The other page works out whether grading pays, and the answer there is almost always that it only pays on a 10. This page is the part that was missing: how to tell whether you are holding one, before you spend the fee finding out.
Everything below was read off a grading company's own published standard on August 15, 2026. Where the hobby says one thing and no company published it, that is said plainly instead of repeated.
Beckett is the only company that still publishes both the four component scores and how they combine, and its rules are the most useful thing on this page. Source
Three perfect attributes and one bad one is a bad card. That third example is a card that is perfect on three of four criteria and comes back a 7. When you are checking your own card, you are not looking for how good it is. You are looking for the worst thing on it, because that is the only number that matters.
Beckett published this in February 2018 and last modified it in November 2020. It was still live and its grade names still match the current scale when this was read on August 15, 2026, but it is an old article and worth treating as the rule of thumb it is.
Centering is the only thing on this list you can measure at home instead of judging, so start here. It is also where the companies disagree most.
How it is measured. A card's centering is found by measuring the distance between the photo and the edge of the card from opposite sides. A card described as 50/50 top to bottom and 60/40 left to right has 60 percent of the border on one side of the photo and 40 percent on the other. PSA says so. Source
Front centering, as each company publishes it
| Grade | PSA | CGC | Beckett (BGS) | SGC | TAG |
|---|---|---|---|---|---|
| Pristine 10 | not published | 50/50 | 50/50 all around | 50/50 | about 51/49 |
| Gem Mint 10 | about 55/45 | about 55/45 | 55/45 both ways (their 9.5) | 55/45 or better | about 55/45 |
| Mint 9 | about 60/40 | 60/40, sports and non-sports only | 55/45 both ways | 60/40 or better | about 60/40 |
| Near Mint-Mint 8 | about 65/35 | 65/35, sports and non-sports only | 60/40 both ways | 65/35 or better | about 65/35 |
| Near Mint 7 | about 70/30 | 70/30, sports and non-sports only | 65/35 both ways | 70/30 or better | about 70/30 |
Back centering. Note how much wider the tolerances are, and how many are simply absent
| Grade | PSA | CGC | Beckett (BGS) | SGC | TAG |
|---|---|---|---|---|---|
| Pristine 10 | not published | not published | 55/45 or better | not published | about 52/48 |
| Gem Mint 10 | about 75/25 | 75/25 | 60/40 or better (their 9.5) | not published | about 65/35 |
| Mint 9 | about 90/10 | 90/10, sports and non-sports only | 70/30 or better | not published | about 75/25 |
| Near Mint-Mint 8 | about 90/10 | not published | 80/20 or better | not published | about 95/5 |
| Near Mint 7 | about 90/10 | not published | 90/10 or better | not published | not published |
PSA gives itself room, and tells you so. A 5% leeway is given to the front centering minimum standards for cards which grade PSA 7 or better. PSA gives its own example: a card meeting every other requirement for a 9 and measuring 60/40 off center on the front automatically meets the front centering standard for a 9. Source
PSA and CGC both allow roughly 75/25 on the back of a card they will still call a 10, against 55/45 on the front. A back that looks badly off to you may not be the problem. A front that looks slightly off probably is.
Its rubric splits the back tolerance by category: about 65/35 for TCG at Gem Mint against about 70/30 for sports, and about 75/25 for TCG at Mint 9 against about 90/10 for sports. The front number is the same for both. Every other company publishes one set of numbers and leaves you to assume it covers Pokemon.
From grade 9 down its wording is conditional: the 60/40 and 90/10 figures are given 'for sports and non-sports cards'. No equivalent number is published for trading cards at 9 and below, which get qualitative language instead. So if you are trying to work out whether a card is a CGC 9 or a CGC 8, there is no published tolerance to measure against.
No back tolerance appears at any grade on its scale. Its scale also carries an asterisk on 'four sharp corners' at grades 8.5, 9, 10 and Pristine, and the footnote that asterisk points to does not render anywhere on the page as served.
The gap between a 10 and a 9 is smaller than people expect and it is written down. Two companies state it in a single sentence each.
