Yahoo! Auctions Japan Deals — Below Past Sold Prices

What it does: Works out what your keyword usually sells for on Yahoo! Auctions Japan (median of ended auctions with a bid), then returns live auctions at or below 80% of it.

You enter: keywords, e.g. Nikon Z6 (add a minimum price like ¥50,000 so accessories don't skew it); threshold

You get: per keyword: sold median (sold comps) and range, how many sold auctions it rests on, live listings checked, deal count and rate — 120 of each by default, configurable up to 600. Per deal (on by default): price now, buy-now price, bids, end time, discount, and whether it is payable today.

Price: $0.02 per keyword with a usable sold-price basis + $0.002 per individual deal delivered; deals are on by default. Disable includeIndividualDeals for the summary only. Run charge limits stop paid delivery and produce a free notice. A keyword whose sold prices are too few or too mixed is refused and not charged; a keyword with a usable median is charged even when it finds no deals.

Example: enter Nikon Z6 with a ¥50,000 floor → sold median ¥146,696 from 60 sold auctions · 28 deals among the 60 live listings checked · best 50.2% below the median (real run, 2026-08-09)

Unofficial — not affiliated with Yahoo! Auctions. Reads public pages only.

How the discount is worked out

1. Read ended auctions for the keyword — only ones that received a bid, because an ended auction with no bids is an asking price nobody accepted. Their median is the basis (120 sold auctions by default, up to 600 with maxSoldComps). To bound sparse searches, collection stops after the page that reaches four times this limit in source rows, even if fewer sold comps were found.
2. Read the live listings (120 by default, up to 600 with maxLiveListings).
3. discountPct = (1 - price / soldMedian) x 100.
4. Return everything at or below your threshold (80% of the sold median by default), plus the price spread.

Two guards, because a basis can lie in two ways

Too few sold prices. minComps (default 8) refuses a keyword whose median rests on a handful of sales.

Too many kinds of thing. A camera search also returns its batteries, cases and manuals. maxCompSpread (default 4.0) compares the top and the bottom of the middle 50% range of the sold prices, and refuses a keyword whose two ends are more than 4x apart, pointing at the fix — an input like this:

{ "keywords": ["Nikon Z6"], "priceMinJpy": 50000 }

Both sides use the same inclusive price band. Ended auctions are read from allowed keyword/offset pages and filtered locally; live auctions use Yahoo's permitted price selectors. Yahoo declares the actual page width. Auction IDs are deduplicated before building the sold-price basis.

With a band, soldListingsFound is an exact matching ended-auction count only after a complete, consistent source scan; otherwise it is null. It counts ended auctions, including ones with no bids; soldCompCount counts the bid-bearing, priced sample used for the median. soldListingsScanned and soldAcquisition show scan reach, duplicate counts, stop reason and whether the total is exact. A bounded sample containing no matches does not prove that the whole keyword has no matches. Saved inputs, limits, fee events and the original price/deal fields remain supported; clients that assume soldListingsFound is always an integer must also accept null.

basisMedianJpy is a summary-only display alias for the usable soldMedianJpy. It keeps the overview separate from individual deal rows; the original median field remains present.

In both cases the run returns basisUsable: false with the reason and that keyword is not charged. mixedKeywordLikely flags the remaining tell: in a coherent market only a small share of the live listings sits far under the median, so a high dealRatePct means the keyword is still mixed.

What comes back

One summary row per keyword, then (on by default) one row per deal. The following is a historical sample, not a current price quote or a guarantee of today's counts:

Measured on 2026-08-09 (real run) — { "keywords": ["Nikon Z6"], "priceMinJpy": 50000 }

{
  "type": "deal_summary",
  "keyword": "Nikon Z6",
  "basisUsable": true,
  "soldCompCount": 60,
  "soldListingsFound": 327,
  "soldMedianJpy": 146696,
  "soldPriceJpy": { "min": 69501, "p25": 99122, "median": 146696, "p75": 236250, "max": 294000, "average": 163864 },
  "liveListingCount": 60,
  "liveListingsFound": 90,
  "dealThresholdPct": 80,
  "dealCount": 28,
  "actionableDealCount": 28,
  "dealRatePct": 46.7,
  "mixedKeywordLikely": true,
  "bestDeal": {
    "type": "deal",
    "id": "l1237151687",
    "title": "Nikon Z6 ボディY267-Y3",
    "priceJpy": 73000,
    "buyNowPriceJpy": 73000,
    "bidCount": 0,
    "endTime": "2026-08-10T12:12:48+00:00",
    "discountPct": 50.2,
    "actionable": true,
    "priceIsFloor": false
  },
  "malformedCards": 0,
  "duplicateCardsDropped": 1,
  "checkedAt": "2026-08-09T17:31:41.046207+00:00"
}

actionable is the one to read: a running auction's price only goes up, so a discount against it is a discount against a floor. Only a buy-now price at or below the threshold can be paid today — dealCount is the honest superset, actionableDealCount the part worth acting on.

Source coverage and page problems

The current live page embeds structured listing data. The older HTML-card format is also supported, including its card boundaries, promoted duplicates and exact tab total. malformedCards and duplicateCardsDropped ship with every summary. An unreadable, loading, repeated or inconsistent zero-result page fails rather than becoming "no deals". A readable explicit zero is a valid result; a missing comparison side is not.

Pricing

$0.02 per keyword scanned, plus $0.002 per individual deal returned (deals are on by default; turn them off with includeIndividualDeals). So one keyword that finds 28 deals costs $0.02 + 28 x $0.002, about 7.6 cents. You pay per keyword; there is no monthly fee. A keyword whose basis is unusable is never charged.

The run's maximum charge limits paid delivery. No new keyword search starts when another usable-basis summary cannot be paid for. Individual deals are returned only within the remaining deal budget. If this truncates delivery, a free run_notice row reports noticeFoundDeals and noticeDeliveredDeals. The summary's dealCount remains the number found among the checked live listings, including when individual output is disabled; it is not a promise that every candidate row was delivered. A delivery/charge receipt mismatch fails the run rather than being reported as success.

Read-only public pages, no login, throttled requests. The seller's name and id are never returned. Nothing is stored between runs.

If something goes wrong

• Wrong number or a failed run? Open a ticket on the Issues tab. I read every one and reply within 2 business days (Japan time).
• You never get a fake "empty" result. If the site can't be read, the run fails and says so.
• A keyword that can't be measured = no charge. Too few or too mixed sold prices and the keyword costs nothing. A keyword that could be measured costs $0.02 even when nothing is underpriced right now — that is the answer you asked for.
• Checked every week. An automatic test runs this tool weekly; if the site changes, I fix it.
• Public pages only. No login, no personal data, and it goes easy on the site.

More tools by the same author

• Yahoo! Auctions Japan Sold Price Summary & Monthly History
• Yahoo! Auctions Japan Bulk Listings — Live or Ended
• Yahoo! Auctions Japan Monthly History — Legacy Task/API
• Mercari Japan Price Summary — Live or Sold Out
• Daikokuya Japan Buyback — What a Shop Actually Paid
• Hifido Japan Used Audio — What It Actually Sold For
• Japan Used Camera & Lens — Dealer Price vs Sold Price
• Used iPhone & Android Prices Japan — Dealer vs Sold

All tools (Japan marketplaces, real estate, jobs, racing, prediction markets): https://apify.com/jpmarketdata

Disclaimer

Unofficial, independent tool — not affiliated with, endorsed by, or sponsored by Yahoo! Auctions. Product names and logos belong to their owners and only say where the data comes from. Data is read from public pages, for market research; check before you act on it.