You forecast that your home renovation will be finished within three months and put it at 70%. In week two, the foreman says the tiles will arrive a week late.
Some people immediately slash their estimate to 10%: "That's it, it's going to overrun." Others don't budge: "It's only a week, still 70%." These are overreaction and underreaction.
Work out how much to move
Think about it: among renovations that finish on time, how many hit a late delivery along the way? And among those that overrun? Suppose overrunning projects are three times as likely to hit this kind of delay. The likelihood ratio is 1/3.
70% is odds of 7 to 3. Multiply by 1/3 and you get 7 to 9, which is about 44%. The sensible answer is somewhere around 44%: a clear move down, but nowhere near 10%.
Overreaction: getting carried away by one headline
This tends to happen with news that is fresh, vivid, or emotional. One day a rumor hits the stock market and someone drops "the index will be up by year-end" from 60% to 20%, then jumps back to 65% when the rumor is denied the next day.
- Sign: your probability line zigzags, and every swing feels like "this time is different."
- Cause: treating one piece of news as decisive, ignoring that it could easily show up either way.
Underreaction: refusing to let go
This tends to happen when evidence piles up gradually. Your project has reported falling behind three weeks in a row, yet you're still holding 75% on "finished on time," telling yourself it'll catch up next week.
- Sign: long before the outcome, the evidence was clearly leaning one way, but your number barely moved.
- Cause: attachment to your original view, or each piece of evidence seems "too small" on its own, so none of them gets used.
How to check yourself
Before each update, ask: "If my original view were wrong, how likely would I be to see this news?" Then look back at your probability line. Big back-and-forth swings usually mean overreaction; realizing at the end that you "should have moved earlier" usually means underreaction.
Good updating neither bends with every breeze nor digs in: let each piece of evidence push you exactly as hard as it deserves.