Quantitative Models and Construction Methods 1. Model Name: PB-ROE Model - **Model Construction Idea**: The PB-ROE model calculates residuals from a time-series regression, where the residual represents the deviation of the market's actual valuation from the fundamental fair valuation. This deviation is defined as the PB-ROE valuation deviation[1][8][16] - **Model Construction Process**: The time-series PB-ROE model is based on the following regression equation: $ Ln(P/B) = a + b \cdot ROE + C \cdot RealInterest + d \cdot Inflation $ - $Ln(P/B)$: Logarithm of Price-to-Book ratio - $ROE$: Return on Equity - $RealInterest$: Real interest rate - $Inflation$: Inflation rate - $a, b, C, d$: Regression coefficients The residuals from this regression represent the PB-ROE valuation deviation, which is used to assess market sentiment and risk appetite. - When the deviation > 0, actual PB is higher than the fair value, indicating high market sentiment and increased risk appetite - When the deviation < 0, actual PB is lower than the fair value, indicating low market sentiment and decreased risk appetite[8][16] - **Model Evaluation**: The PB-ROE valuation deviation is positively correlated with the next week's index return, showing statistical significance. This suggests that the model effectively captures market sentiment and provides actionable signals for tactical positioning[9][16] 2. Model Name: PB-ROE Valuation Deviation-Based Positioning Model - **Model Construction Idea**: The model uses the PB-ROE valuation deviation to determine weekly tactical positioning. Historical mean ± 1 standard deviation is used as the threshold to classify market conditions and generate position signals[2][10][20] - **Model Construction Process**: - Historical mean ± 1 standard deviation of PB-ROE valuation deviation is calculated - Positioning signals are determined as follows: 1. Deviation > Mean + 1 SD: High position 2. Mean < Deviation < Mean + 1 SD: Low position 3. Mean - 1 SD < Deviation < Mean: Medium position 4. Deviation < Mean - 1 SD: Medium-high position - Backtesting results show that this positioning strategy effectively reduces drawdowns and improves returns[10][20] - **Model Evaluation**: The model demonstrates strong historical performance in reducing risk and enhancing returns, making it a valuable tool for tactical asset allocation[20] --- Model Backtesting Results 1. PB-ROE Model - Current PB-ROE valuation deviation: 0.1542[11] - Current standard deviation multiple: 0.8815[12] 2. PB-ROE Valuation Deviation-Based Positioning Model - Current weekly position signal: Low position (0%-30%)[11]

Group 1: Core Insights - The report introduces the PB-ROE model, which establishes a linear relationship between the logarithm of price-to-book (PB) ratio and return on equity (ROE) under the assumption of no dividends and clean surplus [6][8] - The model is effective in explaining current stock pricing and predicting future return differences, indicating that abnormal ROE will converge to the required return rate over the long term [9][11] - The empirical analysis of the A-share market shows that the PB-ROE model has a strong explanatory power, with an adjusted R-squared value exceeding 0.80, indicating that PB is primarily driven by earnings, real interest rates, and inflation [20][21] Group 2: Trading Strategies - The report outlines a position timing strategy based on the PB-ROE valuation deviation, categorizing market conditions into four states: very high, relatively high, relatively low, and very low valuation deviations [41][43] - The backtesting results from January 3, 2014, to February 27, 2026, show that the timing strategy significantly reduces drawdowns, with cumulative returns of 283.45% compared to a benchmark of 122.61% [47][48] - The report emphasizes the importance of adjusting positions based on valuation states to optimize trading outcomes [41][43] Group 3: Key Trading Signals - The report provides statistical analysis of trading signals based on valuation deviations, indicating that high valuation deviation areas yield high win rates and odds for long positions, while mid-high valuation areas are more favorable for short positions [57][59] - The performance of trading signals varies significantly across different valuation deviation regions, with specific thresholds identified for optimal trading strategies [56][60] - The analysis suggests that