macrocephalopod

Yes you should take differences for yields, not ratios (this was obviously written by someone with pre-2007 “interest rates cannot be negative” brain)

This A more math-y answer is: see the "quest for invariance" https://papers.ssrn.com/sol3/papers.cfm?abstract_id=1753788 …

So to find spread betw trailing 30day vol of SPY & 10yr yield: SPY: column of 30 LogRets. LogRet = ln(PriceToday/PriceYesterday). TrailVol = stdev(column) * sqrt(252). 10yr: I have a col of YieldDiffs (YieldToday-YieldYestrday). TrailVol = stdev(column)*sqrt(252). Is this close?

Yep. Depending upon what you’re using it for, it might make sense to get an instrument that tracks the 10yr rate (e.g. 10yr govt bond index) and find the total return vol. The changing duration may be an “issue,” but it also might be important information to have.

yeah i am going to echo what corey said here. for a lot of things where maturity and/or duration is constant-ish price vol is definitely a better measure if observable. otherwise do yield vol but make sure you are properly translating it to price vol using the changing duration