In one line
A base rate is a count of what has already happened in cases like this one, reported with how many cases there were — and it is the only forward-looking form of statement this desk makes.
How it works
"The last 24 RBI decisions moved NIFTY about 0.6% either way, and more than 1% in 7 of them" is a base rate. Every part of it is measured: the class of event, the count, the middle of the distribution and how often the tail was reached. Nothing in it is a claim about the next one.
THE COUNT IS NOT DECORATION. A rate over nine cases and a rate over nine hundred read identically on a page and mean entirely different things, so this desk prints the count beside every rate and greys the rate below a floor. The floor is thirty for session-based figures and twelve for events, and the lower event floor is deliberate arithmetic rather than a relaxation: there are eight policy decisions in a year and two hundred and fifty sessions, so a thirty-case floor would grey every event on the desk and the section would print nothing.
Each sub-measurement carries its OWN count, which is usually smaller than the headline one. The session after an event has no reading for the newest event; the comparison with what the option market was charging exists only for events the chain archive reaches back to. Reporting either against the headline count would be borrowing a sample, and this desk refuses that everywhere.
What it is not
A base rate is not a probability for tomorrow, and it is not improved by being read as one. It is also not a claim that the past distribution holds: markets change, and the honest reading of "the last 24 moved 0.6%" is exactly that sentence and nothing more. Where a desk cannot even say that — a first-of-its-kind event, a class with three observations — the right answer is no number, not the nearest available one.