OPEN-SOURCE SCRIPT
已更新 Probabilistic Projection
Most projection tools extend the recent trend or calculate a single average return. Probabilistic Projection uses a more comprehensive approach: it examines the full range of returns that followed comparable historical conditions.
The method can be understood through one question:
When this market had a similar trend and level of volatility in the past, what happened over the following 6 months, 1 year or 2 years?
The indicator collects these historical forward returns, builds a complete probability distribution, and displays:
• The median historical outcome
• A configurable central range of outcomes
• The quality of the available sample
Similar historical conditions receive additional weight, but they are never allowed to dominate the model. At least 65% of the calculation always comes from the complete historical distribution.
This balance is important. Using only a few similar periods can produce unstable conclusions, while using only a long-term average ignores the current market environment. Probabilistic Projection combines both sources of information while automatically reducing the influence of unreliable samples.
WHY THIS APPROACH IS ROBUST
The model is designed to reduce several common sources of error:
• It evaluates a distribution of outcomes instead of producing one deterministic target.
• It uses confirmed daily data rather than incomplete higher-timeframe values.
• It remains independent of the chart timeframe.
• It limits the influence of trend and volatility conditioning.
• It gives less weight to conditional information when the sample is insufficient.
• It pulls unstable estimates back toward the global historical median.
• It widens the projection range when data quality is lower.
• It distinguishes sample quality from prediction accuracy.
• It uses standard market prices, including on non-standard charts.
A one-year projection therefore remains approximately one year whether the indicator is viewed on a 5-minute, daily or weekly chart.
HOW THE DISTRIBUTION IS BUILT
Three projection horizons are available:
• 6 months: 126 trading days
• 1 year: 252 trading days
• 2 years: 504 trading days
For each monthly projection step, the script measures what the market actually returned over the same forward horizon at different points in its history.
Historical sample endpoints are separated by approximately 21 trading days. The script can use up to 225 observations at each step, depending on the available history.
The final distribution combines three components:
1. Complete historical distribution
Every valid observation contributes to the baseline. This ensures that the model retains information from different market environments, including periods that do not resemble the current one.
2. Trend context
The latest confirmed daily close is compared with a 202-day simple moving average.
Historical observations that began in the same broad trend state can receive additional weight. Trend conditioning cannot contribute more than 12% of the total calculation.
3. Volatility context
The model calculates realized volatility from 21-day daily returns and ranks it over approximately three years.
Historical periods with a similar volatility rank can also receive additional weight. Volatility conditioning cannot contribute more than 28%.
The combined conditional contribution is capped at 35%. The complete historical distribution always retains at least 65% of the total weight.
If too few comparable observations are available, the corresponding trend or volatility weight is automatically reduced or disabled.
READING THE PROJECTION
Solid central line
The solid line represents the median of the reliability-adjusted distribution.
Because expected returns are less stable than return dispersion, the conditioned median is partially pulled back toward the global historical median. This limits the influence of temporary or poorly represented market conditions.
Dashed boundaries
The dashed lines represent a configurable central range:
• 60%: P20 to P80
• 70%: P15 to P85 (default)
• 80%: P10 to P90
The 60% setting focuses on more central outcomes. The 80% setting includes a broader range of historical scenarios.
When data quality is lower, the distance between the median and the boundaries is expanded. The projection becomes more cautious rather than displaying the same apparent precision with a weaker sample.
Shaded range
The optional shaded area makes the distribution easier to read. It represents a percentile-derived range of historical scenarios, not a confidence interval or guaranteed future coverage.
Each point is calculated from the return distribution corresponding to its own horizon. The connected lines help visualize how the median and dispersion evolve over time; they are not individual simulated market paths.
INFORMATION PANEL
Annualized median projection
The median return at the selected terminal horizon is converted into an annualized percentage. This makes the six-month, one-year and two-year projections easier to compare.
It describes the center of the modeled historical distribution. It is not an expected return or a performance promise.
Data quality
The data-quality score evaluates whether enough historical information is available. It considers:
• The number of valid observations
• The overlap between long-horizon observations
• An estimate of the effective independent sample size
• The minimum sample required to display the projection
This score measures the strength of the historical sample, not the probability that the projection will be correct.
CONFIRMED DATA AND LIVE PRICES
The statistical model uses completed daily observations. The optional regime channel also uses confirmed source-timeframe values.
The projected returns are rescaled to the live standard-market close. The displayed price levels can therefore move while the current chart bar is open, but the underlying historical distribution remains based on confirmed data.
On Heikin Ashi and other non-standard charts, the model uses the standard instrument rather than synthetic candle prices.
OPTIONAL REGIME CHANNEL
The optional regime channel provides separate higher-timeframe context and does not alter the probabilistic projection.
It evaluates:
• Price relative to fast and slow moving averages
• Momentum
• Direction of the slow moving average
• Percentage distance from the slow average
• ATR-based expansion
Its boundaries use the larger of a minimum percentage distance or an ATR-based distance. Optional smoothing changes only the visual transition between confirmed levels.
INTENDED USE
Probabilistic Projection is designed primarily for medium- and long-term analysis of broad equity indices.
It can help users:
• See the range of outcomes that historically followed comparable conditions
• Compare potential upside and downside asymmetry
• Understand how uncertainty expands with the projection horizon
• Identify when the historical evidence is limited
• Add a statistical framework to portfolio and risk decisions
No historical method can know the future. Market distributions change, long-horizon samples overlap, and unprecedented events cannot be represented.
Probabilistic Projection is a descriptive probability model, not a price target, trading signal or standalone trading system.