PSA 9 is defined as a card showing only ONE of a very slight wax stain on the reverse, a minor printing imperfection, or slightly off-white borders. Two of those and it is not a 9. The 10 wants four sharp corners, sharp focus and full original gloss, and must be free of staining of any kind. Note the one thing it does NOT rule out: PSA says an allowance may be made for a slight printing imperfection at 10 if it does not impair the overall appeal of the card. So a faint print flaw is not automatically a 9. Source
At Mint 9 Beckett allows clean gloss with one or two tiny scratches barely noticeable to the naked eye, and one faint unobtrusive metallic print line. At 9.5 and above the surface must be free of scratches and metallic print lines entirely. On a holo that single print line is the whole difference, and it is something you can actually go and look for. Source
CGC defines Gem Mint 10 as a card that scored 10 overall but where one criterion did not meet Pristine. Pristine 10 is reserved for cards that are flawless under 10 times magnification. Two different labels, and only one of them is the top. Source
Only one company still prints four component scores on the label, and only that company publishes how they combine into the final grade. One other scores per attribute without calling them subgrades. The arithmetic below is the most useful thing in this section, because it tells you which single flaw decides everything.
Four subgrades on the label: centering, corners, edges, surface. Four 10s earns the Black Label. Subgrades are a paid option on the Base tier, and there is an extra fee for any 10s.
Discontinued. CGC's help center states plainly that subgrades are no longer offered.
No subgrade product. PSA uses qualifiers instead, which cap the grade rather than sit beside it: OC off center, MC miscut, ST staining, PD print defect, OF out of focus, MK marks.
No subgrades on its scale or services pages.
Per-attribute scoring and a 1000 point score. Corner and edge wear are scored on all four corners on both faces, for eight scores each.
The hobby and the grading companies do not use the same words, and the most important one is a word no company uses at all.
The word every collector uses does not appear in any of the five majors' published standards. PSA calls it chipping, and its definition is written for exactly the kind of card Pokemon prints: it says chipping is most noticeable on sets with colored borders, because of the contrast between the border and the lighter cardboard underneath the thin layer of ink, and that it can come from aging, handling, or even dull blades at the factory. So a card can arrive chipped from the factory, and it counts the same as the chipping you caused. Source
Beckett is the one that grades it in steps you can check yourself: at 9 it allows unobtrusive specks of chipping on the borders, at 8 specks visible to the naked eye, and at 9.5 and above the edges must be virtually flawless. So visible white on an edge is roughly an 8 ceiling. Beckett
A straight line from the printing process, caused by a scratch on the printing plate holding ink that never reaches the card. Not damage, not your fault, and it still costs the grade. TAG
No company will tell you foil scratches more easily than plain card stock. What they do is grade a foil-only flaw separately at every single level. Beckett names metallic print lines the whole way down its surface scale: none at all at Pristine, one faint unobtrusive line allowed at Mint 9, a few by Near Mint 7. SGC uses its own term, refractor lines, and allows a minor one at Mint 9 and none at Pristine. So the flaw that only foil gets has its own published tolerance, which is the practical version of the thing nobody states outright. Beckett and SGC
A rounded indentation from the printing process. TAG
This is the one rippers actually hit. CGC gives the tell: packaging roller damage shows as two vertical indentations on BOTH faces of the card, while a pinched layer from manufacturing shows only on the holo side. Check the back before you blame yourself. CGC
White patches, dots and circular discolored spots left by the printing process. PSA files all of these under its PD qualifier. PSA and TAG
The distinction is worth knowing because it decides how bad the news is. A wrinkle shows on one side only. A crease is usually visible on both. PSA puts a number on the crease: it can drop a card one to five grades. TAG and PSA
Indentations along an edge, sometimes from a rubber band. PSA
A pit is a circular surface mark, a dent is a depression in the surface, a scratch takes the appearance of straight or curved lines. TAG
A skewed factory cut, so the card is not a true rectangle. Named by all five majors and allowed progressively further down every scale. all five
Holofoil showing through areas the manufacturer did not intend. PSA recognizes it and treats it as a print quality matter rather than an authenticity one. CGC states outright that it does not recognize so called holo bleeds. This is the clearest case on the page of the same card getting two different readings depending on where you send it. PSA and CGC disagree
Things that are specific to modern Pokemon and sourced to a grading company or to The Pokemon Company itself, rather than to hobby lore.
