2021-22 NBA Hoops Premium Box Set

• G-001 EXTRACTION_SPEC 5.2 BASE-CARD IDENTITY COLLISION, naming both releases: hoopspremium-bkb-2021-22 (this release) and paninihoops-bkb-2021-22 describe THE SAME 270 physical cards. This release is the /199 Panini-Direct Premium Box Set printing of that checklist; the parent is the pack-issued unnumbered printing (paninihoops-bkb-2021-22 carries base_print_run=unnumbered). A source states the repackaging in words rather than leaving it to be inferred: cardboardconnection, "2021-22 NBA Hoops Premium Box Set Basketball takes the entire base set from the main Hoops and packages it as a high-end release." (https://www.cardboardconnection.com/2021-22-nba-hoops-premium-box-set-basketball-cards); Panini America, "Hoops Premium Box Set includes a complete 270 card set which includes Base, Base Rookie, and Base Hoops Tribute all numbered to 199!" (https://blog.paniniamerica.net/available-thursday-2021-22-panini-hoops-nba-premium-box-set/). MEASURED, NOT ESTIMATED, against data/paninihoops-bkb-2021-22/base_cards.csv with csv.DictReader: POSITIONAL overlap (row index against row index, comparing card_number+player_name+team_or_school) = 270 of 270. BY-CARD-NUMBER overlap (keyed on card_number, comparing player_name+team_or_school) = 270 of 270. Zero card numbers are shared-but-differing; zero card numbers appear on only one side; the multiset of 270 (player_name, team_or_school) pairs is IDENTICAL on both sides. BOTH FIGURES ARE REPORTED DELIBERATELY, because they can diverge badly and the positional one can understate identity to the point of being misleading: the 2020-21 pair of this same line reads 175/270 positionally and 269/270 (270/270 after its 3b->98 correction) by card number, purely because one row was numbered differently and shifted every row after it. In 2021-22 the two measures agree because the run is a clean contiguous 1-270 on both sides. Built per 5.2 rather than deduplicated or double-counted. Note the same physical object is ALSO recorded twice inside paninihoops-bkb-2021-22: as a base-set parallel row "Premium Box Set" print_run 199 (its gap G-008) and as a sets.csv row "Hoops Tribute Premium Box Set Autographs" (its gap G-018). This release IS the separate release record; a resolver pass should decide which side keeps them.
• G-002 parallels.csv is written with its header and ZERO data rows, which is a finding and not an unfinished file (EXTRACTION_SPEC 2.3). VERIFIED FOR THIS SEASON RATHER THAN ASSUMED FROM THE PRIOR TWO. EVIDENCE FOR THE ABSENCE: https://www.tcdb.com/Inserts.cfm/sid/290813/2021-22-Hoops enumerates the parent set's complete inventory under the verbatim headings "Insert Sets (101)" and "Parallel Sets (23)". A whole-page search for the string "Premium" returns exactly two entries in the entire inventory - "Premium Box Set" (sid 296164) among the 23 parallels, and "Hoops Tribute Premium Box Set Autographs" (sid 296224) among the 101 inserts. There is no "Premium Box Set Gold", no "Premium Box Set Holo", no sub-parallel of either set of any kind. The page was read twice under two distinct URLs (plain, and with a harmless &sort=0 style parameter appended) so the second reading is not a 15-minute cache hit on the first. cardboardconnection's product page describes only the /199 base cards and the one /10 Tribute autograph and names no parallel (https://www.cardboardconnection.com/2021-22-nba-hoops-premium-box-set-basketball-cards); Panini America's own launch post likewise describes only the complete 270-card set /199 plus one autograph /10 (https://blog.paniniamerica.net/available-thursday-2021-22-panini-hoops-nba-premium-box-set/). This matches hoopspremium-bkb-2019-20 and hoopspremium-bkb-2020-21, both also parallel-free. It remains true that no source says in words "this box set has no parallels"; the absence is inferred from a complete enumeration plus two product descriptions, not from a statement. One apparent counter-signal was investigated and rejected - see G-014.
