Discussion about this post

User's avatar
Henry Zhang | QuantSignals.xyz's avatar

The gap between the 103% aggregate and the 41% median is doing more work than the post grants it. By the appendix, the Fisher index zero-fills missing opposite-week payments, while the median diagnostic only chains organizations positive in both weeks, so entry, exit and churn land differently in the two statistics. That means the spread is partly about concentration and survival, not only size, and the same asymmetry could drive the apparent cooling from 140% to 80% annualized. What I would want is the decomposition: how much of the deceleration is surviving cohorts slowing versus cohort turnover?

Ready for more?