Valuation accuracy
Is Courtyard's valuation any good?
Every card-rip app values your card with its own algorithm, and that algorithm decides what a buy-back is worth. Courtyard is the only one we track that publishes its number on a page you can read. So we took 9 of its valuations and checked each against independent sold comps — matching grade and parallel, which turns out to be the whole game.
What Courtyard's automated 84.6% standing offer actually pays, against real market
84.8% of market value
On 4 cards where we could resolve a clean comp, the standing offer landed at 81.3%–86.4% of true market. Courtyard never publishes a buy-back rate; this is the number, and it holds tighter against the market than it does against Courtyard's own stated values. Sell back and you give up roughly 15% for instant liquidity — a real cost, and a legible one.
This page reverses two of our own earlier findings
The first two versions measured Courtyard's Market Value against the recent sales Courtyard itself displays, because that was all we had. On that basis we published a population threshold, then that the Confidence flag carried no information. Both are dead. They were artifacts of treating those displayed comps as ground truth — and the displayed comps turn out to be the least reliable thing on the screen. Against independent sold data the valuation is good and the flag works. We would rather show the reversal than bury it.
The valuation, against independent sold comps
Sorted by accuracy. On the 4 cards with a sold-derived comp at the matching grade and parallel, the median error is 3.1 points and the mean is 4.7.
Swipe the table
| Card | Their flag | Courtyard | Market | Ratio | Comp |
|---|---|---|---|---|---|
| 2001 Pokemon VS #013 Bugsy's Scizor - 1st Edition Holo, PSA 10 2 sales per month; PSA 10 population 330, total 1,197 | High | $4,026.90 | $3,950 | 101.9% | sold |
| 2024 Topps Chrome Sapphire #CSAYY Yoshinobu Yamamoto Autograph-GOLD /50, PSA 10 the /50 parallel is scarce by construction; the base autograph trades weekly | High | $5,922 | $5,800 | 102.1% | ask |
| 1954 Topps #128 Hank Aaron, PSA 3.5 4 sales, pop 235, average $5,421.55, range $4,921.20-$6,000.00 | High | $5,522.20 | $5,650 | 97.7% | sold |
| 2019 Pokemon Sun & Moon Unbroken Bonds #205 Gardevoir & Sylveon GX - Holo Full Art, PSA 10 1 sale per day; 30 PSA 10 sold listings on file | High | $5,057.60 | $5,259.81 | 96.2% | sold |
| 2023 Panini Spectra Color Blast #1 Keyonte George, PSA 10 THIN: 1 sale per year, exactly 1 PSA 10 sold listing on file. Grades 7, 8, 9 and 9.5 all show no data at all. | Low | $2,123 | $2,375 | 89.4% | sold |
| 2006 Pokemon EX Dragon Frontiers #100 Charizard - Holo Gold Star, PSA 1 THIN AT THIS GRADE. PriceCharting publishes grade tiers from 7 upward and has no PSA 1 figure at all. | High | $4,107.40 | $4,699.99 | 87.4% | ask |
| 1996 Upper Deck Jordan Tribute #MJ35 Michael Jordan - MJ Impressions, PSA 8 PSA 8 population 31, total PSA population 132 — a genuinely small market | Medium | $37.70 | $63 | 59.8% | guide |
A sold comp is built from completed sales and is the only kind we draw conclusions from. An ask is an active listing — an upper bound nobody has paid. A guide is a model output, not a sale. The worst row in the table rests on a guide, on a card worth $37.70.
The Confidence flag does work — we had that backwards
Measured against real market rather than against Courtyard's own displayed comps, the flag is directionally right.
| Flag | Cards | Mean error | Ratios |
|---|---|---|---|
| High | 5 | 4.6 pts | 101.9% · 102.1% · 97.7% · 96.2% · 87.4% |
| Medium | 1 | 40.2 pts | 59.8% |
| Low | 1 | 10.6 pts | 89.4% |
The 5 cards flagged High average 4.6 points of error against the market. The 2 flagged anything else average 25.4. With 2 cards in the second group that is a direction, not a measurement — but it is the opposite direction to what we published, so it goes up.
What is still broken is the display, not the number
The clearest case in the whole exercise
1954 Topps #128 Hank Aaron, PSA 3.5. Its Courtyard page shows a price history with no sales at its own grade, PSA 10 rows in their place, and a $5 sale sitting beside a $295,850 one for the same card. Confidence: High.
And the valuation is 97.7% of the real last sold price — four independent sales averaging $5,421.55 against Courtyard's $5,522.20. The number is fine. The evidence shown beside it is garbage.
Why that still matters
A buyer cannot audit a valuation they are given bad evidence for. Showing $5 and $295,850 as comparable sales, under a High confidence flag, invites exactly the conclusion we drew twice and had to withdraw — that the number behind it must be broken too. It is not. Fixing the display costs Courtyard nothing and would make its best feature checkable.
Two we could not resolve, and the trap they show
Match the parallel, not just the card and grade
Card 5 is the one to learn from. SportsCardsPro lists the 2024 Topps Chrome Sapphire Yoshinobu Yamamoto Autograph at $2,601.19 for PSA 10 — but that is the BASE autograph. The Courtyard slab is the AUTOGRAPH-GOLD /50 parallel, and the matching eBay listing is $5,800. Comparing Courtyard's $5,922 to the base card would have produced a fake 227% error. The parallel and serial must match, not just the player, set and grade.
2018 Panini Prizm #78 Trae Young - PURPLE PULSAR 24/35, PSA 9
Courtyard says $1,929.70. The comps found were 2018 Prizm Trae Young #78 SILVER at PSA 10 — Card Ladder last sold $125.00 on a population of 2,395, and Fanatics/eBay asks of $249-$700. That is a different parallel AND a different grade. and the most instructive miss in the set. Courtyard priced this at $1,929.70 off a single displayed comp of $1,929.69. A /35 Purple Pulsar at PSA 9 and a pop-2,395 Silver at PSA 10 are different products; comparing them would have manufactured a 1,540% error. Needs a Purple Pulsar /35 PSA 9 comp specifically.
2018 Topps 1983 Topps Baseball Rookies #83-1 Shohei Ohtani - GOLD, PSA 9
Courtyard says $5,299.40. The eBay product page for #83-1 exposes a Parallel/Variety selector — Base, Black, Blue, Gold, Red — and the listing captured ($430.00, PSA 9) is the BASE. Courtyard holds the GOLD. Needs the GOLD parallel filtered at PSA 9. Courtyard has this at $5,299.40 against its own comps median of $3,650, so it is the card we most want a clean number for.
Method, and what this does not measure
9 valuations captured from Courtyard card pages, each checked against an independently sourced comp at the matching grader, grade and parallel. Sold-derived comps come from PriceCharting, SportsCardsPro and Card Ladder; where none existed at the card's own grade we say so rather than substituting a nearby grade. This measures whether the app's stated value is right. Whether the app pays that value is the separate question of who pays closest to market, which needs a second app measured before a ranking means anything.