A method and system for quantifying and deducing the SCE gravitational market trend of fused macro-meso-high-frequency data
Patent Information
- Application Number
- CN202611143588.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-07-30
- Publication Date
- 2026-09-18
AI Technical Summary
其中，绝对水平分采用 60 个月滚动窗口百分位法计算，将当前值置于滚动历史序列中做百分位排名，线性映射至目标区间，解决数据分布随时间漂移的问题；趋势动能分通过近 5 期数据的线性回归斜率除以历史波动率得到，消除不同指标的波动差异；预期差分基于 EMA 自适应预期计算实际值与预期值的偏离度，捕捉超预期变化
实现异构指标的量纲统一：通过 “水平 - 动能 - 预期” 三维标准化引擎，将不同单位、不同量级、不同波动特性的宏观、产业、资金指标统一映射至 - 100~+100 的可比区间，从根本上解决了跨维度加权失真的技术问题，且滚动窗口机制保障了评分体系的时间稳定性。
Smart Images
Figure CN122779955A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of financial data processing and quantitative analysis technology, and in particular to a standardized aggregation processing method for multi-source heterogeneous economic indicators, specifically a sector driving force scoring method based on macro-industry-capital three-dimensional aggregation. Background Technology
[0002] With the increasing demand for quantitative investment research and industry monitoring, the quantitative assessment of the driving forces of industry sectors has become a core requirement in the field of financial data analysis. Existing sector scoring and driving force analysis technologies generally suffer from the following technical shortcomings: The heterogeneous dimensions of multi-source data make direct aggregation impossible: Macroeconomic indicators, industrial chain indicators, and market capital indicators vary greatly in units, magnitudes, and volatility characteristics, making it difficult for traditional methods to perform weighted calculations on a unified scale, leading to distortions in cross-dimensional comparisons. Static and fixed weighting systems result in poor adaptability: Existing multi-factor scoring systems mostly use fixed weights, failing to dynamically adjust the influence weights of each dimension with changes in the macroeconomic cycle, significantly reducing the effectiveness of scoring during economic cycle transitions. Insufficient ability to capture marginal changes: Existing methods only focus on the absolute level and first-order trend of indicators, neglecting trend acceleration and expected difference dimensions, failing to identify marginal inflection points of driving forces in advance, exhibiting significant lag. Lack of cross-validation with entity data: Most scoring systems rely solely on closed-loop calculations based on financial market data, making them susceptible to interference from market sentiment and unable to be objectively validated using high-frequency data from the real economy, posing a risk of "data idling." In summary, existing technologies cannot simultaneously meet the multiple requirements of "unified dimensions, cycle adaptability, marginal sensitivity, and entity cross-validation," necessitating a new type of sector-driven force-based scoring system. Summary of the Invention
[0003] The purpose of this invention is to overcome the shortcomings of existing technologies and provide a sector-driven force scoring method based on a three-dimensional aggregation of macro-industry-capital. This method achieves dimensional unification of heterogeneous indicators through a three-dimensional standardization engine, combines macro narrative adjustment and dynamic weighting mechanisms to achieve periodic adaptation, and introduces physical data shadow lines for cross-validation. This improves the accuracy, stability, and forward-looking nature of sector-driven force assessment from a technical perspective. To achieve the above objectives, this invention adopts the following technical solution: A sector-driven force scoring method based on a three-dimensional aggregation of macroeconomics, industry, and capital includes the following steps: The original indicators undergo three-dimensional standardization: For each input economic and financial indicator, sub-scores are calculated from three dimensions: absolute level, trend momentum, and expectation difference. These scores are then weighted and aggregated according to preset weights to obtain the SCE standardized score for each indicator, with a uniform score range of -100 to +100. Specifically, the absolute level score is calculated using a 60-month rolling window percentile method, placing the current value in the rolling historical sequence for percentile ranking and linearly mapping it to the target range, thus addressing the issue of data distribution drift over time. The trend momentum score is obtained by dividing the linear regression slope of the last 5 periods of data by the historical volatility, eliminating fluctuation differences between different indicators. The expectation difference is calculated based on EMA adaptive expectation to determine the deviation between the actual and expected values, capturing unexpected changes. When the expectation difference cannot be calculated, its weights are automatically distributed proportionally to the other two dimensions. Three-axis dimensional classification and initial score calculation: All standardized indicators are categorized into three dimensions based on data attributes: the macro gravity axis (X-axis), the industry gravity axis (Y-axis), and the capital gravity axis (Z-axis). The average score of the indicators within each dimension is taken as the initial score for that dimension. The initial score for the industry axis needs to be corrected by the inventory warning coefficient, and the score is adjusted in the same direction based on the historical percentile of finished goods inventory.
