Skip to main content

pkmnprices

Python client for the Pkmn Prices API. Pokemon TCG card pricing from TCGPlayer, Cardmarket, and eBay.

Sync and async clients, both built on httpx. Typed responses, typed errors, and iterators that page through results for you. Python 3.10+.

Install

pip install pkmnprices

Usage

from pkmnprices import PkmnPrices

client = PkmnPrices("pk_your_key_here")

page = client.cards.list(name="charizard", per_page=10)

card = client.cards.get(page.data[0].id)
for price in card.prices:
    symbol = "€" if price.currency == "EUR" else "$"
    print(f"{price.source}: {symbol}{price.market_price}")

client.close()

The client is also a context manager:

with PkmnPrices("pk_...") as client:
    health = client.health()

Async

import asyncio
from pkmnprices import AsyncPkmnPrices

async def main():
    async with AsyncPkmnPrices("pk_...") as client:
        page = await client.cards.list(name="charizard")
        async for card in client.cards.iterate(name="charizard"):
            print(card.name)

asyncio.run(main())

Get an API key from https://pkmnprices.com/dashboard.

Options

PkmnPrices(
    "pk_...",         # API key, sent as the x-api-key header
    max_retries=2,    # retries on 429 rate limits and 5xx/network errors
    timeout=30.0,     # per-request timeout in seconds
)

Rate-limit 429s are retried with backoff. Credit-limit 429s (credit_limit_exceeded) are not, since they don't reset until the next day.

Pagination

List endpoints return a Page (.data, .pagination). Listing endpoints (eBay, Cardmarket, and TCGplayer) return a CursorPage. Both resources expose iterators so you don't track pages or cursors:

for card in client.cards.iterate(name="charizard"):
    print(card.name)

all_sets = client.sets.list_all(language="english")

for sale in client.cards.listings.iterate_ebay(789, graded=True, grader="PSA", grade="10"):
    print(sale.title, sale.price)

for offer in client.cards.listings.iterate_cardmarket(789, condition="Near Mint", variant="Reverse Holo"):
    print(offer.seller, offer.price, offer.language)

for offer in client.cards.listings.iterate_tcgplayer(789, condition="Near Mint"):
    print(offer.seller_name, offer.price, offer.shipping_price)

TCGplayer offers default to price_asc, item price alone, so the first row can be a low-price, high-shipping offer that is not the cheapest to buy. total_asc and total_desc order by price + shipping_price, which is how TCGplayer's own site orders offers:

for offer in client.cards.listings.iterate_tcgplayer(789, sort="total_asc"):
    print(offer.price + (offer.shipping_price or 0), offer.seller_name)

A cursor is bound to the sort family it was issued under, so do not reuse a total_* cursor with a price_* sort or the reverse. min_price and max_price filter item price under every sort.

Sealed products carry the same three listing sources, under client.sealed.listings:

for offer in client.sealed.listings.iterate_tcgplayer(5678):
    print(offer.seller_name, offer.price, offer.quantity)

for offer in client.sealed.listings.iterate_cardmarket(5678):
    print(offer.seller, offer.price, offer.opened)

for sale in client.sealed.listings.iterate_ebay(5678, sort="price_desc"):
    print(sale.title, sale.price, sale.sold_at)

Sealed Cardmarket offers take no variant: every sealed row is written with an empty one, so the filter could only ever exclude everything. They are also single-language — the offers returned are the ones in the set's own language, so a Cardmarket product page showing offers in other languages lists more than this endpoint does.

Sealed TCGplayer offers are normally condition "Unopened" with an empty printing, so those two filters rarely narrow anything. Sealed eBay sales are never graded, so graded, grader, and grade aren't accepted there and grader/grade come back None. The async client mirrors all of these on AsyncPkmnPrices.

Which card a comp is really about

A variant and its base card can map to one source product page. When that happens, both serve the same sales, and the titles describe whichever printing the seller actually sold. Every eBay comp says which case it is:

for sale in client.cards.listings.iterate_ebay(17679):
    if sale.attribution == "shared":
        continue  # another card's evidence
    print(sale.variant, sale.title, sale.price)

exact means the source page belongs to this card alone. shared means the sale appears under at least one other card too, so it prices the group rather than this entity. unknown means no printing is recorded for the comp: it was collected before printings were recorded, or its title did not narrow the sale to one printing. A feed that is entirely shared is not evidence about the card you asked for.

variant also works as a filter, so a card mapped for more than one printing can be read one printing at a time:

holo = client.cards.listings.all_ebay(789, variant="Holofoil")

Two grades can print the same number

A CGC Pristine 10 and a CGC Gem Mint 10 both carry grade == "10", and so do a BGS Black Label 10 and a plain BGS 10. The higher tier sells well above the lower one, so grade_qualifier tells them apart: "Pristine", "Black Label", or None for the tier with no name of its own, which is nearly every comp. Filtering grade="10" returns every tier. To get one tier, add grade_qualifier:

black_labels = client.cards.listings.all_ebay(
    789, grader="BGS", grade="10", grade_qualifier="black_label",
)
plain_tens = client.cards.listings.all_ebay(
    789, grader="BGS", grade="10", grade_qualifier="none",
)

"none" also returns comps whose tier was never recorded.

