Plain definition
Forecast bias is a consistent tendency for forecasts to sit above or below actual results over repeated periods. It differs from random error because its direction persists and may reflect incentives, a weak model, or ignored evidence.
Illustrative example
Monetary examples use Saudi riyals (SAR). The same principles apply in other currencies.
A sales team forecasts in each of three months; actual sales are , , and . The repeated overstatement is not cured by saying the average error is small. Management reviews pipeline probabilities and whether bonus incentives reward optimistic submissions.
Why it matters
Persistent optimism or pessimism distorts resource allocation even when individual errors appear modest. Detecting direction lets leaders fix incentives or model inputs before the next commitment.
Practical consequence
Keep forecast versions, calculate signed as well as absolute error across many periods, and trace a recurring direction to its assumptions. Do not “correct” bias by mechanically lowering every forecast when the market has genuinely changed.