ListRight
CERT No. 0000001 · FOUNDING SERIES

Price the exact card.
Not the look-alikes.

Photograph a card. ListRight drafts the eBay title, pulls the real sold comps, and shows every comp it kept or rejected — so you review the price before you list.

ListRight is web-based subscription software for trading-card sellers. It is a listing-preparation tool — it does not buy, sell, ship or process payment for any card, and no physical goods are shipped.

Real engine output, cached run Jul 25 2026 — 40 comps fetched, 12 graded slabs excluded (graded median $140), 28 raw sales kept, median $77.50. Not a customer result. Card art is unbranded on purpose; the audit report names the real card.

Start free — 5 cards, no card needed See the $19 founding seat

11 things no other card tool does

  • Priced off eBay sold listingsNot a market feed you're asked to trust
  • Reads accepted Best OffersThe prices eBay hides from everyone else
  • Shows the comps it rejectedStruck through, with the reason
  • Strips multi-card lotsA lot is not a comp for your single
  • Separates graded from rawNever averaged together
  • Serial parallels matchedA /99 is priced off /99 comps — or it tells you
  • Title tokens by rule, not guess75 unit tests behind them
  • 1st Bowman is never tagged RCThe mislabel that costs money both ways
  • Budgets the 80-char limitBefore eBay truncates it for you
  • Median, IQR, n and confidenceNot one number with no working shown
  • Writes the description tooNo external links — eBay policy
§ 01How it compares

Eleven things other
card tools don't do

You have options. Most of them scan a card and hand you a number — the scanning was never the hard part. Here is exactly where they stop and this keeps going.

$53Gap on one mispriced card
40Sold comps pulled per card
Most card tools ListRight
Price sourceA market feed or their own databaseeBay sold listings — what buyers actually paid
Accepted Best OffersHidden by eBay, so invisibleRead and included
Multi-card lotsCounted as compsStripped before pricing
Graded vs rawMixed togetherSeparated
Serial parallelsIgnored — priced like base cards/99 comps for a /99 card — flagged when comps are thin
Title tokensAI-written or typed by handRule-based and unit-tested
1st BowmanOften tagged as a rookieIts own token — never RC
Rejected compsNot shownShown, struck, with the reason
DescriptionYou write it, or paste a templateWritten for you, policy-safe
The priceOne numberMedian, IQR range, n, confidence
Character limitYou find out when eBay truncatesBudgeted to 80 before you see it

Those title rules aren't a claim — they're 75 tests, and they read like the hobby talks:

test_first_bowman_adds_1st_before_brand
test_no_rc_token_when_not_a_rookie
test_print_run_tolerates_full_serial_and_punctuation
test_variety_not_duplicated_when_brand_contains_it
test_listing_title_keeps_grade_even_with_long_brand
test_listing_description_has_no_urls

That last one exists because eBay treats external links in a listing as a policy violation.

§ 02The output

What comes back on every card

Three things, ready to paste — and the comp table that justifies the number, including the rows it refused to use.

Reads the card
2018 Panini Prizm
#280 Luka Doncic
Silver Prizm · RC
Gem MT 10
CERT 47182930
01eBay title80-char limit enforced
2018 Panini Prizm #280 Luka Doncic Silver Prizm RC PSA 10 Gem Mint
66/80
02PriceFrom sold comps, shown below
DatePriceSold listing
2026-07-11$96.002018 Prizm Luka Doncic Silver #280 PSA 10
2026-07-04$88.50Luka Doncic 2018 Panini Prizm Silver RC PSA 10
2026-07-02$31.002018 Prizm Luka Doncic Silver #280 — raw Raw · not PSA 10
2026-06-28$84.00Doncic Silver Prizm rookie PSA 10 Gem Mint
n=14 · MEDIAN $84.00 · IQR $71–$96 · 90D · 2 EXCLUDED
03DescriptionNo external links — eBay policy
2018 Panini Prizm #280 Luka Doncic — Silver Prizm Rookie Card, graded PSA 10 Gem Mint. Certification number 47182930, verified against PSA's population report: 1,842 graded PSA 10, none higher. Card is sealed in its original PSA holder and has not been removed. Ships double-boxed with tracking and insurance.

Price a PSA 10 off raw sales and you list an $84 card at $31. That struck-through row is one ListRight refused to price on — graded and raw are never averaged together.

Illustrative example of the token rules and comp handling — not a customer result.

§ 03The title

Buyers search "Auto." Not "Autograph."

Miss the token a buyer actually types and your card doesn't appear in the search that would have sold it. Every card app writes a title. ListRight writes it from rules — the same value tokens, in the hobby's order, inside the 80-character limit, every single time.