• G-003 EXTRACTION_SPEC 5.4 UNIT CHECK AT RELEASE LEVEL - THE TRAP THIS LINE HAS NOW SPRUNG IN ALL THREE SEASONS, AND IT RECURS HERE UNCHANGED. cardboardconnection prints "Product Configuration:** 271-card factory set" on the same page as "Full base set (270 cards)", "1 Hoops Tribute Autograph" and "Every box contains the full 270-card set numbered to 199 copies." (https://www.cardboardconnection.com/2021-22-nba-hoops-premium-box-set-basketball-cards). 271 is CARDS PER SEALED BOX: the 270-card base set plus the one Tribute autograph. 271 is therefore written into NO field; base_set_size is 270, sourced from tcdb Total Cards "270" (https://www.tcdb.com/ViewSet.cfm/sid/296164/2021-22-Hoops-Premium-Box-Set), from Panini America's "a complete 270 card set", and from cardboardconnection's "Full base set (270 cards)", and corroborated by counting 270 rendered rows and tiles one by one. Both load-bearing strings were re-fetched under a narrow verbatim prompt that demanded them inside quotation marks (EXTRACTION_SPEC 4.1) before either was believed. For the record of how this trap has behaved across the line: 2019-20 = 305 and 303, and those two belonged to a DIFFERENT product (Premium Stock); 2020-21 = 271, belonging to this product; 2021-22 = 271 again, same product, same page, same words. A fourth season should expect it again.
• G-004 TASK TRAP 3 TESTED: "NBA Hoops Premium Stock" IS A DIFFERENT, ONGOING RETAIL LINE AND NOTHING FROM IT IS MERGED HERE. Nothing in this release derives from a Premium Stock source. A targeted search for a 2021-22 Hoops Premium Stock product returned NO such product: the Premium Stock results returned were breakninja's "19-20 Panini NBA Hoops Premium Stock Basketball Checklist" and an eBay listing for a "2023-24 Panini NBA Hoops Premium Stock Basketball Bundle Box", with nothing for 2021-22. tcdb's 2021-22 Hoops parent inventory (https://www.tcdb.com/Inserts.cfm/sid/290813/2021-22-Hoops) contains no Premium Stock entry at all. That agrees with hoopspremium-bkb-2020-21 G-004, which found no 2020-21 Premium Stock product either, and with hoopspremium-bkb-2019-20 G-002/G-003, which established that the 2019-20 Premium Stock Box Set is a separate buildable release needing its own brand decision. reference/brands.csv still carries no row for a Premium Stock line; that escalation is already open as G-003 on hoopspremium-bkb-2019-20 and is NOT re-decided here. NEGATIVE RESULT CAVEAT: absence in a search is weaker than absence in an enumeration, so this records "no 2021-22 Premium Stock product was found", not "none exists".
• G-005 license_status WAS NOT SUPPLIED BY THE TASK BRIEFING and the catalogue is inconsistent on it, so the value written here is a convention choice and is recorded as one - EXACTLY the open question the briefing asked to be noted rather than resolved. EXTRACTION_SPEC 2.1 says license_status comes "as given in your task prompt"; this prompt gave product_name, season, manufacturer and league only. This release writes "licensed (NBA/NBPA)", matching its own line's hoopspremium-bkb-2019-20 and hoopspremium-bkb-2020-21 byte for byte. The parent product for the SAME season, data/paninihoops-bkb-2021-22/release.csv, writes "licensed (NBA)" for what is physically the same checklist under the same licence (measured: 270/270 identical, see G-001). WHAT THE SOURCES ACTUALLY SAY: nothing. No source consulted in this pass - tcdb sid 296164, tcdb sid 296224, cardboardconnection, blog.paniniamerica.net - prints the word "NBPA", the phrase "NBA Players Association", or any licensing statement at all about either product; the distinction between the two values is not sourced on either side and is a catalogue convention, not a fact either release found. I did NOT reconcile it unilaterally. The columns that actually gate the loader (licensed=yes, league=NBA) are unambiguous and identical on both sides. A repair pass should pick one value for the pair hoopspremium/paninihoops per season, or source the difference.