when valuation deviations exceed certain thresholds, the probability of successful trades increases, highlighting the model's utility in identifying trading opportunities [59][61] Group 4: Industry Rotation - The report applies the PB-ROE model to industry analysis, revealing that different industries exhibit distinct valuation deviation characteristics and their correlation with future returns [67][68] - A long-short industry portfolio strategy is proposed, where positions are adjusted based on the valuation deviation of each industry, leading to significant outperformance of the long portfolio compared to a benchmark [71][76] - The findings indicate that industries with positive correlation coefficients should have increasing positions as valuation deviations rise, while those with negative coefficients should see decreasing positions [68][69]

Quantitative Models and Construction Methods 1. Model Name: PB-ROE Model - **Model Construction Idea**: The PB-ROE model calculates residuals from a time-series regression, where the residual represents the deviation of the market's actual valuation from the fundamental fair valuation. This deviation is defined as the PB-ROE valuation deviation[1][8][15] - **Model Construction Process**: 1. The time-series PB-ROE model is based on the Wilcox & Philips (2005) framework, which modifies the cross-sectional PB-ROE model by incorporating macroeconomic variables for time-series return analysis[14][15] 2. The regression equation is: $ Ln(P/B) = a + b \cdot ROE + C \cdot RealInterest + d \cdot Inflation \tag{1} $ - $Ln(P/B)$: Logarithm of Price-to-Book ratio - $ROE$: Return on Equity - $RealInterest$: Real interest rate - $Inflation$: Inflation rate 3. The residual from equation (1) is defined as the PB-ROE valuation deviation, which measures the extent to which the actual PB deviates from the fair PB based on fundamentals[8][15] 4. Interpretation of the deviation: - When deviation > 0, actual PB is higher than the fair PB, indicating high market sentiment and increased risk appetite - When deviation < 0, actual PB is lower than the fair PB, indicating low market sentiment and reduced risk appetite[8][9][15] - **Model Evaluation**: The PB-ROE valuation deviation is statistically significantly positively correlated with the next week's index return, demonstrating its effectiveness in capturing market sentiment and guiding tactical positioning[9][15] --- Model Backtesting Results 1. PB-ROE Model - **Valuation Deviation on 2026/2/27**: 0.1645, above the historical mean + 1 standard deviation, indicating high market sentiment and suggesting a high allocation (80%-100%) for the week of 2026/3/2-2026/3/6[3][10] - **Current Standard Deviation Multiple**: 1.0467[11] --- Quantitative Factors and Construction Methods 1. Factor Name: PB-ROE Valuation Deviation - **Factor Construction Idea**: The PB-ROE valuation deviation measures the market's actual valuation relative to the fair valuation derived from fundamentals. It serves as a proxy for market sentiment and risk appetite[1][8][15] - **Factor Construction Process**: 1. Calculate the residual from the time-series PB-ROE regression model (equation 1 above) 2. Define the residual as the PB-ROE valuation deviation[8][15] 3. Use historical data to establish thresholds for tactical positioning: - Deviation > Mean + 1 Standard Deviation: High allocation - Mean < Deviation < Mean + 1 Standard Deviation: Low allocation - Mean - 1 Standard Deviation < Deviation < Mean: Medium allocation - Deviation < Mean - 1 Standard Deviation: Medium-high allocation[10][19] - **Factor Evaluation**: - The factor is positively correlated with the next week's index return, with the highest deviation group showing the most significant momentum effect. Lower deviation groups also exhibit positive returns due to safety margins, indicating limited downside risk and potential investment opportunities[18][19] --- Factor Backtesting Results 1. PB-ROE Valuation Deviation - **Deviation Group Performance**: - Highest deviation group (Group 4): Significantly higher next-week returns, reflecting strong momentum effects - Lower deviation groups: Positive returns due to safety margins and limited downside risk[18] - **Historical Backtesting**: The PB-ROE valuation deviation-based positioning strategy effectively reduces drawdowns and enhances returns over time[19]

