20231229-东吴证券-金工定期报告_估值异常因子绩效月报_17页_980kb

Summary of Report: Valuation Anomaly Factor Performance

Report Overview

This report analyzes valuation anomaly factors—EPD (Valuation Deviation), EPDS (Slow Valuation Deviation), and EPA (Abnormal Valuation)—constructed using a combination of CTA Bollinger Band strategies and fundamental valuation metrics. The focus is on performance metrics from 2010 to 2023 and backtesting results to improve risk-adjusted returns.

Key Performance Metrics

The factors demonstrate strong historical performance across A-shares (excluding BSE stocks). Here are the key annualized performance stats to December 2023:

In December 2023, the EPA factor showed a positive return of 1.01% for a combined long/short portfolio, with the long portion at -0.083% and short at -1.85%.

Factor Construction Methodology

The factors are built by combining mean reversion principles from Bollinger Bands with discounted cash flow (DCF) models:

• EPD is derived from price-to-earnings (PE) ratios, using rolling mean and standard deviation to identify valuation deviations.
• EPDS incorporates cross-sectional regression to remove stock-specific valuation changes and industry effects.
• EPA further refines the model by excluding Beta, value, and growth style factors to filter out non-mean-reverting influences, aiming to capture true valuation anomalies.

Backtesting from 2010 to 2022 shows:

• Optimizations improved information ratios and reduced maximum drawdowns, with EPA achieving the highest risk-adjusted returns (e.g., 87.0% win rate, 15.3% drawdown after style removal).

Risk Considerations

The report highlights that all metrics are based on historical data, which may not reflect future performance. Single-factor strategies are volatile and require integration with risk management techniques. No guarantee of future results.

This summary captures the essence of the report's findings and methodology.