Trading-card market intelligence

Buy cards below market — before the market notices.

Cardillions watches the major marketplaces and auction houses around the clock, computes the true landed cost of every listing — hammer price, buyer's premium, tax, shipping — and alerts you the moment a card is available under its Card Ladder value, or a player starts moving before the weekly update catches up.

The edge is speed and coverage

Card Ladder is the number the hobby negotiates against — but it updates on roughly a one-week lag, and nobody can watch five auction feeds at once. Software can.

~1 week

The reference lag

A sale takes about a week to show up on Card Ladder and become the new "value." Everything that happens inside that window is edge.

85%

The trade-show floor

Most sellers take ~85% of Card Ladder for cash in person. An online deal only matters if the landed cost beats that floor.

22%

The hidden premium

A $1,000 hammer at Goldin actually costs $1,220. Sticker price is a lie — only landed cost tells you if a listing is under value.

The board: best current buys

The screen the alerts feed — every open opportunity on your watchlist, priced all-in, deepest discount first. Under 75% of value is act-now; under 85% beats the trade-show floor.

What the dashboard shows

Illustrative rows — the live board runs on your watchlist and the day's real listings.

CardWhereLandedValueDiscountCloses
2023 chromium tri-color, PSA 10Whatnot$610$92033.7% offACT NOW45m
2024 flagship silver rookie, PSA 10Goldin$1,086$1,51528.3% offACT NOW1h 10m
2024 flagship base rookie, PSA 10eBay$315$43928.2% offACT NOWBuy now
2019 base rookie, PSA 9Alt$1,176$1,43217.9% offDEALtonight

Landed = price with buyer's premium, tax, and shipping. Every discount is judged against the reference value pinned at detection time, and dual-listed slabs are de-duplicated by grading cert number so one card can't fire twice.

Two alerts. One goal: get in first.

Every live listing and closing auction is matched to its exact card and grade, priced all-in, and compared to the last several Card Ladder sales.

Underpriced listing

Below the floor, all-in

  • Reference value: average of the last several Card Ladder sales for the exact card and grade.
  • Landed cost: price × (1 + buyer's premium) + tax + shipping.
  • Flag under 85% of value — the trade-show floor — with an act-now tier at 75%.
  • Cross-platform de-dup by grading cert number, so a dual-listed slab doesn't fire twice.
Momentum spike

Before the ladder moves

  • Sales velocity and price trend tracked per player and card over rolling 7- and 30-day windows.
  • Flag when either breaks above its recent average by a set margin.
  • Get in before the weekly Card Ladder update reprices the card.
  • Optional signals: game logs, injury news, social mentions, live-stream activity.

Worked example — same card, three prices

Card Ladder value $1,000 (average of the last 5 sales). Trade-show cash price ≈ $850.

WhereHammerPremiumLanded costVerdict
Goldin auction$70022%$854Marginal — even with a trade show
Alt auction$65020%, no tax$78022% under value — alert

Landed-cost calculator

The core math, live. Pick a platform (buyer's premium fills in automatically), enter the price and the Card Ladder value, and see whether it's actually a deal.

Landed cost$780
% of Card Ladder value78%
DEAL — under the 85% floor

Under 75% → act now · under 85% → deal (beats the trade-show floor) · 85–100% → marginal · over 100% → above market.

Beyond discounts: which cards could 3x?

The newest screen doesn't hunt $185 off — it hunts multiples. Cardillions fits the historical growth in each watched card's recorded sales, annualizes it, and extends it twelve months. Bigger cards only, 3x+ flagged.

One-year projection

The uptick, fitted and extended

  • A robust trend fit — the median of every pairwise growth rate between sales — so one shill bid or junk lot can't tilt the line.
  • Annualized and projected twelve months out, with the multiple stated plainly: 2.1x, 3.1x.
  • Bigger cards only: $1,000+ on the fitted value, so small-dollar noise never makes the list.
  • 3x+ potential flagged — the cards worth a deeper look, first.
Shown with its receipts

Every number carries its basis

  • Sales count, history span, and fit noise sit beside every projection — a hot quarter annualized reads very differently from a full year's climb.
  • Too little history — under 6 sales, or under 90 days — and Cardillions shows nothing rather than inventing a number.
  • A straight-line read of recorded sales: not a promise, and not a buy call. The buy math stays on landed cost vs value.
  • The full forecasting model trains on the history the feeds accumulate — this is the baseline it has to beat.

Worked example — three watched cards, one fit

Illustrative sales histories, run through the product's own math.

CardFitted todayTrend / yrIn one yearBasis
2018 flagship rookie, PSA 10$5,264+214%$16,541 3.1x15 sales · 365 days
2003 icon rookie, PSA 10$11,636+109%$24,268 2.1x20 sales · 377 days
2024 phenom parallel, PSA 10$1,493+415%$7,697 5.2x20 sales · 90 days — thin

The Whatnot blind spot

Whatnot runs enormous volume in live-stream auctions, and none of those sales feed into Card Ladder. Cards regularly hammer below the reference number because the buyer pool in a single stream is thin — and the sale never drags the value down afterward. It's the biggest untracked pool of mispriced cards in the hobby, and it's on the Cardillions roadmap.

Marketplace coverage

What the buyer actually pays on top of the hammer — the column that decides whether a listing is really under value.

PlatformBuyer paysNotes
eBay0%Largest volume by far. Official Browse API for live listings.
Alt20% (auctions)Graded cards only, Delaware vault — no sales tax. 0% on fixed price.
Fanatics Collect22%Weekly + premier auctions, vault, Sotheby's partnership on grails.
Goldin22%$19 minimum premium. Premium sports, TCG, and pop culture.
Whatnot0%Live-stream auctions. Not indexed by Card Ladder — the blind spot.
Heritage20%Vintage and six-figure cards; premium applies up to $500k.
MySlabs0%Lowest-fee graded marketplace — peer-to-peer mispricing lives here.
REA · Lelands · Mile High · SCP~20%Vintage-focused houses with overlapping calendars; public archives.

Fee schedules change frequently; numbers are verified against each platform before the engine relies on them.

The first build

Start narrow, prove the matching and the math, then widen.

  1. Inventory feeds. eBay Browse API first; scrapers for Alt, Fanatics Collect, and Goldin auction pages. Whatnot last — highest value, hardest technically.
  2. Value layer. Card Ladder as the anchor, Card Hedger's API as the automated cross-check, Alt's sold-history export as a third signal.
  3. Fee engine. A per-platform table of premiums, tax, and shipping so every listing is compared on landed cost — the calculator above, industrialized.
  4. Alerts. Text or Discord: card, platform, landed cost, Card Ladder value, discount percentage.
  5. Upside projections. The historical uptick in recorded sales, fitted and extended a year out, 3x+ flagged — the first cut is built; the full ML forecaster follows once the feeds accumulate history.
  6. Scope. One sport, a watchlist of 50–100 cards, then widen.