Statistic Arbitrage Analysis Based on Goodwill-Net-Income Factor

We study a statistical arbitrage of the goodwill-to-net-income (GTNI) factor in China A-share market in 2016-2024. We build an industry-adjusted, dollar-neutral long-short portfolio on a cross-sectional ranking based on the daily prices of all tradable A-shares of the ordinary. Three economically oriented filters are used to improve the quality of the signals by screening out firms that earned no profits in the previous three years, no movement of goodwill over the last three years, and negative goodwill balances. Refined (2016-2020) provides an annualized 6.93 return, Sharia of 1.28 and a maximum drawdown of 8.52. Out-of-sample (2021-2024), the performance declines to -2.15% per year with the worst drawdown of -20.63, which is sensitive to the changes in the regulation and regimes. In-sample validity is validated by nine robustness checks, such as alternative rebalancing schedules, variant specifications of signals with winsorization, industry and factor neutralization, transaction-cost stress tests, subsample/event-window analysis, formal inference (Newey-West, Fama-MacBeth, SPA/HLZ), bootstrap validation, capacity/crowding tests, and data-integrity screens, and attributes the out-of-sample worsening to structural breaks other than parameter fragility. Our findings demonstrate the potential and vulnerability of accounting-related issues in emerging markets and the critical nature of regular model reviews to quantitative investors who track risks surrounding goodwill.