Core Insights - The article discusses the development and application of the Kronos model, a Time-Series Foundation Model (TSFM) specifically designed for financial market data, particularly K-line data [3][9][17] - Kronos aims to address the challenges of low signal-to-noise ratio and strong non-stationarity in financial time series data, which often hinder the performance of general-purpose models [3][9] - The model employs a two-phase framework: K-line tokenization and autoregressive pre-training, allowing it to effectively learn the complex "language" of financial markets [12][13][17] Summary by Sections Introduction to TSFM - TSFMs have emerged from the success of large-scale language models in NLP and CV, focusing on pre-training on diverse time series data to create a general-purpose model adaptable to various tasks [2][6] - The key advantages of TSFMs include their generalization and transfer learning capabilities, enabling them to learn universal time patterns and trends from vast datasets [2][6] Overview of Kronos Model - Kronos is tailored for financial K-line data, utilizing a "domain pre-training + fine-tuning" approach to deeply understand financial market characteristics [3][9] - The model's architecture includes a specialized tokenizer and a large autoregressive Transformer model, which learns the syntax and dynamics of financial data [9][12][17] Performance Evaluation of Kronos - Initial tests of the Kronos standard model on major A-share indices showed a high correlation between predicted and actual closing prices, with a Spearman correlation coefficient of 0.732 for the 5-day forecast [4][19] - The model's predictive performance improved significantly when fine-tuned, achieving a Spearman correlation of 0.856 for the same forecast [4][39] Application of Kronos in Timing Strategies - The article explores the application of Kronos in constructing timing strategies based on predicted closing prices, specifically for the CSI 1000 index [30][33] - The strategy generated positive returns, but it missed significant upward trends since July 2025, indicating a reliance on prior index reversal logic [30][33] Enhanced Performance with Fine-Tuning - A fine-tuned version of Kronos demonstrated a 33.9% return in 2025, with an annualized excess return of 9%, outperforming the original method by over 20 percentage points [5][42] - The fine-tuning process involved adjusting model parameters and rolling adjustments to better adapt to market conditions, leading to improved predictive accuracy [34][42] Conclusion - Kronos represents a significant advancement in financial time series forecasting, effectively capturing the complexities of financial data and translating predictions into actionable investment strategies [17][42]

Quantitative Models and Construction Methods 1. Model Name: Multi-dimensional Timing Model (Version 1) - **Model Construction Idea**: The model integrates four dimensions—funding, technical, valuation, and sentiment—to provide directional views on the A-share market[1][2] - **Model Construction Process**: The model combines signals from the four dimensions to determine market timing decisions. Each dimension includes specific indicators, such as option PCR, implied volatility, and futures positions for sentiment, and Bollinger Bands and individual stock movements for technical analysis[30] - **Model Evaluation**: The model demonstrated strong performance in capturing upward trends while avoiding significant market volatility[2][10] 2. Model Name: Multi-dimensional Timing Model (Version 2) - **Model Construction Idea**: This version expands the original model by adding a fundamental dimension to capture bottom-buying opportunities and enriching the sentiment dimension with new indicators[1][84] - **Model Construction Process**: - The sentiment dimension was expanded to include futures basis and main funds indicators - The fundamental dimension was introduced to identify bottom signals based on macroeconomic indicators like CPI, PMI, and EPU - The model integrates five dimensions: funding, technical, valuation, sentiment, and fundamentals[84] - **Model Evaluation**: The expanded model achieved higher annualized returns and maintained similar levels of volatility and drawdown compared to the original version[85][88] --- Model Backtesting Results 1. Multi-dimensional Timing Model (Version 1) - **Annualized Return**: 24.57%[14] - **Annualized Volatility**: 