Comp currency, seller country and sale type

eBay comps carry their own currency. Card comps from ebay.com are USD, and German and Spanish card comps from ebay.de and ebay.es are EUR, so read it per row rather than assuming dollars. Sealed comps are always USD.

seller_country is the seller's location as an ISO 3166-1 alpha-2 code ("US", "JP", "GB"), or None when it isn't known. Cardmarket offers carry it too, read from the offer's item-location flag. Both listing feeds accept the same two filters, as a list or a comma-separated string of up to 30 codes:

domestic = client.cards.listings.all_ebay(789, seller_country=["US", "PR"])
no_china = client.cards.listings.all_cardmarket(789, exclude_seller_country="CN,HK")

seller_country leaves out rows with no known country; exclude_seller_country keeps them. A malformed code raises BadRequestError.

sale_type says how an eBay comp sold: "auction", "fixed_price" or "best_offer_accepted". For an accepted offer, price is what the buyer paid, not the list price. It is None on older comps, comps mirrored from PriceCharting and sealed comps, and some older accepted offers still carry the list price with nothing marking them. Any sale_type filter drops the None rows, so passing all three values keeps only comps whose sale type is known:

auctions = client.cards.listings.all_ebay(789, sale_type="auction")
known = client.cards.listings.all_ebay(
    789, sale_type=["auction", "fixed_price", "best_offer_accepted"],
)

Polling for new comps

Credits are charged per row returned, so re-reading a page of comps you already hold to find out that nothing changed is the expensive way to stay current. since returns only what arrived after a point you name:

checkpoint = "2026-09-01T02:40:15.126147Z"

page = client.cards.listings.ebay(789, since=checkpoint)
for sale in page.data:
    print(sale.title, sale.price)
if page.data:
    checkpoint = page.data[0].ingested_at

Checkpoint on ingested_at, not sold_at. They are different: sold_at is when the sale happened, ingested_at is when we collected it, and a collection run regularly brings in sales that are weeks old. A sale-date bound would step over those permanently.

The bound is exclusive, so passing back the ingested_at you were given never repeats that row. since also accepts a bare YYYY-MM-DD (midnight UTC), which suits a backfill more than a poll: re-running it the same day returns the same rows, and pays for them again.

A card with nothing new returns an empty data list.

TCGplayer freshness

updated_at on a TCGplayer offer is not a freshness signal. It moves only when that listing's own price, quantity or seller details change, so an offer that has been live and unchanged for a month keeps a month-old updated_at however recently it was confirmed.

snapshot_at is the freshness field: when that product's listings were last confirmed against TCGplayer. It is the same for every row in a response, since a snapshot replaces a product's listings wholesale. Listings refresh daily, so a snapshot_at well over a day old means that product's last fetch did not succeed and you are looking at the previous snapshot.

One thing snapshot_at cannot tell you: listings are collected from a US vantage point with no shipping-destination filter, so a response can contain offers TCGplayer's own site hides from you when you browse it from elsewhere.

Cardmarket special attributes

Cardmarket sells more than one kind of good under a single card. Every Cardmarket offer carries four booleans, and they are always present:

for offer in client.cards.listings.iterate_cardmarket(789):
    if offer.graded:
        print(offer.grader, offer.grade)  # "PSA", "10"
    if offer.signed or offer.altered:
        continue  # not a clean card

A signed, altered, graded or opened offer is real, and it is returned, but it does not contribute to the market price. A signed and altered Near Mint copy at EUR 200 must not set the Near Mint price of a card whose clean copies sell for EUR 3,800, and a slab is priced for the slab rather than for the card.

The practical consequence: the cheapest row you get back is not necessarily the market_price. Filter these out before deriving a price yourself.

grader and grade are named to match the graded eBay sale shape, so "PSA 10" reads the same whichever source it came from. Both are None unless graded is true, and can be None even then — Cardmarket flags a slab without always naming the grader, and the details are read from free-text seller comments.

opened is the sealed-product member of the set: the seller's own comment says the item is not sealed. It is always False on a card. A Jungle booster pack whose every other offer sits between EUR 799 and EUR 950 carries one at EUR 10.00 reading "Not Sealed (open, just the booster)" — a real price for an opened pack, not a EUR 10 Jungle booster pack.

sell_count

The other reason the cheapest row may not be the price. sell_count is the seller's completed sales, and an offer from a seller with no completed sales does not set a price.

for offer in client.sealed.listings.iterate_cardmarket(1430):
    if offer.opened:
        continue  # not the sealed product
    if offer.sell_count == 0:
        continue  # no completed sales; sets no price
    # offer.sell_count is None means no count was recorded -- not zero sales

None is not zero. None means no count was recorded for that row; 0 means Cardmarket reports the seller as having sold nothing. Treat only 0 as disqualifying — and note that if not offer.sell_count catches both, which is the mistake this distinction exists to prevent.

Languages

A card's language comes from its set, and it decides what pricing that card can ever have.