CGC states that off-center errors are among the most common printing anomalies in trading card games. If you are opening packs and finding off-center cards, nothing has gone wrong. That is what the process produces. Source
This is the strongest source on the page because it is not a grading company talking. The Pokemon Company's own support pages claim cards pass a thorough quality assurance process and then state, in the same breath, that cards with minor scratching or printing defects do not qualify for exchange. The manufacturer will not replace a scratched or misprinted card, because a scratched or misprinted card is a normal card. Source
CGC's error guide describes cards slipping into the heat-sealing crimper as the pack is closed, so the crimping machine's pattern is pressed into the card. Paper jams, fold-overs, holo splices and insufficient ink all happen inside sealed packs too. The presses run up to 15,000 sheets an hour on sheets of 100 to 121 cards. Damage arriving sealed is ordinary. Source
Texture has been a Pokemon feature since Black and White: Next Destinies, and CGC lists incomplete, misaligned and missing texture as recurring problems. It also adds a warning that matters more than the grade: most counterfeits share the missing-texture quality, so a card with no texture where there should be some is worth checking against the fake-card guide before you send it anywhere. Source Spotting fakes
CGC names specific ones: misregistration between black layers is particularly plentiful on commons and uncommons from Ultra Prism, albino backs are known from Rebel Clash, Sword and Shield base and Vivid Voltage reverse holos, inverted back texture has so far only been seen on Evolving Skies V holos, and texture-layer-only has only been seen on Crown Zenith Galarian Gallery cards. These are error frequencies. No company publishes a claim that any Pokemon set has recurring centering problems. Source
CGC calls Pokemon Jungle a big offender for rough edges, and names holo roller lines as a recurring problem on earlier Pokemon cards. It also publishes its policy on all of it: depending on the severity and prevalence of a production defect, it grades the card with that in mind. A flaw that every copy of a card has is treated differently from one only yours has. Source
What the companies themselves tell you to do before submitting, and where their own advice stops.
PSA's own getting-started advice: put the card in a penny sleeve and then a semi-rigid holder before you start handling it to inspect it. Source
CGC published its own graders' method in writing. The lamps use an LED light, described as a nice clean white light, which is what you want for grading cards. The technique is to hold the card perpendicular to the light source and then move it around so the reflection of the bulb travels across the card. That is how surface flaws show themselves: not by staring at the card flat, but by watching the reflection cross it. Source
This is the only place any of the companies publishes an inspection method as text rather than video. It is written for sports cards and comes from the arm CGC merged in, so treat it as their method rather than a Pokemon standard.
The same transcript says graders remove the card from the semi rigid holder and the penny sleeve, and look at all four corners on both the front and the back. Eight corner checks, not four. Source
CGC specifies the tool: a 10x power loupe with a triplet lens corrected for both color and distortion, which it calls the standard format for professional card graders. Its top grade is defined at exactly that power, as flawless under 10 times magnification. A reader with a 10x loupe is looking at the card the same way the grader will. Source
Beckett grades both faces, PSA and CGC publish separate reverse centering tolerances, and TAG scores corners and edges on both faces for eight scores each. The back is where most people never look and it carries its own numbers.
Because of how subgrades combine, one bad attribute caps everything. Find the worst feature of the card and grade that, then stop.
PSA says outright that a grader must sometimes make a call on a card falling between two grades, and that the final determination is made on experience, eye appeal and market acceptability. It also gives the reverse case: a card that technically falls within the printed standards can still be held back because the market would view it as off center. So the numbers get you close and then a judgment you cannot make from home decides it. Source
CGC names the specific thing you will get wrong: the surface of a card often hides flaws and is one of the tougher areas to analyze at home. If you are going to misjudge your own card, that is where. Source
If your card comes back lower than you expected and you want to know why, PSA's Grader Notes detailing the defects are listed on the Express tier at $149 and above. They are not listed on the $79.99 Regular tier. Budget for a grade with no reason attached. Source
An app cannot grade your card, and the ones that say they can are the ones to be most careful with. What a phone can genuinely do is two things: measure the centering, which is geometry you could check yourself with a ruler, and work out the grade this card has to come back at before submitting it makes any money.
A dozen apps will look at a photo and tell you the card is a 9.5. Not one of them has ever published a test you could check. Every accuracy figure in the category traces back to the company selling the app, and the ones shouting loudest have the least behind them: two separate services advertise 95 percent accuracy with no sample size, no method and no log. The one service that published a real-sounding claim, 87 percent within half a grade across 412 cards sent to PSA and back, promises the full per-card log on a page that does not contain one.
Watch the wording on all of them. The claims are always within one grade. The question you are actually asking is whether the card is a 10 or a 9, so a tool that is right within one grade is, by its own description, no use for the only decision you came to make.