Quantitative Models and Construction Methods 1. Model Name: PB-ROE Model - **Model Construction Idea**: The PB-ROE model calculates residuals from a time-series regression, where the residual represents the deviation of the market's actual valuation from the fundamental fair valuation. This deviation is defined as the PB-ROE valuation deviation[1][8][15] - **Model Construction Process**: The time-series PB-ROE model is based on the following regression equation: $ Ln(P/B) = a + b \cdot ROE + C \cdot RealInterest + d \cdot Inflation $ - $P/B$: Price-to-Book ratio - $ROE$: Return on Equity - $RealInterest$: Real interest rate - $Inflation$: Inflation rate - $a, b, C, d$: Regression coefficients The residuals from this regression represent the PB-ROE valuation deviation, which is used to assess whether the market's actual PB is above or below the fair value based on fundamentals[8][15] - **Model Evaluation**: The PB-ROE valuation deviation is positively correlated with the next week's index return, showing statistical significance. This indicates that the model effectively captures market sentiment and risk appetite[9][15] 2. Model Name: PB-ROE Valuation Deviation-Based Positioning Model - **Model Construction Idea**: The PB-ROE valuation deviation is used to determine weekly portfolio positioning. Historical mean and standard deviation thresholds are applied to classify market conditions and guide positioning decisions[2][10][19] - **Model Construction Process**: - Calculate the PB-ROE valuation deviation using the residuals from the PB-ROE model - Define thresholds based on the historical mean and ±1 standard deviation of the PB-ROE valuation deviation - Positioning rules: 1. Deviation > Mean + 1 SD: High position (80%-100%) 2. Mean < Deviation < Mean + 1 SD: Low position 3. Mean - 1 SD < Deviation < Mean: Medium position 4. Deviation < Mean - 1 SD: Medium-high position - Backtesting results show that this strategy effectively reduces drawdowns and improves returns[10][19] --- Model Backtesting Results 1. PB-ROE Model - The PB-ROE valuation deviation as of February 6, 2026, was 0.159, exceeding the historical mean by +1 standard deviation. This indicates a high market sentiment phase, suggesting a high position (80%-100%) for the week of February 9-13, 2026[3][11] - The current standard deviation multiple is 1.02, close to the threshold of 1. If it falls below 1, a signal to reduce positions will be triggered[11] 2. PB-ROE Valuation Deviation-Based Positioning Model - Historical backtesting demonstrates that the positioning strategy based on PB-ROE valuation deviation effectively reduces drawdowns and enhances returns. The highest returns are observed in the group with the highest valuation deviation, reflecting a strong momentum effect during high market sentiment periods. Conversely, even in low-deviation groups, returns are positive due to the presence of a safety margin[18][19] --- Quantitative Factors and Construction Methods 1. Factor Name: PB-ROE Valuation Deviation - **Factor Construction Idea**: The PB-ROE valuation deviation measures the extent to which the market's actual PB deviates from its fair value based on fundamentals. It serves as a proxy for market sentiment and risk appetite[1][8][15] - **Factor Construction Process**: - Use the residuals from the PB-ROE time-series regression model: $ Ln(P/B) = a + b \cdot ROE + C \cdot RealInterest + d \cdot Inflation $ - Define the residual as the PB-ROE valuation deviation - Positive deviation indicates high market sentiment and risk appetite, while negative deviation indicates low sentiment and risk appetite[8][15] - **Factor Evaluation**: The PB-ROE valuation deviation is statistically significantly positively correlated with the next week's index return, demonstrating its predictive power for short-term market movements[9][15] --- Factor Backtesting Results 1. PB-ROE Valuation Deviation - The PB-ROE valuation deviation is positively correlated with the next week's index return, but the relationship is not strictly linear. The highest deviation group exhibits the highest returns, indicating a strong momentum effect. Lower deviation groups also show positive returns due to the presence of a safety margin, highlighting the factor's robustness in different market conditions[18][19]