/250Print run. The denominator is the information — a /25 is a different card from a /250. Always in the title.
RCRookie card. Only when the card actually carries the RC mark. A 1st Bowman prospect is not a rookie — a mistake that costs money in both directions.
1st BowmanFirst Bowman. Its own token, placed before the brand so it reads "1st Bowman Chrome."
PSA 10Grade. The single strongest value token on a slab, and it leads the title.
1/1One of one. A different class of object from a /250 — and never styled like one.
§ 04The price

Priced off sold comps.
Not a "market price."

What everyone else does

Card tools price off a market feed — TCGplayer, or their own database. It's a number, and you're asked to trust it.

What ListRight does

Reads what the card actually sold for on eBay. Strips lots, separates graded from raw, drops outliers, and gives you a median with a range — then shows you every comp behind it.

Including the ones it threw out. That hollow row in the example above — struck price, "raw — not PSA 10" — is a comp ListRight refused to price on. No other tool shows you its own rejects.

§ 05What runs on every card

Seven stages.
No guessing at any of them.

This is the whole pipeline. Nothing here is a language model being asked politely to produce a good title — the identification is structured, and everything after it is deterministic code with tests.

  1. 01 · ReadFront and back. HEIC and JPEG, EXIF rotation corrected, decompression-bomb guarded, downscaled to 1600px before anything sees it. The back resolves year, set and serial the front can't.
  2. 02 · IdentifyStructured extraction, not prose. Twenty fields come back typed — sport, year, brand, set, player, team, card number, parallel, insert, serial, print run, grader, grade, cert — each with a confidence. The model fills a schema; it doesn't write a sentence and hope.
  3. 03 · TokensRules override the model. A legible serial derives the print run whether the card reads 074/250, /250 or 250. A 1st Bowman prospect is forced to not be a rookie, because it isn't one — a distinction most tools get wrong in both directions.
  4. 04 · Title & descriptionAn 80-character budget, spent by priority — plus the description. Grade, player and year are never dropped. Then RC, 1st Bowman and the print run — short, high-signal — outrank the long brand string and the card number. Duplicate variety text is removed when the brand already contains it, and ALL-CAPS grader fields are re-cased without breaking RC, PSA, SP or SSP. The description is written alongside it, with population context and no external links — eBay treats those as a policy violation.
  5. 05 · CompsReal eBay sold listings — 40 pulled per card, including accepted Best Offer prices that eBay hides from you. That's the truest record of what a card actually trades for, and most tools never see it.
  6. 06 · SegmentThe step that decides whether the number means anything. Multi-card lots stripped out. Graded separated from raw. Serial-numbered parallels matched to serial-numbered parallels. Currencies split. Stale sales aged out. A comp set that skips this is just an average of unrelated cards.
  7. 07 · PriceMedian, not mean — one shill sale can't drag it. Range is the interquartile band, so the middle half of real sales. Confidence is computed from sample size and spread, and it's always shown with the numbers behind it, never as a bare adjective.

Graded slabs additionally verify against PSA's public API — identity and population, by cert number.

§ 06The proof

The same card, titled two ways

Typical2023 Bowman Chrome Jackson Holliday Refractor Auto50 ch
ListRightPSA 10 Jackson Holliday 2023 1st Bowman Chrome Auto Refractor /49966 ch
Typicalmaui ahuna bowman auto22 ch
ListRight2023 Bowman Draft Chrome Maui Ahuna 1st Bowman Autograph #CDA-MAH SF Giants75 ch
TypicalLuka Doncic Prizm Silver rookie graded38 ch
ListRight2018 Panini Prizm #280 Luka Doncic Silver Prizm RC PSA 1057 ch

Illustrative pairs showing the token rules. The Ahuna title is a real engine output (live run, 2026-07-28). Not customer results — there aren't any yet.

§ 07The membership

Founding series

Try five cards free, no card required. If it earns a seat, twenty-five people get it at $19 — cheaper than it will ever be again, permanently — and get to tell me what to build next.

Founding Membership
500 Scans · Monthly
Was $39 · Locked for life
USD 19
SERIES 1 · 25 SEATS
Claim a founding seat — $19/mo

Past 500 cards in a month? Email me and I'll raise your cap — I'd rather price it right than have you hit a wall · 5 cards free first, no card required · Cancel anytime · Refund in the first 30 days

DELIVERY · ListRight is web-based software. Access is delivered electronically and is available immediately after successful payment, through your account on listrightapp.com. No physical goods are shipped. Subscription renews monthly at $19 until cancelled; cancel any time.

Honest status: the engine works and is what I use on my own cards. The hosted version is early — you'll be using a beta and talking to me directly. It does not publish to eBay or Whatnot yet; you copy the title and price into your listing. That's next.

§ 08Not yet

Get the launch note

One email when it opens. Nothing else.

Email me when it opens

Would you use this one?

Deciding what to build after launch. One candidate: measure centering, flag corner and edge defects, then tell you whether paying to grade is worth it against what the graded comps actually sell for. If you'd use it, say so — that's how I'll decide.

I'd use "should I grade this?"