cardtrack

Best Pokémon cards to grade

Ranked by expected profit per submissionevery grade outcome weighted by how often the card actually receives it in the PSA population — after current grading fees. The PSA 10 best case is on every row, but it no longer decides the order: a card that gems one time in twenty is not a better bet than one that gems one time in three. Top pick: Gardevoir & Sylveon-GX at an expected +$1,347.

Cards worth grading4,797
With a gem-rate sample4,50794% of the population
Positive expected value2,819
Median gem rate18.3%
PSA fees as of2026-08-26
Reset

Ranked by expected profit54 matching

Cards worth grading, ranked by expected grading profit
#CardRawPSA 9PSA 10CostBest caseGem rateExpectedSample
1Gardevoir & Sylveon-GXGardevoir & Sylveon-GXUnbroken Bonds #205 · Modern (2011–2019)
Expected +$1,347 per submission: 44.8% of 3,981 graded copies came back PSA 10 ($4,550), and a PSA 9 ($1,250) still clears the $1,447 you have in it. Confidence high on that sample.
$1,032$1,250$4,550$415+$3,10344.8%+$1,3473,981
high
2Greninja & Zoroark-GXGreninja & Zoroark-GXUnbroken Bonds #201 · Modern (2011–2019)
Expected +$412 per submission: 45.5% of 3,084 graded copies came back PSA 10 ($1,638), and only a PSA 10 covers the $543 you have in it — a 9 loses money. Confidence high on that sample.
$358$329$1,638$185+$1,09545.5%+$4123,084
high
3SnorlaxSnorlaxUnbroken Bonds #158 · Modern (2011–2019)
Expected +$369 per submission: 16% of 807 graded copies came back PSA 10 ($2,400), and a PSA 9 ($285) still clears the $277 you have in it. Confidence high on that sample.
$84$285$2,400$193+$2,12316%+$369807
high
4Marshadow & Machamp-GXMarshadow & Machamp-GXUnbroken Bonds #199 · Modern (2011–2019)
Expected +$349 per submission: 48% of 2,926 graded copies came back PSA 10 ($1,162), and only a PSA 10 covers the $368 you have in it — a 9 loses money. Confidence high on that sample.
$256$323$1,162$112+$79548%+$3492,926
high
5GreninjaGreninjaUnbroken Bonds #117 · Modern (2011–2019)
Expected +$316 per submission: 16.2% of 253 graded copies came back PSA 10 ($2,150), and a PSA 9 ($180) still clears the $221 you have in it. Confidence high on that sample.
$30$180$2,150$191+$1,92916.2%+$316253
high
6Blastoise-GXBlastoise-GXUnbroken Bonds #218 · Modern (2011–2019)
Expected +$219 per submission: 43% of 1,691 graded copies came back PSA 10 ($882), and only a PSA 10 covers the $246 you have in it — a 9 loses money. Confidence high on that sample.
$137$128$882$109+$63743%+$2191,691
high
7MewMewUnbroken Bonds #76 · Modern (2011–2019)
Expected +$141 per submission: 25.5% of 341 graded copies came back PSA 10 ($570), and a PSA 9 ($160) still clears the $123 you have in it. Confidence high on that sample.
$17$160$570$106+$44825.5%+$141341
high
8Marshadow & Machamp-GXMarshadow & Machamp-GXUnbroken Bonds #221 · Modern (2011–2019)
Expected +$128 per submission: 37.8% of 747 graded copies came back PSA 10 ($626), and only a PSA 10 covers the $177 you have in it — a 9 loses money. Confidence high on that sample.
$71$87$626$106+$44937.8%+$128747
high
9GengarGengarUnbroken Bonds #70 · Modern (2011–2019)
Expected +$126 per submission: 12.5% of 224 graded copies came back PSA 10 ($1,255), and a PSA 9 ($134) still clears the $139 you have in it. Confidence high on that sample.
