pokemon default-pokemon · 5/16/2026, 5:37:25 PM ↓ csv

EU buy price: GREATEST(avg7, trend) × 1.10 from Cardmarket Price Guide. The +10% uplift compensates for the documented bias — per CM API docs, all averages blend non-English and EX-condition listings, understating true NM-EN buyable price. Reserved List cards get an additional +10% uplift (older-card bias is larger). Floor at €0.50.
Shipping: €1.20/card amortised (post-2026 UPU rules; 20–50 card consolidated parcel; tracked). GST 10% on (price + shipping) × FX. Duty 0% (HS 9504.40 trading cards).
FX: ECB EUR/AUD × (1 − fx_spread_pct). Default Wise spread 0.5%.
AU sell price (multi-signal, in priority order):

1. eBay AU sold median — when ≥5 sold records exist for the card+foil in last 90d. Real transactions, gold standard.
2. Dealer cheapest ask × velocity-aware ratio — strong velocity = 0.92, medium = 0.80, weak = 0.65. Foils −0.03. Velocity-first reframe: if there's evidence the card moves, you'll clear at near-dealer prices (just undercutting slightly). No demand evidence = empirical pessimism (the 0.65 fallback).
3. Buylist-as-floor cross-check — when dealer-ratio is used but a buylist offer exists, implied clearing = buylist / 0.55. If signals disagree by ≥30%, average them.

Velocity category drives the ratio AND the score weight:

• strong — eBay sold ≥5/90d OR active buylist OR EDHREC top-500. Score weight 1.0.
• medium — eBay sold ≥2/90d OR EDHREC top-2000. Score weight 0.7.
• weak — no demand signals; speculative. Score weight 0.4.

Undercut risk (count of competing in-stock AU listings for the same card+foil):

• low — ≤2 competing listings. No score penalty.
• medium — 3–5 listings. −5% score penalty.
• high — 6+ listings. −15% score penalty (price-chase risk).

Score formula: profit × velocity_weight × (1 − undercut_penalty) (replaces the older profit × √(velocity × liquidity)).Illiquidity: base 10% × (1 − x) of effective sell price (moderate risk profile) + 10% extra for cards selling above A$150 (slow-clearing high-value tier).
Channel fees: mtg_mate_retail 15%, ebay_au 15%, mtg_mate_buylist 0%. (15% blends ~13% eBay AU FVF + 5–10% promoted-listings cost − partial offset for unpromoted listings — conservative.) Sell-days: buylist 7, retail 60, eBay 45.
Holding cost: landed × 8% annual × (sell_days/365).
Counterfeit risk: 2% baseline, +3% if >€100, +5% if >€300, +5% Reserved List, +1% private seller >€50.
Sanity filters: drop EU price < €0.50, drop ROI > 500%, drop sell:cost ratio > 8.
Time horizon: the scan config's timeHorizonDays is informational only — it does not filter the leaderboard. Use the days column to spot opps slower than your tolerance.
Annualised ROI: simple — roi × 365 / sell_days. Upper bound; real friction (eBay payout holds, dead stock, capital downtime) typically caps realised at 50–70% of this.
Freshness warning: a card's most-recent reprint compresses prices for 6–12 months as supply lands. Levels: high (reprinted <90d), medium (90–365d), low (1–2y), none (>2y or never). Informational, not used in scoring.
Reality not in the model:

• Multi-seller buy on Cardmarket — €1.20/card amortised assumes one bundled parcel; in practice a 100-card buy splits across 8–15 sellers, each charging €1–€8 minimum shipping. True cost is 20–40% higher than modelled.
• eBay AU sample size — for cards without ≥5 recent sold records, sell price uses the empirical dealer×ratio fallback. Watchlist coverage is currently ~20 cards; expand via scan:ebay-au-diy to firm up estimates.
• Operations time, returns rate, payout holds, reprint/ban events, FX volatility, counterfeit fat-tails — all real, none modelled.

Hover over any row to see the per-opportunity math breakdown.