• G-006 CHANNEL IS KNOWN AND SOURCED BUT HAS NOWHERE TO LIVE IN THE SCHEMA, so it is recorded here. Panini America states "The 2021-22 Panini Hoops Premium Box Set will be featured in a Dutch Auction Format!", "Opening Price Starts at $4000. Price Drops every five minutes until product sells out.", "AVAILABLE VIA DUTCH AUCTION: February 10th @ 11AM (CST)", "Limited to 1 Set Per Transaction" and "All Sales are Final." (https://blog.paniniamerica.net/available-thursday-2021-22-panini-hoops-nba-premium-box-set/). tcdb's autograph-set description says "Panini Official Website exclusive sale Feb 10th, 2022" (https://www.tcdb.com/ViewSet.cfm/sid/296224/2021-22-Hoops-Hoops-Tribute-Premium-Box-Set-Autographs). cardboardconnection says "The Premium Collector's Series Set is released in an online-exclusive "Dutch Auction" format on Thursday, February 10 at 12 PM (Eastern)." and "The price starts at $4,000 and decreases every five minutes until it sells out or reaches $750" (https://www.cardboardconnection.com/2021-22-nba-hoops-premium-box-set-basketball-cards). In the taxonomy of EXTRACTION_SPEC 2.3 that is channel=panini-direct with channel_basis=sourced-exclusive and, by the derived map, channel_class=promotion. release.csv has no channel column and parallels.csv has no rows, so NO machine-readable channel value exists anywhere in this release and a consumer filtering by channel will not find it. Identical disposition to hoopspremium-bkb-2020-21 G-007. Note for a schema pass: this is the third consecutive season of this line where a fully sourced channel is unrepresentable.
• G-007 card_tier IS WRITTEN ON 251-270 ONLY, AND DIVERGES FROM THE PARENT RELEASE; the divergence is recorded rather than smoothed. This release writes card_tier="Hoops Tribute" on cards 251-270 and leaves it empty on 1-250. Basis: this product's OWN tcdb checklist renders the literal flag token "TRIB" on exactly those 20 rows and on no others (https://www.tcdb.com/Checklist.cfm/sid/296164/2021-22-Hoops-Premium-Box-Set?PageIndex=3 renders "251 | Luka Doncic | TRIB, SN199 | Dallas Mavericks" through "270 | Dirk Nowitzki | TRIB, SN199 | Dallas Mavericks"), and the parent set renders it too, re-fetched verbatim as "251 | Luka Doncic TRIB | Dallas Mavericks" and "270 | Dirk Nowitzki TRIB | Dallas Mavericks" (https://www.tcdb.com/Checklist.cfm/sid/290813/2021-22-Hoops?PageIndex=3). This is BETTER evidence than the 2020-21 sibling had, which had to read TRIB off the parent only. NOTE the expansion of the token "TRIB" to the words "Hoops Tribute" comes from Panini America ("Base Hoops Tribute") and cardboardconnection ("1 Hoops Tribute Autograph"), NOT from tcdb - EXTRACTION_SPEC 3.3 warns that the fetch layer invents glosses for tcdb flag tokens, so the token is recorded here as the literal string it is. TWO DIVERGENCES: (a) data/paninihoops-bkb-2021-22/base_cards.csv leaves card_tier EMPTY on all 270 rows including 251-270, although its release notes describe the same three segments in prose; (b) the 201-250 rookie block gets no card_tier here, following hoopspremium-bkb-2019-20 G-009 and hoopspremium-bkb-2020-21 G-010 which both left their rookie block empty for the same reason - even though this season Panini America DOES name the segment in words ("Base, Base Rookie, and Base Hoops Tribute"), so a uniform repair pass across the line could now fill it with a citation. I did not change the line's convention unilaterally mid-line. The base run is NOT treated as tiered for print-run purposes: it is uniformly /199, so base_print_run is the integer 199 and not varies-by-tier, and base_card_tiers.csv is correctly absent (EXTRACTION_SPEC 2.5).
