A base rate is the underlying frequency of an event across a relevant history — the answer to "how often does this happen, usually?" before considering anything about the current situation. If a coin in a basket has closed higher over a seven-day window 56% of the time, its seven-day up base rate is 56%.
Why it comes first
Specific evidence feels more persuasive than general frequencies, so people tend to underweight base rates — a well-documented error known as base-rate neglect. A story about why an asset should rise this week can be compelling while telling you almost nothing beyond what the base rate already did. Starting from the base rate and then asking how far the evidence justifies moving away from it is the discipline that keeps forecasts honest.
The base rate as a benchmark
A forecaster who knew nothing about the current market could still forecast the base rate every time. That naive strategy sets the bar:
- If a set of forecasts cannot beat the base rate, it is not adding information, however sophisticated its reasoning.
- Comparing a Brier score to the base-rate forecast's Brier score gives the Brier skill score.
In crypto this matters more than it sounds. Over many historical windows the major assets rose more often than they fell, so a forecaster who always says "up" can post a respectable hit rate without any insight. A headline accuracy figure means little until it is placed next to that naive number.
Choosing the right base rate
The base rate depends on the question: the horizon (seven days is not thirty), the asset, and the period measured. A base rate built on a bull-market sample will overstate how often prices rise in general. Stating which history a base rate comes from is part of reporting it.
In MoonWire analysis
Each prognosis call is issued with the engine's probability and the base rate beside it, and the published record scores the engine against the base-rate forecaster on the same calls. A call only goes out when the evidence clears a conviction gate relative to that base rate.