[0004] Macro narrative weight adaptation: The Bayesian inference engine identifies the state of two core variables, growth and inflation, and outputs the current macro narrative state (stagflation / recovery / overheating / recession). It calls the preset dimension adjustment coefficient matrix to dynamically correct the basic weights of the three dimensions, so that the scoring system adapts to the driving logic of the current economic cycle.
[0005] Dynamic weight adjustment: Calculate the second difference of the initial score of each dimension, i.e. the acceleration of change. After standardizing the absolute value of the acceleration, map it to the coefficient range of 0.8~1.2. The dimension with the greater acceleration has a higher weight, which strengthens the influence weight of the driving factor with the fastest marginal change.
[0006] Multi-dimensional aggregated output: Combining the narrative-adjusted dimensional weights and dynamic weight coefficients, the initial scores of the three axes are weighted, summed, and normalized to obtain the final total SCE driving force score for the segment, with a value range of -100 to +100.
[0007] Optional steps: Overlay of physical shadow lines: Select high-frequency physical quantity indicators such as copper, crude steel, cement, and diesel, synthesize the physical pulse shadow index through the same SCE standardization process, and overlay it into the sector price series as a cross-validation layer of the real economy dimension to identify false trends driven purely by funds. Beneficial effects
[0008] Compared with the prior art, the present invention has the following beneficial technical effects: Achieving dimensional unification of heterogeneous indicators: Through the "level-momentum-expectation" three-dimensional standardization engine, macroeconomic, industrial, and financial indicators with different units, magnitudes, and fluctuation characteristics are uniformly mapped to a comparable range of -100 to +100, fundamentally solving the technical problem of cross-dimensional weighting distortion, and the rolling window mechanism ensures the time stability of the scoring system.
[0009] It has the ability to adapt to macroeconomic cycles: By recognizing and adjusting the coefficient matrix through Bayesian narrative, the system can automatically switch the dimension weights according to the state of the economic cycle. For example, during the stagflation period, the weight of the money aspect is reduced and the weight of the macro aspect is strengthened, while during the recovery period, the weight of the industry aspect is strengthened, which greatly improves the effectiveness of the scoring under different market conditions.
[0010] Improve the sensitivity of marginal inflection point identification: By introducing the dimensions of trend acceleration and expected difference, we break through the limitations of traditional methods that only focus on absolute values and first-order trends, and can capture the marginal changes of driving forces earlier, thus improving the foresight of the score.
[0011] Constructing a cross-validation mechanism between the real economy and finance: The introduction of the shadow line of the real economy's pulse enables the scoring of pure financial data to be verified by high-frequency data of the real economy, effectively filtering out false signals caused by capital sentiment and improving the objectivity and credibility of the scoring results. Attached Figure Description
[0012] Figure 1 is a flowchart of the SCE 3D standardization engine described in this invention; Figure 2 is a schematic diagram of the architecture of the three-axis classification and narrative weight adjustment described in this invention; Figure 3 is an example diagram of the output of the total SCE driving force score for the 18 industry sectors described in this invention; Figure 4 is a schematic diagram showing the effect of superimposing the physical pulse shadow line and the sector price as described in this invention. Detailed Implementation
[0013] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
[0014] Example 1: Standardized Calculation of Single-Indicator SCE Taking the "Year-on-Year Growth Rate of Outstanding Social Financing" indicator as an example, the specific implementation of step S1 is as follows: Absolute Level Score Calculation: Retrieve the monthly data of the indicator for the past 60 months, sort the values from smallest to largest, calculate the percentile ranking of the current value as the 30th percentile, and calculate the absolute level score as -40 using the linear mapping formula: score = percentile × 2 - 100. Trend Momentum Score Calculation: Perform linear regression on the data from the past 5 months, obtaining a slope of 0.3. Calculate the historical volatility of the slope of the indicator over the past 36 months as 1.5, and standardize the momentum value to 0.2. Mapping to the -100~+100 range yields a trend momentum score of +20. Expected Difference Calculation: Construct an adaptive expected sequence using a 12-period EMA. The current actual value is 1 percentage point lower than the expected value. Dividing this by the historical volatility of 0.8 yields a standardized expected difference of -1.25. Mapping to the -30~+30 range yields an expected difference score of -25. Three-dimensional aggregation: Calculated using the default weighting of 0.5:0.4:0.1: -40×0.5 + 20×0.4 + (-25)×0.1 = -20 + 8 - 2.5 = -14.5, meaning the SCE standardized score for this indicator is -14.5. If the expected difference for the current period cannot be calculated due to missing data, its 0.1 weight is proportionally allocated to the other two items, adjusting the weighting to 0.55:0.45:0, and the weighting calculation is performed again.