• G-008 rookie_flag RESTS ON THE PARENT SET, NOT ON THIS PRODUCT'S OWN CHECKLIST - but with a positive control and near-complete row coverage, which is a materially stronger position than the prior two seasons of this line. tcdb renders NO "RC" token anywhere on the Premium Box Set's own rows or tiles (it renders "SN199", and "TRIB" on 251-270), the same behaviour recorded in hoopspremium-bkb-2019-20 G-008 and hoopspremium-bkb-2020-21 G-009: a parallel gallery is a good print-run source and a bad rookie-flag source. rookie_flag was therefore read off the PARENT set sid 290813, for the identical card numbers. COVERAGE ACTUALLY OBSERVED ON THE PARENT: cards 1-76 and 101-176 via Checklist.cfm PageIndex 1 and 2 (both answered that the literal two-character string "RC" appears on no row of the page); cards 71-100 and 171-200 via Gallery.cfm PageIndex 8, 9, 10, 18, 19 and 20 (10 tiles each, all six answered "RC" absent); cards 201-250 via Checklist.cfm PageIndex 3 which lists all 50 as carrying "RC"; cards 251-270 via the same page which lists all 20 as carrying "TRIB". So every one of the 270 numbers was covered by at least one endpoint - no range rests on block structure alone, unlike hoopspremium-bkb-2020-21 G-009 which had no row-level reading for 78-100 and 187-200. POSITIVE CONTROL RUN, because "no RC found" is worthless if the renderer never prints RC: https://www.tcdb.com/Gallery.cfm/sid/290813/2021-22-Hoops?PageIndex=21 was fetched and DOES render it, verbatim "201 - Cade Cunningham RC - Detroit Pistons" and "207 - Santi Aldama RC - Memphis Grizzlies", on all ten tiles 201-210. A narrow verbatim re-fetch of individual parent rows returned "201 | Cade Cunningham RC | Detroit Pistons", "250 | Luka Garza RC | Detroit Pistons", "251 | Luka Doncic TRIB | Dallas Mavericks" and answered "No" to whether "RC" appears on card 251. RESIDUAL WEAKNESS, stated rather than hidden: for cards 211-249 the RC designation rests on a page-level summary listing all 50 rookie names rather than on 39 individually quoted rows, corroborated by the narrow verbatim reading of the block's two endpoints (201 and 250) and by Panini America's "Base, Base Rookie, and Base Hoops Tribute". data/paninihoops-bkb-2021-22/base_cards.csv independently records rookie_flag=yes for exactly 201-250 and no elsewhere, so the two releases AGREE - unlike the 2019-20 pair, where hoopspremium-bkb-2019-20 G-007 records a direct rookie_flag contradiction.
• G-010 THE TWO STRUCTURAL SOURCES MAY BE CLOSER TO ONE OPINION THAN TWO - BUT THIS SEASON THERE IS A GENUINE THIRD, AND IT IS THE MANUFACTURER. tcdb and cardboardconnection agree on every value written here (270 base cards, /199, a 20-card Hoops Tribute autograph set at 251-270 numbered /10, February 10) and cardboardconnection's set inventories for the Hoops line track tcdb closely enough that independence cannot be assumed - the same caution hoopspremium-bkb-2019-20 G-010 and hoopspremium-bkb-2020-21 G-011 raised. WHAT IS DIFFERENT IN 2021-22: Panini America's own launch post (https://blog.paniniamerica.net/available-thursday-2021-22-panini-hoops-nba-premium-box-set/) is a first-party source and is not derived from either, and it independently states the 270-card set, the /199 numbering, the one autograph "numbered to as low as 10", the Dutch auction, the $4000 opening price and the February 10 date. So base_set_size, base_print_run and release_date each rest on three sources of which one is the manufacturer. beckett was NOT reachable in this pass and no beckett page was opened. Marketplace corroboration, seen as search-result titles rather than fetched pages and therefore load-bearing for nothing: "John Wall 2021-22 Panini Hoops Premium Box Set /199 #144 NBA Rockets" (https://www.ebay.com/itm/325048360459) matches this checklist at 144 John Wall Houston Rockets; "2021-22 Panini Hoops Premium Box Set /199 CADE CUNNINGHAM PISTONS 201" (https://www.ebay.com/itm/125275482134) matches at 201 Cade Cunningham Detroit Pistons and at rookie_flag=yes; an eBay catalogue product page titled "2021-22 Panini NBA Hoops Premium Box Set - 270 Cards (#/199) 1 Auto (#/10)" (https://www.ebay.com/p/6052434604) restates the whole shape. The cross-check that carried real weight is data/paninihoops-bkb-2021-22, built from a different source mix, which agrees with this release on all 270 card numbers, players, teams and the rookie block.
