When a product goes out of stock, the conventional assumption is that the sale is simply lost. Demand transference challenges that assumption. A predictable and quantifiable portion of that lost demand migrates to substitute products within the same assortment — and with the right analytical framework, that behavior can be measured, modeled, and acted on.
This project presents a full demand transference analysis applied to New York Yankees licensed apparel, using 18 months of illustrative sales and inventory data across four core SKU families and three retail channels.
Note: All data in this project is illustrative and intended to demonstrate the methodology and analytical framework. It is not reflective of actual Yankees or MLB retail results.
Licensed sports apparel is one of the most demand-volatile retail categories. Sales spike around Opening Day, playoff runs, player events, and national broadcasts — and stockouts tend to occur precisely at peak demand, the worst possible time.
Key question this analysis answers:
When a Yankees Home Pinstripe Jersey in size XL goes out of stock, where does that demand actually go — and how much revenue can be recovered?
| Metric | Value |
|---|---|
| OOS Episodes Analyzed | 847 |
| Average Transfer Rate | 62% |
| Average Lost Demand | 38% |
| Total Recovered Revenue (18 months) | $1,257,155 |
| Total Truly Lost Revenue (18 months) | $647,287 |
| Estimated Recovery Opportunity | $173,000 – $230,000 |
- Jersey-to-Jersey substitution captures 20–42% of lost demand — the highest transfer rate observed
- Replica Jersey Tees serve as a demand floor, capturing 12–18% of overflow from OOS jerseys
- In-stadium retail has the highest transfer rate (71%) vs. e-commerce (58%), reflecting in-person purchase urgency
- Playoff periods show the highest transfer rates (79%), making pre-emptive substitute stocking highest-ROI during postseason
Transfer Rate (A → B) = ΔSales_B during OOS_A ÷ Lost_Demand_A
- ΔSales_B — incremental sales uplift of substitute product B during the OOS window of product A
- Lost_Demand_A — estimated unfulfilled demand, modeled via SARIMA baseline forecast
A SARIMA (Seasonal AutoRegressive Integrated Moving Average) model is calibrated against 12 months of pre-event history, with adjustments for:
- Day-of-week and game-day effects
- Promotional calendar (playoffs, All-Star, special events)
- Channel traffic patterns (in-stadium vs. online)
- Comparable non-OOS benchmarks
Substitute uplift is credited to transference only when the correlation coefficient between OOS timing and uplift timing exceeds 0.70, ensuring signal integrity. Rates are aggregated using a weighted average across all OOS episodes per product pair.
Transfer rates are intentionally asymmetric — the rate from Product A → B will not equal B → A. This reflects real consumer behavior: transfer depends on the purchase intent of the customer who encountered the stockout, not simply the product relationship. Symmetric rates in a transfer matrix are a modeling red flag.
| OOS Product | Home Pinstripe Jersey | Away Gray Jersey | Road Alt. Jersey | Replica Jersey Tee |
|---|---|---|---|---|
| Home Pinstripe Jersey | — | 38% | 20% | 18% |
| Away Gray Jersey | 42% | — | 22% | 14% |
| Road Alt. Jersey | 35% | 28% | — | 12% |
| Replica Jersey Tee | 22% | 19% | 8% | — |
Darker = higher substitution rate. Values represent 18-month weighted averages.
| Period | Avg. Transfer Rate | Lost Demand | Inventory Risk |
|---|---|---|---|
| Opening Day +2 Weeks (Mar–Apr) | 74% | 26% | 🔴 Critical |
| Regular Season (Apr–Sep) | 63% | 37% | 🟠 High |
| Playoffs (Oct) | 79% | 21% | 🔴 Critical |
| Off-Season (Nov–Feb) | 48% | 52% | 🟡 Moderate |
| Special Events | 41% | 59% | 🔴 Extreme |
demand-transference/
│
├── README.md ← You are here
│
├── whitepaper/
│ └── DT_WhitePaper.pdf ← Full white paper
│
├── sql/
│ └── yankees_demand_transference.sql ← PostgreSQL DDL + sample data + 8 queries (in progress)
│
├── dashboard/
│ └── Yankees_Demand_Transference_Dashboard.xlsx ← Excel dashboard (in progress)
│
└── images/
├── sankey_demand_flow.png ← Demand flow diagram
└── transfer_rate_matrix.png ← Heat map matrix
The PostgreSQL schema includes 7 tables:
| Table | Description |
|---|---|
products |
SKU catalog with family, size, and pricing |
channels |
Retail channels (stadium, store, e-commerce) |
seasons |
MLB calendar periods with risk classifications |
inventory_snapshots |
Daily on-hand inventory with auto-computed OOS flag |
sales_transactions |
Daily sales by SKU and channel |
oos_events |
Discrete stockout episodes with estimated lost demand |
transfer_rates |
Aggregated substitution rates with affinity scores |
- Transfer Rate Matrix — cross-SKU family substitution rates
- OOS Episode Summary — duration, lost demand, lost revenue
- Revenue Recovered vs. Truly Lost — by SKU family
- Substitute Uplift Detection — compares sales during vs. pre-OOS
- Seasonal Transfer Rate Analysis — by period and channel
- Top Substitute Pairs — ranked by affinity score
- Channel Revenue Recovery — transfer rates by retail channel
- Asymmetry Check — flags any symmetric rate pairs as modeling errors
Five-tab workbook aligned to the white paper:
| Tab | Contents |
|---|---|
| Dashboard | Master view — KPI tiles, transfer matrix, financial summary, seasonal & channel breakdowns |
| Transfer Matrix | Full 12-row rate table with affinity scores, substitute tiers, asymmetry check |
| OOS Events | 12 sample OOS episodes with dates, duration, lost demand and revenue |
| Financial Summary | Revenue recovered vs. lost by SKU family + recovery opportunity analysis |
| SQL Query Reference | Quick-reference card mapping analytical questions to queries |
- Implement a Transfer Rate Dashboard — Track rates by SKU pair, channel, and season; alert when high-velocity SKUs drop below 21-day cover
- Define Substitute Pairs for All High-Velocity SKUs — Designate top-2 substitutes per SKU to drive automatic safety-stock uplift rules
- Run Seasonal Inventory Reviews Tied to the MLB Calendar — 60-day pre-season reviews with worst-case OOS scenario modeling before peak periods
Personal portfolio project — illustrative analysis demonstrating demand transference methodology applied to licensed sports apparel retail.
This project is for portfolio and educational purposes. Data is illustrative and not reflective of actual New York Yankees or MLB retail operations.