$26$134$1,255$113+$1,11712.5%+$126224
high
10Gardevoir & Sylveon-GXGardevoir & Sylveon-GXUnbroken Bonds #225 · Modern (2011–2019)
Expected +$118 per submission: 40.2% of 1,293 graded copies came back PSA 10 ($700), and only a PSA 10 covers the $271 you have in it — a 9 loses money. Confidence high on that sample.
$164$170$700$107+$42940.2%+$1181,293
high
11Reshiram & Charizard-GXReshiram & Charizard-GXUnbroken Bonds #194 · Modern (2011–2019)
Expected +$103 per submission: 42.8% of 3,587 graded copies came back PSA 10 ($525), and only a PSA 10 covers the $201 you have in it — a 9 loses money. Confidence high on that sample.
$96$126$525$105+$32442.8%+$1033,587
high
12Gardevoir & Sylveon-GXGardevoir & Sylveon-GXUnbroken Bonds #130 · Modern (2011–2019)
Expected +$102 per submission: 48.9% of 4,025 graded copies came back PSA 10 ($409), and only a PSA 10 covers the $144 you have in it — a 9 loses money. Confidence high on that sample.
$70$100$409$74+$26548.9%+$1024,025
high
13Gardevoir & Sylveon-GXGardevoir & Sylveon-GXUnbroken Bonds #204 · Modern (2011–2019)
Expected +$95 per submission: 53.2% of 1,649 graded copies came back PSA 10 ($450), and only a PSA 10 covers the $185 you have in it — a 9 loses money. Confidence high on that sample.
$111$87$450$74+$26553.2%+$951,649
high
14Lucario & Melmetal-GXLucario & Melmetal-GXUnbroken Bonds #224 · Modern (2011–2019)
Expected +$94 per submission: 55.5% of 847 graded copies came back PSA 10 ($385), and only a PSA 10 covers the $155 you have in it — a 9 loses money. Confidence high on that sample.
$81$88$385$74+$23055.5%+$94847
high
15Reshiram & Charizard-GXReshiram & Charizard-GXUnbroken Bonds #217 · Modern (2011–2019)
Expected +$87 per submission: 39.8% of 4,092 graded copies came back PSA 10 ($725), and only a PSA 10 covers the $322 you have in it — a 9 loses money. Confidence high on that sample.
$215$179$725$107+$40239.8%+$874,092
high
16Muk & Alolan Muk-GXMuk & Alolan Muk-GXUnbroken Bonds #197 · Modern (2011–2019)
Expected +$71 per submission: 52.8% of 1,844 graded copies came back PSA 10 ($400), and only a PSA 10 covers the $182 you have in it — a 9 loses money. Confidence high on that sample.
$108$95$400$74+$21852.8%+$711,844
high
17Marshadow & Machamp-GXMarshadow & Machamp-GXUnbroken Bonds #198 · Modern (2011–2019)
Expected +$63 per submission: 51.5% of 922 graded copies came back PSA 10 ($300), and only a PSA 10 covers the $115 you have in it — a 9 loses money. Confidence high on that sample.
$42$49$300$73+$18551.5%+$63922
high
18Greninja & Zoroark-GXGreninja & Zoroark-GXUnbroken Bonds #222 · Modern (2011–2019)
Expected +$60 per submission: 45.3% of 1,119 graded copies came back PSA 10 ($403), and only a PSA 10 covers the $177 you have in it — a 9 loses money. Confidence high on that sample.
$103$110$403$74+$22545.3%+$601,119
high
19Greninja & Zoroark-GXGreninja & Zoroark-GXUnbroken Bonds #107 · Modern (2011–2019)
Expected +$57 per submission: 48.4% of 3,389 graded copies came back PSA 10 ($300), and only a PSA 10 covers the $123 you have in it — a 9 loses money. Confidence high on that sample.
$50$75$300$73+$17748.4%+$573,389
high
20Greninja & Zoroark-GXGreninja & Zoroark-GXUnbroken Bonds #200 · Modern (2011–2019)
Expected +$55 per submission: 54.6% of 1,326 graded copies came back PSA 10 ($290), and only a PSA 10 covers the $140 you have in it — a 9 loses money. Confidence high on that sample.