The method can be understood through one question:
When this market had a similar trend and level of volatility in the past, what happened over the following 6 months, 1 year or 2 years?
The indicator collects these historical forward returns, builds a complete probability distribution, and displays:
• The median historical outcome
• A configurable central range of outcomes
• The quality of the available sample
Similar historical conditions receive additional weight, but they are never allowed to dominate the model. At least 65% of the calculation always comes from the complete historical distribution.
This balance is important. Using only a few similar periods can produce unstable conclusions, while using only a long-term average ignores the current market environment. Probabilistic Projection combines both sources of information while automatically reducing the influence of unreliable samples.
WHY THIS APPROACH IS ROBUST
The model is designed to reduce several common sources of error:
• It evaluates a distribution of outcomes instead of producing one deterministic target.
• It uses confirmed daily data rather than incomplete higher-timeframe values.
• It remains independent of the chart timeframe.
• It limits the influence of trend and volatility conditioning.
• It gives less weight to conditional information when the sample is insufficient.
• It pulls unstable estimates back toward the global historical median.
• It widens the projection range when data quality is lower.
• It distinguishes sample quality from prediction accuracy.
• It uses standard market prices, including on non-standard charts.
A one-year projection therefore remains approximately one year whether the indicator is viewed on a 5-minute, daily or weekly chart.
HOW THE DISTRIBUTION IS BUILT
Three projection horizons are available:
• 6 months: 126 trading days
• 1 year: 252 trading days
• 2 years: 504 trading days
For each monthly projection step, the script measures what the market actually returned over the same forward horizon at different points in its history.
Historical sample endpoints are separated by approximately 21 trading days. The script can use up to 225 observations at each step, depending on the available history.
The final distribution combines three components:
1. Complete historical distribution
Every valid observation contributes to the baseline. This ensures that the model retains information from different market environments, including periods that do not resemble the current one.
2. Trend context
The latest confirmed daily close is compared with a 202-day simple moving average.
Historical observations that began in the same broad trend state can receive additional weight. Trend conditioning cannot contribute more than 12% of the total calculation.
3. Volatility context
The model calculates realized volatility from 21-day daily returns and ranks it over approximately three years.
Historical periods with a similar volatility rank can also receive additional weight. Volatility conditioning cannot contribute more than 28%.
The combined conditional contribution is capped at 35%. The complete historical distribution always retains at least 65% of the total weight.
If too few comparable observations are available, the corresponding trend or volatility weight is automatically reduced or disabled.
READING THE PROJECTION
Solid central line
The solid line represents the median of the reliability-adjusted distribution.
Because expected returns are less stable than return dispersion, the conditioned median is partially pulled back toward the global historical median. This limits the influence of temporary or poorly represented market conditions.
Dashed boundaries
The dashed lines represent a configurable central range:
• 60%: P20 to P80
• 70%: P15 to P85 (default)
• 80%: P10 to P90
The 60% setting focuses on more central outcomes. The 80% setting includes a broader range of historical scenarios.
When data quality is lower, the distance between the median and the boundaries is expanded. The projection becomes more cautious rather than displaying the same apparent precision with a weaker sample.
Shaded range
The optional shaded area makes the distribution easier to read. It represents a percentile-derived range of historical scenarios, not a confidence interval or guaranteed future coverage.
Each point is calculated from the return distribution corresponding to its own horizon. The connected lines help visualize how the median and dispersion evolve over time; they are not individual simulated market paths.
INFORMATION PANEL
Annualized median projection
The median return at the selected terminal horizon is converted into an annualized percentage. This makes the six-month, one-year and two-year projections easier to compare.
It describes the center of the modeled historical distribution. It is not an expected return or a performance promise.
Data quality
The data-quality score evaluates whether enough historical information is available. It considers:
• The number of valid observations
• The overlap between long-horizon observations
• An estimate of the effective independent sample size
• The minimum sample required to display the projection
This score measures the strength of the historical sample, not the probability that the projection will be correct.
CONFIRMED DATA AND LIVE PRICES
The statistical model uses completed daily observations. The optional regime channel also uses confirmed source-timeframe values.
The projected returns are rescaled to the live standard-market close. The displayed price levels can therefore move while the current chart bar is open, but the underlying historical distribution remains based on confirmed data.
On Heikin Ashi and other non-standard charts, the model uses the standard instrument rather than synthetic candle prices.
OPTIONAL REGIME CHANNEL
The optional regime channel provides separate higher-timeframe context and does not alter the probabilistic projection.
It evaluates:
• Price relative to fast and slow moving averages
• Momentum
• Direction of the slow moving average
• Percentage distance from the slow average
• ATR-based expansion
Its boundaries use the larger of a minimum percentage distance or an ATR-based distance. Optional smoothing changes only the visual transition between confirmed levels.
INTENDED USE
Probabilistic Projection is designed primarily for medium- and long-term analysis of broad equity indices.
It can help users:
• See the range of outcomes that historically followed comparable conditions
• Compare potential upside and downside asymmetry
• Understand how uncertainty expands with the projection horizon
• Identify when the historical evidence is limited
• Add a statistical framework to portfolio and risk decisions
No historical method can know the future. Market distributions change, long-horizon samples overlap, and unprecedented events cannot be represented.
Probabilistic Projection is a descriptive probability model, not a price target, trading signal or standalone trading system.
發行說明
Minor Update發行說明
Minor Update發行說明
Minor Update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。
Plan the trade ⚡ Trade the plan
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。
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。
Plan the trade ⚡ Trade the plan
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。