一文搞懂港股A股差异：从市场特征到风险模型

Quantitative Models and Construction Methods 1. Model Name: Barra Risk Model - Model Construction Idea: The model identifies and quantifies common factors (industry and style factors) and idiosyncratic factors affecting stock returns, aiming to decompose return sources and provide risk constraints such as industry and style neutrality[119][121]. - Model Construction Process: - Data Processing: - Align reporting periods due to varying fiscal years among Hong Kong-listed companies[120]. - Standardize accounting standards using Wind GSD reports, converting to IFRS[120]. - Normalize financial reporting currencies using time-series exchange rates[120]. - Exclude dual-counter trading stocks and REITs for consistency[120]. - Factor Construction: - Constructed for three pools: Southbound Stock Connect, full Hong Kong market, and full market excluding penny stocks[121]. - Includes 10 major style factors (e.g., Beta, Momentum, Size, Earnings Yield) and industry factors[121]. - Factors are standardized using median-based outlier removal and normalized to standard normal distribution[122]. - Factor returns are calculated using Weighted Least Squares (WLS) with free-float market cap weights[122]. - Formula for WLS: min Σ sqrt(w_i,t-1) * (r_i,t - Σ β_i,k,t-1 * f_k,t + Σ γ_i,m,t-1 * g_m,t)^2 where weights are the fourth root of free-float market cap proportions[123]. - Factor Definitions: - Beta: Systematic risk derived from regression of stock returns against market returns[125]. - Momentum: Weighted log returns over a 500-day period[125]. - Size: Logarithm of total market capitalization[125]. - Earnings Yield: Inverse of P/E ratio[125]. - Volatility: Derived from residual standard deviation and monthly return ranges[125]. - Growth: Based on regression of past five years' revenue and earnings per share[125]. - Value: Inverse of P/B ratio[125]. - Leverage: Includes metrics like market leverage and debt-to-asset ratio[125]. - Liquidity: Based on trading volume relative to free-float shares over different time horizons[125]. - Model Evaluation: The model demonstrates good applicability in the Hong Kong market, with stable R² values across different pools[126][152]. --- Model Backtesting Results 1. Barra Risk Model - R² Values: - Southbound Stock Connect: Average R² = 36.7% since 2014[126][152]. - Full Hong Kong Market: Average R² = 17.4%[126][152]. - Full Market (Excluding Penny Stocks): Average R² = 21.9%[126][152]. - Factor Cumulative Returns: - Positive: Beta, Momentum, Liquidity, BP, Earnings Yield[130][149]. - Negative: Size, Growth, Non-Linear Size[130][149]. - Mixed: Volatility (positive for full market, negative for Southbound Stock Connect), Leverage (positive for full market, negative for Southbound Stock Connect)[130][149]. --- Quantitative Factors and Construction Methods 1. Factor Name: Penny Stock Screening - Factor Construction Idea: Screen stocks based on price, financial, and behavioral metrics to identify and exclude penny stocks[117][118]. - Factor Construction Process: - Price Metrics: - Rolling one-month average closing price < 1 HKD and market cap < 10 billion HKD[117]. - Rolling one-month average closing price < 1 HKD and average trading volume in the lowest 20% of the market[117]. - Financial Metrics: - ROE and debt-to-asset ratio thresholds[118]. - Behavioral Metrics: - History of stock splits or rights issues in the past year[118]. - Exclusion of "Three Highs": High equity pledges, high debt ratios, and frequent fundraising activities[118]. - Factor Evaluation: Helps mitigate risks associated with high-volatility, low-liquidity stocks prevalent in the Hong Kong market[117][118]. --- Factor Backtesting Results 1. Penny Stock Screening - Market Characteristics: - As of June 2025, 56% of Hong Kong stocks are priced below 1 HKD, with 14.5% below 0.1 HKD[112][113]. - Penny stocks exhibit high volatility and low liquidity, making them risky for investment[112][117]. 2. Barra Risk Model Factors - Cumulative Returns by Factor: - Beta: Positive across all pools[130][149]. - Momentum: Positive across all pools[130][149]. - Size: Negative across all pools[130][149]. - Earnings Yield: Positive across all pools[130][149]. - Volatility: Positive for full market, negative for Southbound Stock Connect[130][149]. - Growth: Negative across all pools[130][149]. - Value: Positive across all pools[130][149]. - Leverage: Positive for full market, negative for Southbound Stock Connect[130][149]. - Liquidity: Positive across all pools[130][149].