Compsniper：具備 API 的 130point 替代方案，可查詢 eBay 已售評級卡牌成交資料（2026）

If you need a 130point alternative with an API, you've probably already found the problem: 130point is a lookup website, built for one search at a time in a browser. There's no documented developer API. You can't wire it into a card shop's pricing sheet, a collection tracker or a grading-ROI calculator.

This post shows how to get the same kind of data, eBay sold comps for graded and raw cards, plus the real accepted Best Offer price, from a REST API you can call from Python or JavaScript.

What card people actually need from comps

If you've priced cards for a while, your checklist looks like this:

1. The exact card. Right player, year, set, number and parallel. Not the base version when you searched the refractor.
2. The right grade. A PSA 9 sale in a PSA 10 search ruins a median.
3. What the buyer really paid. On Best Offer sales eBay shows the asking price, not the accepted one.
4. Enough history. Slow cards might sell twice a month, so 90 days is often too thin.
5. Population. How many 10s exist decides whether a 10 is rare or common.

Here's how to get each one.

1. eBay sold comps for a graded card

curl -H "Authorization: Bearer $COMPSNIPER_API_KEY" \
  "https://api.compsniper.com/v1/scrape?keyword=2023+topps+chrome+wembanyama+psa+10&count=240&bestOfferPrice=true"

One request returns up to 240 sold listings with sold price, date, condition, shipping and seller, plus a ready-made summary: median, 25th and 75th percentile, min, max and average shipping.

The part that matters for cards: an AI cleanup step removes wrong matches before the median is calculated. That means other grades, lots, different parallels and "card plus slab plus case" bundles. In a blind 40-search benchmark, 5.4% of the sales it returned were the wrong product, against 25.7% to 31.6% for other eBay sold data APIs tested. Want the raw feed exactly as eBay lists it? Add relevance=false.

2. The real price on Best Offer sales

Card sales close on offers all the time. Add bestOfferPrice=true (as above) and each accepted-offer sale gains acceptedOfferPrice, the price the buyer actually paid.

A real example from October 25, 2026: a PSA 10 Charizard VMAX 074/073 was listed at $349.00 and sold on an offer for $307.00. Without the accepted price, that comp is 12% too high.

3. PSA prices by grade and the population report

For graded cards, PSA's own data adds what eBay can't: prices across every grade and the population.

# Find the card's PSA spec ID
curl -H "Authorization: Bearer $COMPSNIPER_API_KEY" \
  "https://api.compsniper.com/v1/psa/search?keyword=charizard+base+set&count=5"

# Prices by grade, recent sales and the population for that card
curl -H "Authorization: Bearer $COMPSNIPER_API_KEY" \
  "https://api.compsniper.com/v1/psa/card?specId=5217539&count=20"

/v1/psa/card returns:

• priceByGrade: for each grade, the number of sales, average, min, max and latest price
• population: total graded, the gem rate (share of PSA 10s) and the count for every grade
• items: recent sales with grade, cert number, sale type and date

You can filter to one grade with grade=10 (half grades and raw work too). There's also /v1/psa/set for the population of every card in a set in one call.

4. More than 90 days of history

For cards that don't move every day, GET /v1/research/sold goes back up to 3 years, between any two dates:

curl -H "Authorization: Bearer $COMPSNIPER_API_KEY" \
  "https://api.compsniper.com/v1/research/sold?keyword=wembanyama+topps+chrome+psa+10&startDate=2023-11-05&endDate=2026-11-04&sort=lastSoldDescending&limit=50"

It's included on every plan, free too.

Putting it together in Python: a grading ROI check

Is a raw card worth sending to PSA? Compare what raw copies sell for against what PSA 9s and 10s sell for, weighted by the gem rate.

import os
import requests

API = "https://api.compsniper.com"
H = {"Authorization": f"Bearer {os.environ['COMPSNIPER_API_KEY']}"}

def get(path, **params):
    r = requests.get(API + path, headers=H, params=params, timeout=90)
    r.raise_for_status()
    return r.json()

card = get("/v1/psa/card", keyword="charizard base set unlimited holo 4/102")
by_grade = {g["grade"]: g for g in card["priceByGrade"]}
gem_rate = card["population"]["gemRate"]

raw = get("/v1/scrape", keyword="charizard base set unlimited holo 4/102 -psa -bgs -cgc", count=100)
raw_median = raw["summary"]["median"]

psa10 = by_grade.get("10", {}).get("averagePrice")
psa9 = by_grade.get("9", {}).get("averagePrice")
print(f"raw median ${raw_median}, PSA 9 avg ${psa9}, PSA 10 avg ${psa10}, gem rate {gem_rate}%")

Swap in your own grading fee and shipping and you have a real answer in a few lines.

Same idea in JavaScript

const H = { Authorization: `Bearer ${process.env.COMPSNIPER_API_KEY}` };
const get = async (path, params) => {
  const res = await fetch(`https://api.compsniper.com${path}?${new URLSearchParams(params)}`, { headers: H });
  if (!res.ok) throw new Error(`API error ${res.status}`);
  return res.json();
};

const comps = await get("/v1/scrape", {
  keyword: "2023 topps chrome wembanyama psa 10",
  count: "240",
  bestOfferPrice: "true",
});
console.log("median", comps.summary.median, "from", comps.items.length, "sales");

Beyond eBay

Graded cards also sell at auction houses. The same key returns realized prices from Goldin, Fanatics Collect, Pristine Auction and MySlabs (/v1/goldin/sold, /v1/fanatics/sold, /v1/pristine/sold, /v1/myslabs/sold), and TCGplayer sold history for TCG cards (/v1/tcgplayer/sold).

Try it

Get a free key at compsniper.com (100 requests a month, no card). More detail: Sports card price API, Pokemon card price API and PSA API docs. Code: ebay-sold-listings-api on GitHub.

What's the one comp you always double-check by hand before you trust it: grade, parallel or Best Offer price?