$67$87$290$73+$15054.6%+$551,326
high
21Pheromosa & Buzzwole-GXPheromosa & Buzzwole-GXUnbroken Bonds #192 · Modern (2011–2019)
Expected +$50 per submission: 50.3% of 1,583 graded copies came back PSA 10 ($340), and only a PSA 10 covers the $163 you have in it — a 9 loses money. Confidence high on that sample.
$90$90$340$73+$17750.3%+$501,583
high
22Marshadow & Machamp-GXMarshadow & Machamp-GXUnbroken Bonds #82 · Modern (2011–2019)
Expected +$49 per submission: 48.7% of 2,254 graded copies came back PSA 10 ($275), and only a PSA 10 covers the $108 you have in it — a 9 loses money. Confidence high on that sample.
$35$50$275$73+$16848.7%+$492,254
high
23Pheromosa & Buzzwole-GXPheromosa & Buzzwole-GXUnbroken Bonds #215 · Modern (2011–2019)
Expected +$45 per submission: 42% of 512 graded copies came back PSA 10 ($312), and only a PSA 10 covers the $109 you have in it — a 9 loses money. Confidence high on that sample.
$36$40$312$73+$20342%+$45512
high
24Dedenne-GXDedenne-GXUnbroken Bonds #219 · Modern (2011–2019)
Expected +$37 per submission: 33.6% of 393 graded copies came back PSA 10 ($357), and only a PSA 10 covers the $122 you have in it — a 9 loses money. Confidence high on that sample.
$48$80$357$74+$23533.6%+$37393
high
25Celesteela-GXCelesteela-GXUnbroken Bonds #228 · Modern (2011–2019)
Expected +$36 per submission: 25.5% of 188 graded copies came back PSA 10 ($411), and only a PSA 10 covers the $86 you have in it — a 9 loses money. Confidence high on that sample.
$12$25$411$74+$32425.5%+$36188
high
26Lucario & Melmetal-GXLucario & Melmetal-GXUnbroken Bonds #120 · Modern (2011–2019)
Expected +$36 per submission: 50.9% of 2,625 graded copies came back PSA 10 ($240), and only a PSA 10 covers the $107 you have in it — a 9 loses money. Confidence high on that sample.
$35$45$240$72+$13350.9%+$362,625
high
27Lucario & Melmetal-GXLucario & Melmetal-GXUnbroken Bonds #203 · Modern (2011–2019)
Expected +$36 per submission: 53% of 1,015 graded copies came back PSA 10 ($235), and only a PSA 10 covers the $117 you have in it — a 9 loses money. Confidence high on that sample.
$45$70$235$72+$11753%+$361,015
high
28Pheromosa & Buzzwole-GXPheromosa & Buzzwole-GXUnbroken Bonds #1 · Modern (2011–2019)
Expected +$34 per submission: 50.5% of 1,061 graded copies came back PSA 10 ($200), and only a PSA 10 covers the $87 you have in it — a 9 loses money. Confidence high on that sample.
$15$43$200$72+$11350.5%+$341,061
high
29Pheromosa & Buzzwole-GXPheromosa & Buzzwole-GXUnbroken Bonds #191 · Modern (2011–2019)
Expected +$33 per submission: 53.8% of 420 graded copies came back PSA 10 ($215), and only a PSA 10 covers the $95 you have in it — a 9 loses money. Confidence high on that sample.
$23$26$215$72+$11953.8%+$33420
high
30Honchkrow-GXHonchkrow-GXUnbroken Bonds #202 · Modern (2011–2019)
Expected +$27 per submission: 57.4% of 169 graded copies came back PSA 10 ($170), and only a PSA 10 covers the $78 you have in it — a 9 loses money. Confidence high on that sample.
$6$15$170$72+$9357.4%+$27169
high
31Koga's TrapKoga's TrapUnbroken Bonds #211 · Modern (2011–2019)
Expected +$11 per submission: 50.7% of 428 graded copies came back PSA 10 ($213), and only a PSA 10 covers the $109 you have in it — a 9 loses money. Confidence high on that sample.