21.54%[14] - **Maximum Drawdown**: -28.46%[14] - **Sharpe Ratio**: 1.14[14] 2. Multi-dimensional Timing Model (Version 2) - **Annualized Return**: 26.69%[85] - **Annualized Volatility**: 21.48%[85] - **Maximum Drawdown**: -28.46%[85] - **Sharpe Ratio**: 1.24[85] --- Quantitative Factors and Construction Methods 1. Factor Name: Futures Basis (Sentiment Dimension) - **Factor Construction Idea**: The futures basis reflects price information in the futures market and acts as a sentiment amplifier during extreme market conditions[3][32] - **Factor Construction Process**: - Basis = Futures Price - Spot Price - Annualized basis rate is calculated to reduce the impact of contract expiration - Weighted average of the four contracts (current month, next month, current quarter, next quarter) based on open interest[32] - **Factor Evaluation**: The factor is suitable for mean-reversion strategies, with signals generated during overbought or oversold conditions[40] 2. Factor Name: Main Funds (Sentiment Dimension) - **Factor Construction Idea**: This factor captures the flow of main funds in the stock market, reflecting high-selling and low-buying behavior[3][50] - **Factor Construction Process**: - Signals are derived from smoothed 20-day moving averages of net fund inflows and institutional active buying - Positive signals indicate buying opportunities, while negative signals suggest selling[51][54] - **Factor Evaluation**: The factor is effective for momentum strategies, with a high win rate but relatively low payoff ratio[57] 3. Factor Name: Fundamental Bottom Signal (Fundamental Dimension) - **Factor Construction Idea**: This factor identifies bottom-buying opportunities based on macroeconomic indicators, assuming that poor fundamentals often precede market recoveries[4][76] - **Factor Construction Process**: - Signals are triggered when CPI, PMI, and EPU simultaneously indicate weakening fundamentals - A one-quarter window after the bottom signal is used for long positions[76][83] - **Factor Evaluation**: The factor demonstrates high win rates and payoff ratios in bottom-buying scenarios, significantly outperforming the benchmark[83] --- Factor Backtesting Results 1. Futures Basis - **Annualized Return**: 19.06%[50] - **Annualized Volatility**: 20.85%[50] - **Maximum Drawdown**: -31.31%[50] - **Sharpe Ratio**: 0.91[50] 2. Main Funds - **Annualized Return**: 8.75%[60] - **Annualized Volatility**: 19.19%[60] - **Maximum Drawdown**: -30.57%[60] - **Sharpe Ratio**: 0.46[60] 3. Fundamental Bottom Signal - **Annualized Return**: 12.48%[83] - **Annualized Volatility**: 13.27%[83] - **Maximum Drawdown**: -22.62%[83] - **Sharpe Ratio**: 0.94[83]

Group 1: Macro Analysis - The narrative that the dollar will enter a prolonged decline akin to the 70s and 80s needs reassessment, as the fastest decline of the dollar may have already passed[13] - The U.S. economy's growth rate relative to Europe and Japan remains superior, suggesting potential dollar strength in the medium term[13] - The dollar index has shown a long-term divergence from the U.S. economic share, with the index rising despite a declining economic share post-2008 financial crisis[15] Group 2: Fixed Income Market Insights - In July, the bond market is expected to face downward pressure, with credit outperforming rates[29] - Government bond net financing is projected to increase to between 1.5 trillion and 1.7 trillion yuan in July due to accelerated local government bond issuance[27] - The average decline in the 10-year government bond yield from 2021 to 2024 is approximately 4.4 basis points, indicating a trend of decreasing yields[29] Group 3: Equity Market Trends - The total position of stock funds increased to 94.90%, up by 97 basis points from the previous week, indicating a bullish sentiment[36] - The average return of stock funds this week was 1.31%, reflecting positive market performance[38] - The Hang Seng Index saw a decline of 1.52%, suggesting a mixed outlook for Hong Kong equities[39] Group 4: CIPS Regulatory Changes - The People's Bank of China is revising CIPS rules to enhance participant management and flexibility, allowing for easier access to the system[43] - The CIPS system processed 821.69 million transactions worth 175.49 trillion yuan in 2024, marking a 42.60% increase year-on-year[42] - The new rules include risk management requirements and clarify the roles of domestic and foreign participants in the CIPS framework[43]