Test one yourself. You can test one yourself in five minutes and it costs nothing. Scan the same card twice, then scan a card you already know is damaged. An app worth paying for gives you the same answer twice and notices the damage. Readers report neither. One scanned a brand new card, deliberately dented it, scanned it again and got the same grade with no mention of the dent. Another scanned three slabbed cards, a Beckett 9.5, a PSA 10 and a Black Label 10, and the app returned 7.5, 6 and 8.5. A third destroyed a card on purpose and it still came back a 9.
There is a reason it does not work, and it is the camera rather than the software. When researchers at Liverpool John Moores built a corner grader, they gave up on cameras and switched to a flatbed scanner because they needed constant lighting, and even then they reached 83 percent on corners alone. A team at Hull built a robot with a Canon body and two softboxes to get around what they named the shadow problem. The companies that genuinely grade by machine use photometric stereo, laser topography and microscopes. And PSA prints the quiet part on its own Photograde page: the scanning process does not always pick up every detail or defect. That is PSA warning you about its own archival scans. Your phone, held over a card on a kitchen table, is a weaker instrument than the one PSA is disclaiming.
PSA bought an AI grading company. Genamint, in April 2021, and its founder became a PSA vice president. So the company holding more graded card data than anyone on earth has owned a computer vision grading operation for five years. Its free app has been downloaded enough to carry fifty four thousand ratings. It identifies your card, prices it, checks certification numbers and builds your submission. It will not tell you what grade the card is going to get. Five years, all that data, and the people who actually assign the number still will not put a prediction in your hand.
CGC says out loud what the machines are genuinely for. Its own help pages describe AI automating the time-consuming parts of grading, and it names them: attributing the card, and measuring the centering. Attribution and geometry. Not the grade.
In July 2026 a collector put thirty Pokemon cards through five of these apps, sent the cards off to be graded, and published the per-card results instead of just the summary. The apps landed inside one grade most of the time, between 83 and 97 percent, which sounds fine until you look at what happened on the cards that came back Gem Mint. Not one app called a single 10 correctly. Every one of them under-graded every gem card. A Mega Sharpedo that came back a 10 was called 8.4, 9.0, 7.3, 7.0 and 7.0.
That is the opposite of the failure everybody expects. The worry is always that an app talks you into a submission by flattering the card. On this data it does the reverse and talks you out of the only cards worth sending. Either way it is the same problem: the number the app gives you is not connected to the number you get back.
Thirty cards, one person, and the reference grades came from TAG rather than PSA, so treat it as one honest experiment and not a settled study. It is also the only published independent test of these apps on Pokemon cards that exists, which is the more telling thing about it.
Jade Lizard Software
Measures the centering ratio off a photo and checks it against the published thresholds for PSA, BGS, SGC, CGC, TAG and ACE. It does not predict a grade, and does not claim to.
The manual guides are the trustworthy part. Reviewers report the automatic edge detection giving different readings for the same card, and the ads are aggressive enough that several people complain about them.
App Store • iPhone only
Jade Lizard Software
Works out your break-even grade: the lowest grade that still turns a profit once grading, shipping, insurance and selling fees are paid. It looks at no photograph at all.
The only Pokemon-only app here, and the only one that answers the question this page is actually about. Last updated September 2025, so it is not being actively worked on.
App Store • iPhone only
Collectors Universe
Identifies cards, checks certification numbers, builds and tracks a submission. It does not assess condition at all, and does not pretend to.
By far the biggest rating base of any app here, and it is the grading company's own. Worth having once you have decided to send something in. No help at all in deciding, and it does not claim to be.
SGC will grade a card, slab it and stand behind the number for fifteen dollars. CGC Economy is twenty. Against that, one popular predictor charges $9.99 a week, which is $519 a year, to guess. At those two numbers a subscriber has to correctly talk themselves out of about thirty five bad submissions a year, three a month every month, using a tool readers say scores two visibly different cards the same. Nobody ripping packs on weekends submits thirty five marginal cards a year.
Several of these listings also have a billing complaint problem that is documented in the stores themselves. On one, a user reported a signup screen showing $0.00 and a $34 charge landing immediately, and the developer replied in public that this has been reported multiple times by different users. On another, a free trial charged a lifetime membership on the spot.
