The Newsvendor Challenge

Your role

Order today. Learn demand tomorrow.

You manage a campus pastry counter. Every morning brings a new forecast or a change in the unit economics. Choose how many pastries to prepare before demand is revealed, then learn which input moved the optimal quantity.

1

Identify the change

Compare the morning’s brief with the baseline.

2

Predict, then choose Q

First predict the direction of Q*, then commit a quantity.

3

Explain the movement

Use immediate feedback to connect the changed input to Q*.

The eight-morning challenge

What will change?

After the baseline, five mornings change exactly one input at a time. The seventh combines two opposing changes, and the eighth is a capstone. This lets you learn what moves Q* before solving the full problem.

Your mission Maximize cumulative profit over eight rounds. After each result, compare your choice with that morning’s Q* on the same realized demand.

Demand level: μ
Uncertainty: σ
Economics: C_u/C_o

Time: about six minutes. No calculator required.

Round 1 of 8

Your profit$0

Gap vs Q*—

Morning 1 brief

Regular weekday

A normal day at the campus counter.

Baseline

Mean demand: 100
Std. deviation: 20
Sell: $10
Cost: $4
Salvage: $1

Demand will be revealed only after you commit.

Before calculating: where should Q* sit relative to mean demand?

Below mean At mean Above mean

Order quantity, Q

units

Need a decision hint?

Compare the marginal costs:

C_u = r − c = $6 for a unit you could have sold but did not stock.

C_o = c − s = $3 for a unit ordered but left unsold.

The critical ratio is 0.667. Because this is above 0.5, Q* lies above mean demand. Greater uncertainty makes that upward buffer larger.

Decision log

Your mornings so far

Order, demand, outcome, and profit for completed rounds
Morning	Market brief	Order Q	Demand	Outcome	Profit
Your first result will appear here.

Eight mornings complete

Your policy, explained

Your total profit $0 Decision quality: avg. 0 units from Q* · Predictions: 0/8

Adaptive Q* policy $0 Uses that morning’s optimal order

Mean-demand policy $0 Orders that morning’s forecast mean

Comparing replays: Market scale changes between versions, so compare your distance from Q* and prediction score—not raw profit totals.

The long-run solution

Why Q* changed across mornings

1

C_u = selling price − unit cost

2

C_o = unit cost − salvage value

3

CR = C_u/(C_u+C_o) sets the target demand percentile

4

Q* = μ + z(CR)σ combines the forecast, economics, and uncertainty

Swipe the table to compare your Q with Q*

Brief	μ	σ	CR	Prediction	Your Q	Q*