Define the event
Use a binary event or measurable threshold. Include a deadline and a rule that another person could apply. Example: “Will I complete three planned exercise sessions between Monday and Sunday?”
Personal psychohistory
Use your past to see what you are likely to do next.
Build a personal record, find recurring conditions, issue forecasts before events, and score them afterward. With enough resolved cases, intuition becomes a model.
The method
Nine steps from memory to measurable foresight.
Begin with repeated, clearly defined behaviors in comparable contexts. Every resolved case strengthens your reference class and makes the next forecast more informative.
Build the reference class
Find past episodes that match the behavior, time horizon, and context. Include ordinary cases, not only memorable successes or failures. When personal data are sparse, widen the class and keep the estimate closer to 50 percent.
Calculate the personal base rate
Count how many comparable episodes ended yes and how many ended no. The simplest base rate is yes divided by total cases. The Forecast Lab adds light smoothing so a tiny sample does not create false certainty.
Measure present similarity
Compare time, place, resources, people, mood, workload, health, and environmental cues. The more the present resembles the reference class, the more weight the old base rate deserves.
Separate intention from evidence
Intention matters, but it is not execution. Compare what you want now with what you actually did under similar constraints, then list the forces likely to help or obstruct you.
Issue the continuity forecast
Estimate what will happen if no deliberate change is made. This is the branch already implied by your history, current evidence, and existing friction.
Change the system deliberately
Specify an if-then trigger, action, resource, environmental change, and accountability mechanism. Once the system changes, issue a separate active forecast.
Update when the evidence changes
When relevant evidence arrives before the deadline, enter a new probability and preserve the old one. Record exactly what changed and why the number moved.
Resolve, score, and improve
Apply the resolution rule, record yes as 1 or no as 0, and calculate the Brier score. Review many forecasts together until your stated probabilities begin to match observed frequencies.
Minimal mathematics
Three tools that turn intuition into a record.
The mathematics makes assumptions visible, preserves them, and allows one forecast to be compared with the next.
(yes + 1) / (cases + 2)
This simple Laplace-style estimate prevents a tiny history from yielding absolute certainty. With 3 successes and 1 failure, it gives 4/6, or about 67 percent, rather than 75 percent.
(probability - outcome)²
For a binary event, outcome is 1 for yes and 0 for no. Lower average scores are better. The score rewards accuracy and appropriately restrained confidence.
Probability compared with frequency
Across enough comparable forecasts, events assigned 70 percent should occur roughly 70 percent of the time. Calibration alone is not the whole of forecast quality, but it exposes systematic overconfidence or underconfidence.
First model: the calculator uses transparent, general-purpose weights. Treat its number as a starting probability, then replace generic assumptions with the evidence produced by your own resolved forecasts.
Personal instrument
The Personal Forecast Lab
Forecast entries stay in this browser unless you export them. The ledger belongs to you and grows more useful each time an outcome is honestly resolved.
Local forecast ledger
Keep every miss beside every hit. A model learns only from a complete record.| Question | Deadline | Forecast | Evidence | Outcome | Brier | Action |
|---|
Resolved forecasts: 0 | Mean Brier score: not available
A 21-day entry practice
Build the first personal dataset.
Twenty-one days is the first observation cycle, not a magic number. Its purpose is to produce a usable sequence of observations, forecasts, and outcomes.
Build the record
Choose one recurring behavior. Log whether it occurred, the time, place, preceding cue, barrier, mood, and consequence. Make no attempt to improve the pattern yet.
Issue daily forecasts
Each morning, predict the day's behavior with a probability. Each evening, resolve it. Note where intention, context, and interruption diverged.
Test one intervention
Keep the outcome definition stable. Add one clear if-then plan or environmental change. Record passive and active forecasts separately.
Failure modes
Seven distortions that weaken foresight.
The correction is procedural: improve the question, broaden the record, separate branches, preserve old forecasts, and resolve every outcome by the rule chosen in advance.
Intention substitution
Answering “how much do I want this?” instead of “how often do I complete this under comparable constraints?”
Reference-class shopping
Choosing only past cases that support the preferred prediction or treating the current case as uniquely exempt from base rates.
Regime blindness
Applying old patterns after a major change in health, schedule, environment, relationships, incentives, or available resources.
Outcome vagueness
Using language flexible enough that any result can later be described as a success.
Double-counted evidence
Treating several correlated signs as independent reasons and adjusting the probability repeatedly for the same underlying fact.
Observer effects
Ignoring that tracking, public commitment, reward, shame, or the forecast itself may change the behavior.
Selective resolution
Keeping the impressive hits while deleting, redefining, or forgetting the misses that would lower confidence in the model.
Professional boundary: this personal forecasting exercise is not a medical, legal, financial, or emergency decision system. Use the appropriate qualified service for those decisions.
Research foundation
Where the method comes from
The practice combines the outside view and reference classes, probability tracking, Brier scoring, research on self-prediction, context-linked habits, implementation intentions, and person-specific longitudinal modeling. The sources page identifies the studies and shows how they form the present foundation of personal psychohistory.