ACE Grading publishes a dispute policy that lists what it will not consider when you challenge a grade. Alongside previous grades from other companies it names pre-grade services and, in as many words, AI grading results and external automated assessments. It is the only published statement any grading company has made on the subject, and it is a refusal. Source
Scroll up. Measuring the borders yourself against each company's published tolerance is the same arithmetic the apps are doing, it is free, and you can see your own working.
GemRate publishes how often a specific card actually comes back a 10 across PSA, BGS, SGC and CGC, and PSA's own free app carries the same pop report. Knowing a card gems eight percent of the time is a harder number than any app's guess about your particular copy. Source
For surface, which is what actually decides a 9 against a 10, your eyes beat a phone camera. Hold an LED at a low raking angle and tilt the card. A jeweler's loupe costs about what one month of the expensive app costs.
An app claiming complete accuracy. One says exactly that on its own site, beside patented AI technology with no patent number and no test behind either.
A high star rating with bad written reviews under it. Stars come from a prompt inside the app; a written review costs somebody five minutes. One grader holds 4.6 stars from 529 ratings while seventeen of its twenty five most recent written reviews are one star. A reviewer of that app wrote that it asked for a five star review three times before he had even logged in for the first time.
Reviews traded for scans. A user of one app wrote, in the review itself, that he was posting it to unlock five minutes of the full version. Another wrote that a developer had reached out and graded a card for him free.
Ratings that arrive in a clump. One listing carries a block of five star reviews all posted within about three weeks, from two-name handles, several repeating the same sentence with the same deliberate misspelling, and a few with untranslated German left in the middle of the English.
Any blog post ranking the best grading app. The domains filling those search results were nearly all registered inside the last year by the app makers themselves, and every one ranks its own product first.
Any of it on a Japanese card. Not one app in either sweep mentions Japanese support, and several are trained on sports cards rather than Pokemon.
Ratings, prices and store listings read August 24, 2026. Nothing here is an affiliate link and nobody paid to be on this page.
Two published outcome numbers, and how to look up your own card.
PSA publishes that about 90 percent of modern trading card game cards from 2003 onward have graded at a PSA 8 or higher, and about 80 percent of pre-2003 vintage TCG cards at a 7 or higher, with its own note that all percentages are approximate and every card is graded individually without regard to them. Source
Read that number the way it is written, not the way it feels. It is a floor, not a forecast. Nine in ten clearing an 8 tells you nothing whatever about your odds of a 10, and on most cards an 8 does not cover the fee. This statistic is the most reassuring thing PSA publishes and it is not reassurance.
TAG publishes that its Pristine grade, which it says exceeds the industry standard for a Gem Mint 10, represents less than 1 percent of the cards it has graded. Source
Go and look up your exact card. The most useful thing you can do before submitting is look up how many 10s already exist for your exact card. CGC's population report is open with no account, organized by game, era and set. Beckett's and SGC's reports also open without an account. Source PSA's does not. Its population report redirects to a sign-in page. So the one company whose 10 most people actually want is the one you cannot check without an account.
Things a page like this normally tells you, and what is actually behind each one. This list was longer before a second pass through the sources: the lighting and viewing-angle advice above was very nearly filed here as unpublished, and it turned out CGC prints it.
If the card survived all of that, the next question is whether the fee clears. What grading costs and whether it pays has the current fee table for all five companies and works the break-even out against each card's own price. If it did not survive, it is still a card: where to sell it raw is next door, and a raw card sells fine.
And if the card is from 1999, work out which printing before you decide anything. 1st Edition, Shadowless or Unlimited is a two mark check on the card in front of you, and the three runs are priced a long way apart.
Read from each company's own published standards and error guides on
August 15, 2026. Standards change and the companies do not announce it, so check before you
submit anything expensive. Where two pages of the same company disagreed, the standards page was used:
Beckett runs two live scale pages that disagree at the top grades. The current one gives Pristine 10 as 50/50 front and 55/45 or better on the back; an older one gives 60/40 on the back and describes Gem Mint 9.5 as 50/50 one way and 55/45 the other. This file follows the current page.
PSA's glossary and PSA's standards page disagree on front centering by a whole grade, and in more than one place: the glossary gives Near Mint 7 as no worse than 75/25 against the standards page's 70/30, and Near Mint-Mint 8 as 70/30 against 65/35. The glossary reads like it is one grade shifted throughout. This file follows the standards page everywhere.
Nothing here is a pull rate and nothing here is a promise about your card. The final call is a human
being's judgment, and every company says so.