Quantitative Models and Factor Construction Quantitative Models and Construction Process 1. **Model Name**: Point and Figure Chart (P&F) **Model Construction Idea**: The P&F chart is based on Wyckoff's "cause and effect" principle, which links the accumulation/distribution phase to subsequent price movement. It uses effective price oscillations in a consolidation range to predict the magnitude of price breakout, emphasizing "how long horizontally, how high vertically" in terms of price oscillations rather than time[1][7][9]. **Model Construction Process**: - **Single Point P&F Chart**: Records price movements without considering time. For example, if a stock price moves from 50 to 53, it records 50, 51, 52, and 53 in one column. A reversal in price direction starts a new column[16][19]. - **Three-Point P&F Chart**: Introduces a "reversal value" parameter (Reversal = 3), filtering out minor price movements. This chart is more focused on medium-term trends. The box size (Box Size) is dynamically adjusted based on recent price levels, e.g., a fixed 2% proportion[21][22]. - **Target Price Formula**: $ Target\ Price = Horizontal\ Columns \times Box\ Size \times Reversal\ Value + Starting\ Price $ Horizontal columns represent the number of columns in the trading range, excluding the initial and final trend columns. The starting price can be the lowest/highest price in the range or the last price before the breakout[30][31]. **Model Evaluation**: The P&F chart effectively compresses price consolidation ranges, highlights key breakout signals, and provides a quantitative framework for price target estimation. However, it lacks time dimension and is sensitive to external shocks[15][83]. 2. **Model Name**: Trading Range Quantification **Model Construction Idea**: Defines trading ranges quantitatively to ensure accurate P&F chart predictions. A trading range is identified based on price oscillations and structural rules[42][46]. **Model Construction Process**: - Minimum 5 columns are required for a trading range, with the width not significantly exceeding the height (to avoid extreme oscillations). - **Identification Rules**: - If the right-hand side (RHS) three columns overlap with the current column by more than 20%, and the fourth RHS column overlaps with the first three by more than 50%, the RHS condition is satisfied. - If the RHS four columns overlap with the left-hand side (LHS) first column by less than 20%, the LHS condition is satisfied. - If both RHS and LHS conditions are met, the five columns form a trading range. - **Expansion Rules**: Subsequent columns are added if they overlap with the RHS three columns and their LHS three columns by at least 50% each. - **Width-Height Check**: If the maximum price difference (in points) divided by the width exceeds 1.5, the range is discarded[46][49][50]. **Model Evaluation**: The rules effectively identify trading ranges, ensuring the P&F chart's predictive accuracy. However, extreme oscillations or complex market conditions may challenge its application[51][52]. Model Backtesting Results 1. **P&F Chart**: - **Accuracy**: - For upward targets: 20% achieved with 0-point error, 73% within ±5 points. - For downward targets: 20% achieved with 0-point error, 76% within ±5 points[62][63]. - **Impact of Initial Space**: - Initial space 10%-20%: 23% precision, 82% within ±5 points. - Initial space 20%-50%: 17% precision, 64% within ±5 points. - Initial space >50%: 9% precision, 38% within ±5 points. Predictions for >50% space should be conservatively adjusted by 30%[67][68]. 2. **Quantitative Strategy Using P&F Chart**: - **Annualized Return**: 18.1% (2010-2024), outperforming the Wind All A Index (3.8%) and pure timing strategy (8.5%). - **Maximum Drawdown**: -35.0%, compared to -56.0% for the Wind All A Index and -28.5% for the timing strategy. The strategy improves risk-return characteristics[77][80]. Quantitative Factors and Construction Process 1. **Factor Name**: Effective Oscillation Count **Factor Construction Idea**: Measures the number of effective oscillations in a trading range to estimate breakout potential. More oscillations indicate higher breakout potential[1][7]. **Factor Construction Process**: - Count the number of price oscillations within a defined trading range. - Use the count to estimate the breakout magnitude using the target price formula[30][31]. **Factor Evaluation**: Provides a robust measure of market accumulation/distribution phases, aiding in price target estimation. However, it may be less effective in highly volatile or irregular markets[15][83]. Factor Backtesting Results 1. **Effective Oscillation Count**: - **Prediction Accuracy**: - 70% of targets (upward/downward) achieved within ±5 points in the CSI 800, CSI 1000, and CSI 2000 indices. - Prediction accuracy is independent of market capitalization[62][63][66]. - **Impact of Initial Space**: Predictions for >50% initial space require conservative adjustments to avoid overestimation[67][68].