$37$27$213$72+$10350.7%+$11428
high
32Dedenne-GXDedenne-GXUnbroken Bonds #195 · Modern (2011–2019)
Expected +$10 per submission: 48.7% of 345 graded copies came back PSA 10 ($200), and only a PSA 10 covers the $101 you have in it — a 9 loses money. Confidence high on that sample.
$29$25$200$72+$9948.7%+$10345
high
33Red's ChallengeRed's ChallengeUnbroken Bonds #213 · Modern (2011–2019)
Expected +$9 per submission: 43.8% of 2,369 graded copies came back PSA 10 ($368), and only a PSA 10 covers the $236 you have in it — a 9 loses money. Confidence high on that sample.
$162$181$368$74+$13243.8%+$92,369
high
34Muk & Alolan Muk-GXMuk & Alolan Muk-GXUnbroken Bonds #220 · Modern (2011–2019)
Expected +$7 per submission: 47.6% of 544 graded copies came back PSA 10 ($201), and only a PSA 10 covers the $110 you have in it — a 9 loses money. Confidence high on that sample.
$38$50$201$72+$9147.6%+$7544
high
35Whimsicott-GXWhimsicott-GXUnbroken Bonds #206 · Modern (2011–2019)
Expected +$6 per submission: 56.5% of 177 graded copies came back PSA 10 ($150), and only a PSA 10 covers the $83 you have in it — a 9 loses money. Confidence high on that sample.
$12$10$150$71+$6656.5%+$6177
high
36Blastoise-GXBlastoise-GXUnbroken Bonds #35 · Modern (2011–2019)
Expected +$5 per submission: 51.4% of 1,606 graded copies came back PSA 10 ($154), and only a PSA 10 covers the $90 you have in it — a 9 loses money. Confidence high on that sample.
$18$38$154$72+$6451.4%+$51,606
high
37Dedenne-GXDedenne-GXUnbroken Bonds #195a · Modern (2011–2019)
Expected +$2 per submission: 27.5% of 1,713 graded copies came back PSA 10 ($276), and only a PSA 10 covers the $106 you have in it — a 9 loses money. Confidence high on that sample.
$33$50$276$73+$17027.5%+$21,713
high
38Honchkrow-GXHonchkrow-GXUnbroken Bonds #223 · Modern (2011–2019)
Expected +$1 per submission: 41.4% of 222 graded copies came back PSA 10 ($174), and only a PSA 10 covers the $86 you have in it — a 9 loses money. Confidence high on that sample.
$14$30$174$72+$8841.4%+$1222
high
39Persian-GXPersian-GXUnbroken Bonds #207 · Modern (2011–2019)
Expected +$-0 per submission: 54.7% of 307 graded copies came back PSA 10 ($130), and only a PSA 10 covers the $80 you have in it — a 9 loses money. Confidence high on that sample.
$9$22$130$71+$5054.7%+$0307
high
40Reshiram & Charizard-GXReshiram & Charizard-GXUnbroken Bonds #20 · Modern (2011–2019)
Expected loss −$2 per submission: 25% of 3,025 graded copies came back PSA 10 ($314), and only a PSA 10 covers the $134 you have in it — a 9 loses money. Confidence high on that sample.
$61$75$314$73+$18025%−$23,025
high
41Muk & Alolan Muk-GXMuk & Alolan Muk-GXUnbroken Bonds #61 · Modern (2011–2019)
Expected loss −$3 per submission: 49.7% of 1,571 graded copies came back PSA 10 ($150), and only a PSA 10 covers the $92 you have in it — a 9 loses money. Confidence high on that sample.
$21$33$150$71+$5849.7%−$31,571
high
42Muk & Alolan Muk-GXMuk & Alolan Muk-GXUnbroken Bonds #196 · Modern (2011–2019)
Expected loss −$3 per submission: 55.7% of 713 graded copies came back PSA 10 ($156), and only a PSA 10 covers the $106 you have in it — a 9 loses money. Confidence high on that sample.
$34$40$156$72+$5155.7%−$3713
high
43Venomoth-GXVenomoth-GXUnbroken Bonds #216 · Modern (2011–2019)
