# Inspect a forecast decision with asymmetric mistake costs

Illustrative planning brief; no automatic product import.

Synthetic planning case: A toy decision can commit or wait. Its supplied readiness probability is 0.6. A false commitment costs four declared loss points; waiting when ready costs one. Correct decisions cost zero.

## Decision

Under these hypothetical inputs, committing has expected loss 1.6 points and waiting 0.6, so the declared loss rule favors waiting. A majority probability alone does not choose the action.

## Owned work

- Define actual action consequences
  - Owner role: Decision owner
  - Acceptance evidence: The four-to-one loss convention and zero correct-decision loss are declared.
- Inspect the conditional comparison
  - Owner role: Reviewer
  - Acceptance evidence: At supplied p0.6,1.6 exceeds 0.6 expected loss points.
- Check evidence before real use
  - Owner role: Planning lead
  - Acceptance evidence: Actual probabilities, authority and consequences need independent justification before any commitment.

## Workflow

1. Define actual action consequences. Check: The four-to-one loss convention and zero correct-decision loss are declared.
2. Inspect the conditional comparison. Check: At supplied p0.6,1.6 exceeds 0.6 expected loss points.
3. Check evidence before real use. Check: Actual probabilities, authority and consequences need independent justification before any commitment.

## Judgment

Loss points are local toy preferences, not money, measured harm or a legal/safety assurance. No live task probability is validated.


## Filled manual planning note

Under these hypothetical inputs, committing has expected loss 1.6 points and waiting 0.6, so the declared loss rule favors waiting. A majority probability alone does not choose the action. Commit loss=4(1−p); wait loss=p. They tie when 4−4p=p, so p=0.8. This threshold follows the toy losses and supplied probability, not a universal readiness policy. Loss points are local toy preferences, not money, measured harm or a legal/safety assurance. No live task probability is validated.


## Workflow questions

### Is 0.8 a recommended business threshold?

No. It is the algebraic result of this invented four-to-one loss rule.

### Does the worksheet validate the supplied 0.6 probability?

No. It is a hypothetical input; a real decision needs justified probability and consequence evidence.

## Product connection

Use the owned checks and downloaded brief to discuss this planning decision alongside your TeamBoostAI tasks. Confirm available fields, roles and account features separately. The example is manual; it does not calculate live analytics, create work or run an experiment in the product.

Confirm account availability before adopting this manual outline.

## Original worked case

Synthetic records, manual planning only. No account import or live analytics.

### Inspect the invented case records

Action | Mistake probability under supplied p | Loss if mistaken | Expected loss
--- | --- | --- | ---
Commit | 1−0.6=0.4 | 4 points | 1.6
Wait | 0.6 | 1 point | 0.6

### Reasoning

Commit loss=4(1−p); wait loss=p. They tie when 4−4p=p, so p=0.8. This threshold follows the toy losses and supplied probability, not a universal readiness policy.

### Bounded result

Under these hypothetical inputs, committing has expected loss 1.6 points and waiting 0.6, so the declared loss rule favors waiting. A majority probability alone does not choose the action.

### Distinct decision

This links a binary decision threshold to asymmetric declared mistake costs, distinct from accuracy or probability-error scoring.

### Limits

Loss points are local toy preferences, not money, measured harm or a legal/safety assurance. No live task probability is validated.

### Definitions and method context

- Forecasting: Principles and Practice — accuracy — https://otexts.com/fpp3/accuracy.html — Genuine held-out forecast and point-error evaluation context. Binary scoring examples use their own explicit toy definitions; no real task model is validated.