# Read signed forecast error without hiding misses

Illustrative planning brief; no automatic product import.

Synthetic planning case: Four saved duration forecasts are 4,6,5 and 7 days; actual durations are 6,4,7 and 5. The declared signed error is actual minus forecast.

## Decision

Mean signed error is zero, while every forecast misses by two days. Report the lack of mean directional bias separately from error magnitude.

## Owned work

- Freeze the error convention
  - Owner role: Evaluation owner
  - Acceptance evidence: Actual minus forecast is used for all four cases.
- Inspect direction and magnitude
  - Owner role: Reviewer
  - Acceptance evidence: Zero signed mean is shown beside two-day mean absolute error.
- Bound the interpretation
  - Owner role: Planning lead
  - Acceptance evidence: The note does not call the forecasts exact or validated from these four invented cases.

## Workflow

1. Freeze the error convention. Check: Actual minus forecast is used for all four cases.
2. Inspect direction and magnitude. Check: Zero signed mean is shown beside two-day mean absolute error.
3. Bound the interpretation. Check: The note does not call the forecasts exact or validated from these four invented cases.

## Judgment

Four synthetic cases cannot validate a model, independence or future error. These are illustrative held-out-style records, not real task analytics.


## Filled manual planning note

Mean signed error is zero, while every forecast misses by two days. Report the lack of mean directional bias separately from error magnitude. Signed mean=(2−2+2−2)/4=0 days. Mean absolute error=(2+2+2+2)/4=2 days. Cancellation in the signed mean does not erase individual misses. Four synthetic cases cannot validate a model, independence or future error. These are illustrative held-out-style records, not real task analytics.


## Workflow questions

### What does a positive error mean here?

Actual duration was longer than forecast under the stated actual-minus-forecast convention.

### Does zero bias prove a useful forecast?

No. Magnitude, evaluation context and the decisions affected by misses still matter.

## 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

Case | Predicted days | Actual days | Signed error | Absolute error
--- | --- | --- | --- | ---
A | 4 | 6 | +2 | 2
B | 6 | 4 | −2 | 2
C | 5 | 7 | +2 | 2
D | 7 | 5 | −2 | 2

### Reasoning

Signed mean=(2−2+2−2)/4=0 days. Mean absolute error=(2+2+2+2)/4=2 days. Cancellation in the signed mean does not erase individual misses.

### Bounded result

Mean signed error is zero, while every forecast misses by two days. Report the lack of mean directional bias separately from error magnitude.

### Distinct decision

The guide separates directional cancellation from prediction-error magnitude.

### Limits

Four synthetic cases cannot validate a model, independence or future error. These are illustrative held-out-style records, not real task analytics.

### 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.