• G-011 CONTRADICTION FOUND IN THIS PROJECT'S OWN ALREADY-GATED DATA, reported rather than repaired (it is another release's directory). data/hoopspremium-bkb-2020-21/release.csv carries a dated marker reading "[CORRECTED 2026-09-12: ... Consequence for the overlap figures stated above: the base run is now contiguous 1-270 and positional overlap against paninihoops-bkb-2020-21 is 270/270, not 175/270.]". THAT CLAIM IS FALSE AS THE FILES ACTUALLY STAND, measured with csv.DictReader today: hoopspremium-bkb-2020-21/base_cards.csv still carries the Jeremy Lamb row PHYSICALLY AT ROW INDEX 4 (immediately after card 3, where tcdb rendered "3b"), with its card_number now written as 98. Renumbering the cell did not move the row. So positional (row-index) overlap against data/paninihoops-bkb-2020-21/base_cards.csv is STILL 175 of 270, exactly as before the correction; what the correction actually changed is the BY-CARD-NUMBER figure, which rose from 269/270 to 270/270 - and the marker does not state that new figure either, still describing 269/270 in the prose above it. Net: of the two figures the marker asserts, the positional one is wrong and the by-card-number one is stale. FIX, for whoever owns that release: either reorder that one row so the file is sorted by card_number (which EXTRACTION_SPEC 1.3 / CLAUDE.md forbid doing casually - "never reorder rows"), or amend the marker to say "by-card-number overlap is 270/270; positional overlap remains 175/270 because the row was not moved". The second is almost certainly correct. METHOD NOTE: this is exactly the failure mode the task briefing warned about in the other direction - a positional figure and a by-card-number figure are different measurements and a note that conflates them misleads. It also illustrates EXTRACTION_SPEC 5.3's own lesson: a claim about our own data is the cheapest kind to verify, three lines of csv.DictReader, and this one was asserted without being run.
• G-012 No pack odds and no scarcity tier are recorded for either set. This is a sealed complete factory box set with fixed contents sold as a single unit by Dutch auction, not a pack product - Panini America states "Limited to 1 Set Per Transaction" and cardboardconnection "Full set per box" and "1 per box" - so per-set pack odds do not exist in the ordinary sense; no source consulted states any odds figure. stated_odds, odds_evidence_url, odds_evidence_quote and scarcity_tier are therefore empty on both sets.csv rows. Same disposition as hoopspremium-bkb-2019-20 G-012 and hoopspremium-bkb-2020-21 G-013.
• G-013 ENDPOINT AND FETCH-COST MEASUREMENTS FROM THIS PASS, confirming EXTRACTION_SPEC 3.2 as corrected and adding one number to it. (1) The corrected 3.2 recipe is right and the old 10:1 claim would have lost the tail of every page. Measured on sid 296164: Checklist.cfm?PageIndex=1 delivered complete rows 1-78 of 100 and then cut off mid-row at 79; ?PageIndex=2 delivered 101-176 of 101-200 and cut off mid-row at 177; ?PageIndex=3 is a 70-row page, under the truncation point, and came back complete at 201-270. So the truncation sat at rows 78 and 76 of 100 this pass, inside the 74-86 band six agents measured. (2) The Gallery top-up cost exactly what 3.2 predicts: six Gallery.cfm pages (8, 9, 10, 18, 19, 20) recovered cards 79-100 and 177-200, all six returning a full 10 tiles. TOTAL COST FOR A 270-CARD RUN: 3 Checklist + 6 Gallery = 9 fetches, identical to the 9 hoopspremium-bkb-2020-21 measured for its 270-card run, against 27 for Gallery alone. That is now two independent measurements of the same figure on the same line and it can be budgeted. (3) NO Gallery tile was silently dropped this pass - all six top-up pages and all four parent pages returned 10 of 10 - so the dropped-tile defect hoopspremium-bkb-2020-21 G-012 recorded is intermittent, not systematic. The cache-busting discipline was still applied: every count re-check was issued under a URL varied by a harmless extra parameter, never as a second call to an identical URL. (4) NO 503 was returned on any of the ~20 tcdb fetches in this pass, at a concurrency of 3-4.