[0015] Example 2: Three-axis dimensional classification and weight adjustment Taking the new energy sector as an example, a total of 12 indicators were included in the calculation, among which: Macroeconomic gravity axis (4 items): social financing growth rate, PMI, PPI, industrial added value, calculated average initial score X=12.3; Industry gravity axis (5 items): photovoltaic installation growth rate, power battery production, lithium price year-on-year growth rate, industrial robot production, and finished product inventory percentile. The initial score after correction by the inventory warning coefficient is Y_adj=35.7. Capital Gravity Axis (3 items): Sector turnover, Northbound Capital holding ratio, and price momentum, with an average initial score Z=22.1. The Bayesian engine identifies the current macroeconomic narrative as "recovery," and adjusts the coefficient matrix as follows: X-axis coefficient 1.0, Y-axis coefficient 1.2, Z-axis coefficient 1.0. The default base weights are w_x:w_y:w_z=0.4:0.35:0.25. The acceleration over the last 3 periods for each dimension is then calculated, resulting in dynamic weight coefficients k_x=1.0, k_y=1.1, and k_z=0.9. Finally, the SCE total score is calculated using the formula.
[0016] Example 3: Construction of a physical pulse shadow line Four high-frequency physical indicators—refined copper production, crude steel production, cement clinker inventory ratio, and diesel operating rate—were selected. Each indicator underwent SCE standardization in step S1, and then a physical pulse shadow index was synthesized using equal weighting, with a value range of -100 to +100. This index was overlaid on the same coordinate axis as the sector price index. When the two trends diverged, it indicated that the price trend lacked support from the real economy and there was a high risk of a correction.
[0017] Those skilled in the art will recognize that the present invention can be implemented through a combination of software programs and hardware platforms. The software programs are stored in computer-readable storage media, including but not limited to ROM, RAM, disks, optical discs, etc.
Claims
1. A sector-driven force scoring method based on a three-dimensional aggregation of macro-industry-capital, characterized in that, Includes the following steps: S1 Original Indicator Three-Dimensional Standardization Processing: For each input original economic and financial indicator, calculate the absolute level score, trend momentum score, and expected difference score respectively, and aggregate them according to preset weights to obtain the SCE standardized score of the single indicator. The value range of the SCE standardized score is -100 to +100. S2 Three-Axis Dimension Classification and Initial Score Calculation: All SCE standardized scores are classified into three independent dimensions—Macroeconomic Gravity Axis, Industry Gravity Axis, and Capital Gravity Axis—according to the data source attributes of the indicators, and the initial dimension scores of each dimension are calculated. S3 Macro Narrative Weight Adaptation: Based on the current macro narrative state output by the Bayesian inference engine, the preset dimension adjustment coefficient matrix is called to correct the basic weights of the three dimensions and obtain dimension weights that are adapted to the current economic cycle. S4 Dynamic Weight Correction: Calculate the dynamic weight coefficients for each dimension based on the historical acceleration of the initial scores of each dimension. The values of the dynamic weight coefficients range from 0.8 to 1.
