Coaching practices for Forecast Revision
Describe almost anything you are working through and IX Coach finds the practices whose real-world fit is closest. For Forecast Revision, these are the strongest matches in the current practice library.
Does this sound like the set of challenges you might be facing?
- New information keeps arriving that should obviously move my estimate, but I just cling to the date I first committed to
- Once I’ve made up my mind I dig in and won’t budge until I’m absolutely certain I was wrong
- I commit to one confident number
- I catch myself shaving down the estimate I’m about to present
- Each project wraps up and I move straight on without ever writing down what I’d guessed versus what actually happened
Practices that may help
- Update your forecast incrementally as new evidence arrives
Treat your forecast as a probability that should shift with each new piece of evidence, not a commitment that survives contradiction.
Reference Class Forecasting - Update beliefs frequently and in small increments
When new evidence arrives, adjust your probability estimate — even if the change is small.
Superforecasting - Reference Class Forecasting
Reference class forecasting, developed by Daniel Kahneman and Amos Tversky and formalized by Bent Flyvbjerg, improves forecast accuracy by anchoring on the statistical distribution of outcomes for similar past projects rather than on the details of the current one. The method reliably corrects the optimism bias that inflates cost, time, and benefit estimates in planning — the evidence base here is real and specific. - Forecast a distribution, not a point estimate
Represent your forecast as a range of likely outcomes, not a single predicted number.
Reference Class Forecasting - Distinguish cognitive optimism bias from strategic misrepresentation
Recognize that some forecast inflation is genuine bias and some is deliberate spin — they require different fixes.
Reference Class Forecasting - Conduct post-project debriefs to contribute to the class data
Record actual vs. forecast outcomes honestly — this builds the reference class that future forecasts depend on.
Reference Class Forecasting - Anchor on the base rate before adding inside-view details
Start your forecast from the class median, then adjust — do not start from your narrative and adjust to the base rate.
Reference Class Forecasting - Track your estimates and calibrate
Compare your Fermi estimates to actual figures when you can, and use the gap to improve future estimates.
Fermi Estimation - Estimate in ranges, not point estimates
Instead of "my estimate is 500," say "I think it is between 200 and 2000."
Fermi Estimation - Treat forecasting accuracy as a skill that improves with practice
Take prediction errors as performance feedback, not as proof that forecasting is futile.
Superforecasting
Related concerns
- Forecast Revision Practice
Treat your forecast as a probability that should shift with each new piece of evidence, not a commitment that survives contradiction.
Update your forecast incrementally as new evidence arrives
- How To Make Better Forecasts
Record actual vs. forecast outcomes honestly — this builds the reference class that future forecasts depend on.
Conduct post-project debriefs to contribute to the class data
- Superforecasting On A Budget
Philip Tetlock’s forecasting tournament research found that a subset of ordinary people — "superforecasters" — consistently outperform domain experts and intelligence analysts at probabilistic prediction. They share measurable cognitive and behavioral habits: they think in probabilities, update frequently on evidence, and actively seek disconfirming information. These habits are learnable.
- Decompose Forecasting Question
Break a hard forecasting question into smaller, estimable pieces and aggregate them.
Decompose complex questions into sub-questions
- False Precision Forecast
Treat any precise probability or quantitative forecast with explicit suspicion about whether the model fits the domain.
Beware false precision in forecasts and models
- Goal Vs Forecast
Represent your forecast as a range of likely outcomes, not a single predicted number.
Forecast a distribution, not a point estimate
Describe your situation in your own words to search the complete practice library.