• G-014 AN APPARENT PARALLEL WAS FOUND AND REJECTED AS SOURCE CONTAMINATION; recorded so a later pass does not rediscover it and write it. A free-text eBay seller title reads "2021-22 Panini Hoops Premium Box Set Silver Kevin Huerter /199 #148 Hawks SP" (https://www.ebay.com/itm/188488755253), which on its face suggests a "Silver" parallel OF the Premium Box Set. It is not one, and the explanation is fully determined: "Silver" is a SEPARATE base parallel of the main 2021-22 Hoops set, tcdb sid 296170, listed among the 23 parallels at https://www.tcdb.com/Inserts.cfm/sid/290813/2021-22-Hoops alongside and independent of "Premium Box Set" (sid 296164); data/paninihoops-bkb-2021-22/parallels.csv carries it as its own base-set-scope row with print_run 199 and channel hobby. Both the Silver parallel and the Premium Box Set are numbered /199, so a seller holding one /199 Hoops card has two true-sounding labels to reach for and reached for both. Card 148 is Kevin Huerter, Atlanta Hawks in this checklist, so the number and player check out and only the tier attribution is wrong. This is EXTRACTION_SPEC 3.4 contamination shape four, a real label bolted onto the wrong product. NOTHING was written from it and parallels.csv stays at zero rows (G-002).
• G-015 PRODUCT-NAME VARIANTS SEEN IN PROSE, NONE WRITTEN, NO BRAND ESCALATION NEEDED (EXTRACTION_SPEC 2.1/5.1 - report, do not decide). product_name is written as "NBA Hoops Premium Box Set Basketball", which is what reference/brands.csv carries for brand hoopspremium (brd_01M2B8M6BJKF4BVCZ1A77FRZ27) and what hoopspremium-bkb-2019-20 and hoopspremium-bkb-2020-21 both carry, byte for byte. The sources agree on the identity: tcdb "2021-22 Hoops - Premium Box Set" (https://www.tcdb.com/ViewSet.cfm/sid/296164/2021-22-Hoops-Premium-Box-Set), cardboardconnection page heading "2021-22 NBA Hoops Premium Box Set Basketball Cards Checklist" (https://www.cardboardconnection.com/2021-22-nba-hoops-premium-box-set-basketball-cards), Panini America "2021-22 Panini Hoops NBA Premium Box Set" and "Hoops Premium Box Set" (https://blog.paniniamerica.net/available-thursday-2021-22-panini-hoops-nba-premium-box-set/). TWO VARIANTS EXIST IN PROSE AND ARE DELIBERATELY NOT WRITTEN ANYWHERE: cardboardconnection calls the same object "The Premium Collector's Series Set" in one sentence of the same page whose heading says Premium Box Set, and DA Card World lists the product as "2021/22 Panini NBA Hoops Basketball Premium Collectors Set (Box) /199" (https://www.dacardworld.com/sports-cards/2021-22-panini-nba-hoops-basketball-premium-collectors-set-box) - the same retailer string data/paninihoops-bkb-2021-22 already cites on its Premium Box Set parallel row. Recorded because EXTRACTION_SPEC 2.1 makes the brand name permanent and a later agent meeting "Premium Collectors Set" first should not mint a second brand row from it.

If you know a source that answers one of these, tell us — name the question's label and where to look.