Expected loss −$5 per submission: 46% of 248 graded copies came back PSA 10 ($155), and only a PSA 10 covers the $94 you have in it — a 9 loses money. Confidence high on that sample.
$22$35$155$72+$6246%−$5248
high
44Whimsicott-GXWhimsicott-GXUnbroken Bonds #226 · Modern (2011–2019)
Expected loss −$6 per submission: 43.1% of 253 graded copies came back PSA 10 ($165), and only a PSA 10 covers the $97 you have in it — a 9 loses money. Confidence high on that sample.
$25$37$165$72+$6843.1%−$6253
high
45Persian-GXPersian-GXUnbroken Bonds #227 · Modern (2011–2019)
Expected loss −$7 per submission: 39.3% of 303 graded copies came back PSA 10 ($173), and only a PSA 10 covers the $93 you have in it — a 9 loses money. Confidence high on that sample.
$21$34$173$72+$8039.3%−$7303
high
46WelderWelderUnbroken Bonds #214 · Modern (2011–2019)
Expected loss −$9 per submission: 44.3% of 1,250 graded copies came back PSA 10 ($167), and only a PSA 10 covers the $113 you have in it — a 9 loses money. Confidence high on that sample.
$41$60$167$72+$5544.3%−$91,250
high
47Venomoth-GXVenomoth-GXUnbroken Bonds #193 · Modern (2011–2019)
Expected loss −$22 per submission: 57.1% of 184 graded copies came back PSA 10 ($88), and only a PSA 10 covers the $81 you have in it — a 9 loses money. Confidence high on that sample.
$10$24$88$71+$757.1%−$22184
high
48Green's ExplorationGreen's ExplorationUnbroken Bonds #209 · Modern (2011–2019)
Expected loss −$25 per submission: 45% of 2,393 graded copies came back PSA 10 ($217), and only a PSA 10 covers the $182 you have in it — a 9 loses money. Confidence high on that sample.
$110$115$217$72+$3445%−$252,393
high
49Venomoth-GXVenomoth-GXUnbroken Bonds #12 · Modern (2011–2019)
Expected loss −$26 per submission: 56.6% of 175 graded copies came back PSA 10 ($81), and only a PSA 10 covers the $77 you have in it — a 9 loses money. Confidence high on that sample.
$6$12$81$71+$456.6%−$26175
high
50Celesteela-GXCelesteela-GXUnbroken Bonds #208 · Modern (2011–2019)
Expected loss −$26 per submission: 56.5% of 186 graded copies came back PSA 10 ($80), and only a PSA 10 covers the $78 you have in it — a 9 loses money. Confidence high on that sample.
$7$17$80$71+$256.5%−$26186
high
51Dedenne-GXDedenne-GXUnbroken Bonds #57 · Modern (2011–2019)
Expected loss −$31 per submission: 36.5% of 189 graded copies came back PSA 10 ($100), and only a PSA 10 covers the $83 you have in it — a 9 loses money. Confidence high on that sample.
$12$30$100$71+$1736.5%−$31189
high
52JanineJanineUnbroken Bonds #210 · Modern (2011–2019)
Expected loss −$31 per submission: 45% of 509 graded copies came back PSA 10 ($102), and only a PSA 10 covers the $92 you have in it — a 9 loses money. Confidence high on that sample.
$21$31$102$71+$1145%−$31509
high
53ArcanineArcanineUnbroken Bonds #22 · Modern (2011–2019)
Expected loss −$51 per submission: 10.1% of 158 graded copies came back PSA 10 ($123), and only a PSA 10 covers the $77 you have in it — a 9 loses money. Confidence high on that sample.
$6$26$123$71+$4610.1%−$51158
high
54MewtwoMewtwoUnbroken Bonds #75 · Modern (2011–2019)
Expected loss −$68 per submission: 19% of 247 graded copies came back PSA 10 ($191), and only a PSA 10 covers the $124 you have in it — a 9 loses money. Confidence high on that sample.
$52$25$191$72+$6719%−$68247
high