Language Cards Pricing Plan
English 28,158 USD (TCGplayer, eBay) + EUR (Cardmarket) Free
Japanese 29,660 USD + EUR Pro+
German 13,078 EUR (Cardmarket) only Pro+
Spanish 12,419 EUR (Cardmarket) only Pro+
german = client.cards.list(language="German", currency="eur")

Spelling is normalised: "German", "german", "de" and "DE" all resolve to the same thing, and responses come back in the canonical form ("German"). Spanish takes "Spanish", "Español", "espanol", "spa" and "es".

German and Spanish cards have no USD price and never will, because TCGplayer sells neither. Asking for either with currency="usd" returns an empty list rather than an error, so reach for "eur".

The same rule scopes Cardmarket listings. cardmarket() returns offers in the card's set language only, so an English card never returns German or Spanish offers. cardmarket.com shows every language on one product page, so the cheapest offer here is usually higher than the "From" price a browser shows you. That is this feed being narrower, not a different product.

A free key is limited to English. Asking for Japanese, German or Spanish raises ForbiddenError, and omitting language returns English only rather than the whole catalogue.

German coverage runs from HeartGold & SoulSilver (2010) to current sets, and Spanish from Black & White (2011). Spanish price history only goes back to when Spanish pricing started, so it is shorter than the other languages'. A Spanish card with no Cardmarket offer in Spanish yet has no price, which is missing data, not a zero.

Currency

Every price has a currency field. Pass currency="usd" or currency="eur" to filter, or leave it off to get everything your plan allows. EUR (Cardmarket) prices need a Pro plan; a free key asking for eur raises ForbiddenError.

card = client.cards.get(789, currency="usd")
box = client.sealed.get(5678, currency="eur")

Cardmarket current prices are condition- and printing-specific marketplace prices. Each EUR row has one market_price for its exact condition and variant; for example, a Near Mint Reverse Holofoil price is distinct from a Mint or Normal price. The retired Price Guide low, trend, and avg fields are not returned. Live Cardmarket listings are automatically restricted to the card's language.

Cardmarket Mapping

Card and sealed detail responses expose Cardmarket's stable product identifiers when a mapping is available:

card = client.cards.get(789)
print(card.cardmarket_url)
print(card.cardmarket_product_id)

box = client.sealed.get(5678)
print(box.cardmarket_url)
print(box.cardmarket_product_id)

Both fields are None until the product has been mapped.

Errors

Everything raised subclasses PkmnPricesError, which carries status, code, docs_url, rate_limit, and retry_after. docs_url is the docs page for that class of error as sent by the API, so a 401 points at authentication, a 403 at pricing and a 429 at rate limits; it is None when the response had no body to read it from.

from pkmnprices import ForbiddenError, NotFoundError, RateLimitError

try:
    client.cards.get(789, currency="eur")
except ForbiddenError:
    ...  # needs a higher plan
except NotFoundError:
    ...  # no such card
except RateLimitError:
    ...  # ran out of retries

A 401 comes back with code set to missing_api_key when no key was sent and invalid_api_key when the key doesn't match one, both as UnauthorizedError.

Subclasses: BadRequestError (400), UnauthorizedError (401), ForbiddenError (403), NotFoundError (404), ConflictError (409), CreditLimitError and RateLimitError (429), InternalServerError (5xx), APIConnectionError (network/timeout).

License

MIT

Metadata

Release files for pkmnprices 4.1.0

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for pkmnprices 4.1.0
File Size Uploaded Provenance
pkmnprices-4.1.0.tar.gz 23.1 kB With provenance Details

Built distribution (wheel)

Table of built distributions (wheels) for pkmnprices 4.1.0
File Interpreter ABI Platform
pkmnprices-4.1.0-py3-none-any.whl Python 3 none any Details

Total release size: 45.9 kB

Release files / pkmnprices-4.1.0.tar.gz

Download URL pkmnprices-4.1.0.tar.gz
Size 23.1 kB
Tags Source
SHA-256 checksum
How to use checksums
1f1642d0958849c463b1e796cfa758386de5ede0422b6bc94f1653e927caf18b
BLAKE2b-256 checksum
How to use checksums
f30b096cbe77601250e876f76fd7e4bb632aa07d8786d3a02d12452013c12a3a
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.14

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Oct 18, 2026.

Transparency log

Release files / pkmnprices-4.1.0-py3-none-any.whl

Download URL pkmnprices-4.1.0-py3-none-any.whl
Size 22.8 kB
Tags Python 3
SHA-256 checksum
How to use checksums
487de32f3addc719de0e5e8d163c1cb67fdf70f97817f8c81aa8ffa3f98904f0
BLAKE2b-256 checksum
How to use checksums
f03090e4524dfdf357c45a27537baded67ecbfea61ffee55fa910f3da2006f09
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.14

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Oct 18, 2026.

Transparency log

Security for pkmnprices 4.1.0

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

Full provenance

All 2 files in this release have been uploaded with provenance metadata.

Select from 2 files

Attestation for pkmnprices-4.1.0.tar.gz

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Oct 18, 2026.

Transparency log

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.

To learn more about protecting yourself from malicious actors, visit our package security documentation.

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page