2. S5 Multi-Dimensional Aggregation Output: Combining the corrected dimension weights and dynamic weight coefficients, the initial scores of the three dimensions are weighted and calculated to obtain the final SCE driving force total score.
2. The sector-driven force scoring method based on macro-industry-capital three-dimensional aggregation as described in claim 1, characterized in that, In step S1: The absolute level score is calculated using the rolling window percentile method. The historical sequence of the original indicator for the most recent 60 months is taken, the percentile ranking of the current value in the historical sequence is calculated, and linearly mapped to the range of -100 to +100. The trend momentum score is calculated using the slope standardization method. The slope is obtained by linear regression of the most recent 5 data points. The slope is then standardized by dividing the slope by the historical volatility of the indicator and mapped to the corresponding score range. The expected difference is calculated using the adaptive exponential moving average method. An EMA expected sequence is generated based on historical data, and the difference between the actual value and the expected value is calculated and mapped to the -30 to +30 range after standardization by historical volatility. When the expected difference cannot be calculated, its weight is automatically allocated proportionally to the absolute level score and the trend momentum score.
3. The sector-driven force scoring method based on macro-industry-capital three-dimensional aggregation as described in claim 1, characterized in that, In step S2: The macroeconomic gravity axis corresponds to macroeconomic indicators, including but not limited to high-frequency indicators related to credit, inflation, and growth. The industrial gravity axis corresponds to industry chain indicators, including but not limited to industry output, price difference, diffusion index, and inventory cycle indicators. The capital gravity axis corresponds to market trading indicators, including but not limited to price, trading volume, and capital flow indicators.
4. The sector driving force scoring method based on macro-industry-capital three-dimensional aggregation as described in claim 3, characterized in that, In step S2, the initial score of the industry gravity axis needs to be corrected by the inventory warning coefficient: the correction coefficient is calculated based on the finished goods inventory percentile value. When the inventory is at a historical high, the initial score of the industry axis is corrected downward, and when the inventory is at a historical low, the initial score of the industry axis is corrected upward.
5. The sector-driven force scoring method based on macro-industry-capital three-dimensional aggregation as described in claim 1, characterized in that, In step S3, the macro-narrative state includes at least three categories: stagflation, recovery, and overheating, and the corresponding dimension adjustment coefficient matrix is as follows: Stagflation: Macroeconomic gravity axis adjustment coefficient 1.2, industrial gravity axis adjustment coefficient 1.0, capital gravity axis adjustment coefficient 0.5; Recovery status: Macroeconomic gravity axis adjustment coefficient 1.0, industrial gravity axis adjustment coefficient 1.2, capital gravity axis adjustment coefficient 1.0; Overheating state: Macroeconomic gravity axis adjustment coefficient 1.3, industrial gravity axis adjustment coefficient 1.0, capital gravity axis adjustment coefficient 1.
0.
6. The sector-driven force scoring method based on macro-industry-capital three-dimensional aggregation as described in claim 1, characterized in that, In step S4, the historical change acceleration is calculated by the second difference of the scores of the last three dimensions. The larger the absolute value of the acceleration, the higher the dynamic weight coefficient of the corresponding dimension.
7. The sector-driven force scoring method based on macro-industry-capital three-dimensional aggregation as described in claim 1, characterized in that, The method further includes step S6: A physical data shadow index is constructed by selecting high-frequency physical quantity indicators of copper, crude steel, cement, and diesel, and synthesizing them into a physical pulse shadow line after SCE standardization. The physical pulse shadow line is then superimposed on the sector price series for cross-validation.
8. The sector driving force scoring method based on macro-industry-capital three-dimensional aggregation as described in claim 1, characterized in that, In step S5, the formula for calculating the total SCE driving force score is as follows: Where X, Y_adj, and Z are the initial scores for the macro, industry, and capital axes, respectively; w_x, w_y, and w_z are the basic weights for each dimension; and k_x, k_y, and k_z are the comprehensive coefficients for each dimension after narrative adjustment and dynamic weight correction.
9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the sector driving force scoring method based on the three-dimensional aggregation of macro-industry-capital as described in any one of claims 1-8.
10. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the sector driving force scoring method based on the three-dimensional aggregation of macro-industry-capital as described in any one of claims 1-8.