How the ranking works — and where it can mislead

Expected profit = (gem rate × PSA 10) + (PSA 9 rate × PSA 9) + (sub-9 rate × the card's own sub-9 prices) − raw pricegrading cost. Best case is the PSA 10 leg alone. Grading cost is the PSA tier fee for the graded value plus roughly $20 round-trip shipping, at fees as of 2026-08-26; the paused PSA Value tier is excluded and membership is not included.

Population selection bias. Gem rates are the share of the graded population at PSA 10 (PSA population report; CGC's when PSA has fewer than 10 copies). That population is self-selected: collectors grade their cleanest copies, and near misses get cracked and resubmitted, so it overstates what a random raw copy would achieve. Read a gem rate as an upper bound, and read it against its sample — confidence is high from 100, medium from 30, low from 10 graded copies; under 10, the card is shown as best-case only rather than ranked.

Prices. Raw is CardTrack's blended near-mint price for the exact printing; PSA 9 and PSA 10 come from the card's graded price ladder, with a PSA 9 estimated at 0.4× the 10 when the ladder has no 9. Where the ladder has no price for a lower grade, that leg is valued at raw. None of this is a sold price for your copy — it is a model of what the market has paid, and grading outcomes are not guaranteed.

PSA grading standards

Centering is the one criterion PSA publishes as a hard number, so it is the only part of a grade you can check yourself before paying. The ratios below are the widest border PSA allows at each grade — a card is judged on its worse axis, and the reverse is graded too. Meeting the centering limit does not earn the grade: corners, edges and surface all still have to hold up.

GradeFrontBackWhat the card looks like
PSA 10 GEM-MT55/4575/25Effectively flawless to the naked eye. Four sharp corners, clean edges, full original gloss and no staining. A minor printing flaw can survive if it does not hurt the card's overall look, but wear cannot — this is the grade where centering is tightest and most 10 attempts fail.
PSA 9 MINT60/4090/10Mint to the eye, with one small permitted flaw: a faint surface imperfection, a tiny printing blemish, or slightly off-white borders. Corners and edges still look sharp. The usual difference between a 9 and a 10 is centering or a single point of very light wear.
PSA 8 NM-MT65/3590/10Very light wear that needs looking for. Typically one slightly fuzzy corner, a minor surface mark, or minimally off-white borders. Still presents as a clean card at arm's length.
PSA 7 NM70/3090/10Light wear that is visible without hunting for it — slight fraying at a corner or two, minor surface wear, or a small loss of gloss. No creasing.
PSA 6 EX-MT80/2090/10Obvious but modest handling. Corners show visible rounding, edges show wear, and the surface may have lost some gloss. Colour is still strong and the card is free of creases.
PSA 5 EX85/1590/10Clearly played. Rounded corners, noticeable edge wear and some surface scuffing, with colour beginning to dull. Still whole and presentable.
PSA 4 VG-EX85/1590/10Well handled. Corner rounding is pronounced, surface wear is easy to see, and a very light crease may be present without dropping the card further.
PSA 3 VG90/1090/10Heavy handling. Rounded corners, dulled colour, scuffing or scratching, and light creasing. The card reads as used rather than collected.
PSA 2 GOOD90/1090/10Substantial wear throughout — heavy corner rounding, creasing, surface damage or writing. Intact and identifiable, but plainly worn.
PSA 1.5 FR90/1090/10Severe wear, stopping short of destroyed. Heavy creasing, major surface loss or damage, but the card is complete with no missing pieces.
PSA 1 PRThe floor. Extreme damage — major creasing, staining, tears, missing paper or pieces. Graded so the card can be authenticated and slabbed at all, not because condition is being certified.

Qualifiers

When a single defect is the only thing holding a card back, PSA can grade it with a qualifier — a slab reading PSA 8 (OC) rather than dropping to a 6. Qualified cards usually trade below an unqualified card of the same number, so the letters matter when you are pricing one.

  • OCOff-Centre: Centering is outside the tolerance for the grade the card would otherwise earn.
  • STStaining: A stain the card is otherwise clean enough to have avoided.
  • PDPrint Defect: A flaw introduced when the card was printed, not by handling.
  • OFOut of Focus: The image itself was printed soft — a manufacturing fault, not wear.
  • MKMarks: Ink or pencil marks on the card.
  • MCMiscut: Cut outside the intended borders at the factory.

Grade criteria and centering tolerances are published by PSA. Summaries here are ours. See the official PSA grading standards for the authoritative text.

Grading FAQ

Which Pokémon cards are worth grading?

The cards whose expected graded value — PSA 10, PSA 9 and lower grades each weighted by how often that card actually receives them — beats the raw price plus the grading fee. Right now that is led by Gardevoir & Sylveon-GX (Unbroken Bonds) at an expected +$1,347 per submission.

How is expected grading profit calculated?

Expected value = (gem rate × PSA 10 price) + (PSA 9 rate × PSA 9 price) + (sub-9 rate × the card's own sub-9 prices) − raw price − grading cost. Rates come from the PSA population report (CGC when PSA has fewer than 10 copies); the grading cost is the PSA tier fee for the graded value plus roughly $20 round-trip shipping, at fees as of 2026-08-26.

Why is a card with a huge PSA 10 price not at the top?

Because a PSA 10 price only matters as often as the card gems. A card with fewer than 10 graded copies has no measurable gem rate, so it is listed separately as best-case only rather than ranked above cards with evidence.

Are gem rates reliable?

They are the share of the graded population at PSA 10, and that population is self-selected: people grade their cleanest copies and crack-and-resubmit near misses. Treat the gem rate as an upper bound for a random raw copy, and weight the sample size — the confidence label on each row reflects it.