Investment Rating - The report suggests a high position (80%-100%) for the week of February 2-6, 2026, based on the PB-ROE model indicating a market sentiment that is high and in a phase of valuation expansion [3][9]. Core Insights - The PB-ROE model calculates the residuals, which represent the deviation of actual valuations from fundamental reasonable valuations. A positive residual indicates that the market is optimistic, while a negative residual suggests a pessimistic market sentiment [1][12]. - The A-share market's PB-ROE residuals are positively correlated with the index's performance in the following week, with statistical significance. When residuals exceed historical averages, it indicates extreme market optimism, warranting a high allocation. Conversely, when residuals fall below historical averages, it suggests a strong safety margin with limited downside, allowing for a medium to high allocation [2][8]. Summary by Sections Current Weekly Position Insights - As of January 30, 2026, the overall PB-ROE residual value for the market is 0.18, which exceeds the historical average plus one standard deviation, indicating a high allocation recommendation for the upcoming week [3][9]. PB-ROE Model Methodology - The PB-ROE model, based on the work of Wilcox & Philips, establishes a linear relationship between log PB and ROE under certain assumptions. The model's residuals reflect the market's actual valuation compared to fundamental values, influencing investment decisions based on market sentiment [12][13]. Investment Strategy Based on Residuals - The report categorizes investment positions based on the residual values: 1. Residuals greater than the mean plus one standard deviation suggest a high position 2. Residuals below the mean minus one standard deviation indicate a medium to high position 3. Residuals between the mean and plus one standard deviation suggest a low position 4. Residuals between the mean minus one standard deviation and the mean indicate a medium position - Historical backtesting shows that this timing strategy effectively reduces drawdowns and enhances returns [18][20].

- Model Name: PB-ROE Model - Model Construction Idea: The PB-ROE model calculates the residuals of the time series PB-ROE model, where the residuals represent the deviation of the market's actual valuation from the fundamental reasonable valuation[1][8] - Model Construction Process: - The model uses the regression equation: $Ln(P/B) = a + b \cdot ROE + C \cdot RealInterest + d \cdot Inflation$ - In this equation, $a$, $b$, $C$, and $d$ are parameters, $ROE$ is the return on equity, $RealInterest$ is the real interest rate, and $Inflation$ is the inflation rate[8][13] - The residuals are calculated as the difference between the actual PB and the fundamental reasonable PB[8] - When the residual > 0, the actual PB is higher than the fundamental reasonable PB, indicating high market sentiment and increased risk appetite[1][8] - When the residual < 0, the actual PB is lower than the fundamental reasonable PB, indicating low market sentiment and decreased risk appetite[1][8] - Model Evaluation: The PB-ROE residuals are significantly positively correlated with the next week's index returns, indicating that the model can effectively capture market sentiment and provide tactical positioning signals[8][13] - Model Test Results: - As of January 30, 2026, the overall market PB-ROE residual value was 0.18, exceeding the historical mean + 1 standard deviation, suggesting a high position (80%-100%) for the week of February 2-6, 2026[3][9] - Factor Name: PB-ROE Residual - Factor Construction Idea: The PB-ROE residual is the deviation of the market's actual valuation from the fundamental reasonable valuation, used to gauge market sentiment and risk appetite[1][8] - Factor Construction Process: - The factor is derived from the time series PB-ROE model: $Ln(P/B) = a + b \cdot ROE + C \cdot RealInterest + d \cdot Inflation$ - The residuals are calculated as the difference between the actual PB and the fundamental reasonable PB[8][13] - The historical mean ± 1 standard deviation of the residuals is used to determine the positioning signals[2][8] - Factor Evaluation: The PB-ROE residual is significantly positively correlated with the next week's index returns, indicating its effectiveness in capturing market sentiment and providing tactical positioning signals[8][13] - Factor Test Results: - The PB-ROE residual value as of January 30, 2026, was 0.18, suggesting a high position (80%-100%) for the week of February 2-6, 2026[3][9]