The proof
Sample receipts — live backtest pending dataWe give the model only what was knowable before launch and ask it to separate winners from losers. Here is the record against real sales-rank outcomes. This is the number no backward-looking tool can produce.
4/5
correct calls (sample)
| Product | Prediction | Tier | Real outcome | Result |
|---|---|---|---|---|
| Blind-box collectible line | Winner | Top decile | Rank rose to top 3% in 90 days | ✓ hit |
| Generic wooden stacking set | Loser | Bottom half | Rank decayed, delisted at reset | ✓ hit |
| Licensed character playset | Winner | Top quartile | Sustained top-quartile rank | ✓ hit |
| Novelty gimmick toy | Viral spike | Fast then fade | Spiked, faded in 60 days | ✓ hit |
| Mid-tier building set | Winner | Top quartile | Underperformed — held mid-pack | ✕ miss |