Group 1 - The report highlights that the number of bond funds has steadily increased, currently accounting for approximately 30% of all public funds, with a similar scale proportion [3][14][15] - As of September 12, 2025, the number of medium- and long-term pure bond funds is 2,102, representing 53.75% of total bond funds, while short-term pure bond funds account for 9.36% [18][26] - The average duration of medium- and long-term pure bond funds has risen from 1.61 years in Q1 2023 to 3.98 years by Q2 2025, marking the highest level in nearly a decade [35][46] Group 2 - The report discusses three main methods for calculating the duration of bond funds: weighted average of major bonds, interest rate sensitivity method, and fund net value regression method [34] - The weighted average method is straightforward but suffers from data lag and may not accurately reflect the fund's characteristics due to the low proportion of major bonds disclosed [35][39] - The interest rate sensitivity method provides more accurate results but is limited by the infrequency of data disclosure, which affects its timeliness [42][46] Group 3 - The fund net value regression method is recommended for its higher frequency of data and ability to fit the bond fund's net value trends, although it faces challenges with multicollinearity among explanatory variables [48][50] - The report suggests using Lasso regression for variable selection in the regression model to enhance accuracy and stability [51] - The choice of explanatory variables is crucial, with recommendations to select indices based on government bonds, policy financial bonds, credit bonds, and innovative bonds to ensure granularity [56]