clearer wording across the read lines.
The old "verify on 4H" and "unverified" wording is gone. The cluster, transition, and top read lines now state the relationship directly — for example, "fast frame only; daily structure has not confirmed" — instead of asking you to check another chart yourself.
Unchanged
Filters, phases, temperature scale, cluster detection, and the four universes all work as before. The tool describes structural state. It does not give trade instructions.
Method credit and the MPL-2.0 licence stay the same.
开源脚本
秉承TradingView的精神，该脚本的作者将其开源，以便交易者可以查看和验证其功能。向作者致敬！您可以免费使用该脚本，但请记住，重新发布代码须遵守我们的网站规则。
免责声明
这些信息和出版物并非旨在提供，也不构成TradingView提供或认可的任何形式的财务、投资、交易或其他类型的建议或推荐。请阅读使用条款了解更多信息。
开源脚本
秉承TradingView的精神，该脚本的作者将其开源，以便交易者可以查看和验证其功能。向作者致敬！您可以免费使用该脚本，但请记住，重新发布代码须遵守我们的网站规则。
免责声明
这些信息和出版物并非旨在提供，也不构成TradingView提供或认可的任何形式的财务、投资、交易或其他类型的建议或推荐。请阅读使用条款了解更多信息。