Read four numbers off a DIG report and get the TAG Score a slab would carry, with the error stated out loud instead of hidden.
Built by fitting real certs. Not affiliated with Technical Authentication & Grading.
Every TAG'd card carries a Grade — the familiar 1-10 scale — and a DIG Report that shows its centering and DINGS. Far fewer carry an actual TAG Score: the precise 1000-point number, from a DIG Report Plus, that tells you whether a 10 is a high 10 or a low 10. Most graded cards on the market are unscored — grade only, no DIG Report Plus, no TAG Score.
That's a real gap. A grade-10 card whose centering would score 987 looks identical, in a listing, to one that would barely clear 951 — same slab, same grade, same asking price, because nobody paid for the DIG Report Plus that would tell them apart. This tool reads the centering and DINGS numbers already sitting on every standard DIG Report — no DIG Report Plus required — and estimates the TAG Score the card would carry if it had one.
It doesn't replace TAG's own scoring, and it isn't trying to. It's built first for the grade 10 and Pristine range — the calibration set is real Gem Mint and Pristine certs, so that's where the estimate is most trustworthy and where the buy/walk call is firmest. Find a raw grade-10 slab whose centering says it would clear the buy line, and that's a real, calibrated gem: worth submitting for a DIG Report Plus, worth paying up for, or worth buying outright because the seller is pricing it like an average 10.
Below grade 10, the tool still runs, but it grades on a curve on purpose, and a strict one. A submission costs money, and the estimate is extrapolating past the range it was actually fit on, so it only reads CONSIDER on a card sitting in the top 5% of its own grade's point window — a genuine outlier, not just a decent example, and still just a cue to go comp it, not a green light to pay whatever's asked. The next 5% gets a MAYBE. Everything else at grade 9 and below reads WALK: not because the card is flawed, but because an ordinary example of its grade isn't worth paying TAG to confirm. The gems are out there unscored, mostly at the top of the grade 10 pile — this tool is built to find those first.
These do not move the estimate. Filling them computes the worst wear reading for you and tells you whether the card meets your own bar. A reading of 992 and a reading of 999 score the same, but they are not the same card in hand.
The line refits every time you add a real card. Only clean cards with a printed score belong here — a card with DINGS has an unmeasurable penalty and would poison the fit. Add yours and the estimate gets better for you; if this ever runs on a shared backend, it gets better for everyone.
my.taggrading.com/card/ and add the cert number to the end. Everything you need is on that one page.C around the card image. Type them into Left, Right, Top and Bottom in section 01. The tool checks that each pair adds to 100, so a typo gets caught.B:, written like 52L/48R 50T/50B.One optional extra. Section 02 lets you type the full corner and edge tables, which does not change the estimate but tells you whether the card meets your own bar. It isn't required.
score = 1100.69 − 7.941 × (front − 50) − 1.979 × back
Fitted on ten zero-DINGS cards spanning scores 930 to 988, front centering 50.73 to 57.23 and back centering 51 to 57. In-sample mean error 0.90, worst residual 1.65. Cross-validated 2.17.
| Model (zero-DINGS, n=10) | Cross-val error |
|---|---|
| front only | 4.21 |
| front + back | 2.17 |
| front + worst wear | 4.40 |
Back centering costs about 2 points per percentage point. It took a while to see because the first batch of clean cards all sat between 51 and 53. Once the range widened to 51–57 it separated cleanly. Q8710593 has better front centering than a confirmed Pristine and better wear, but a back of 55 — and it scored 985, not 990+.
Wear does not predict anything on clean cards. Every zero-DINGS card in the set reads 990 or better. An apparent wear effect in an earlier version was DINGS damage in disguise.
Why it cannot be narrowed. M9716831 and X2142466 each carry exactly one back-surface DING. One lost 1.3 points, the other 328. Same type, same count, 327-point spread. Penalty correlates with DING count at +0.17 and with the worst wear reading at +0.40 — both noise. The DIG report says a DING exists and where it is. It never says how bad.
This is the arithmetic behind your own rule. A clean card is predictable to within a point. A card with one DING is a coin flip between losing nothing and losing a third of its score.
The printed grade fixes the window. Your severity call places the card inside it. Judge the card in hand under a single light source, tilted.
| Level | What you see | Measured cards |
|---|---|---|
| 1 | Nothing in hand. The DIG map finds it, your eye does not. | lost 1–4 |
| 2 | Findable if you know where to look. Faint corner whitening or a pinpoint nick under raking light. | lost 7–25 |
| 3 | Visible at normal viewing distance on close inspection. Clear corner white, a short surface line, a definite edge nick. | lost 66–148 |
| 4 | Obvious at arm's length. Several whitened corners, a crease, or a scratch that catches light across the card. | lost 170–258 |
| 5 | The first thing you notice about the card. Heavy creasing, delamination, large surface damage. | lost 294–327 |
Levels 1 and 2 cover nine of the fifteen measured cards; levels 3 to 5 cover six. If you are torn between two levels, pick the lower one — the tool is more forgiving of underestimating than overestimating (mean error 7.8 versus 8.6 when a level off).
Every TAG grade is a fixed 50-point block: 10 Pristine is 990–1000, 10 Gem Mint is 950–989, 9 is 900–949, 8.5 is 850–899, and so on down. Knowing the grade collapses the answer to a 50-point window before any modelling happens.
| Estimate for a DINGS card | Mean error | Worst |
|---|---|---|
| centering ceiling alone | 90.5 | 327 |
| + printed grade | 11.2 | 48 |
| + severity placed in the window | 7.8 | 36 |
On a clean card the grade does something quieter but still useful: it stops the model claiming Pristine on a slab that says Gem Mint, and stops it dipping below the floor of the grade you already hold.
| Check | Go | Caution | Stop |
|---|---|---|---|
| Total DINGS | 0 | — | 1+ |
| Front centering, worse half (10 / 10P only) | ≤ 52.0 | 52.0–52.5 | > 52.5 |
| Back centering, worse half (10 / 10P only) | ≤ 52.0 | 52.0–52.5 | > 52.5 |
| Lowest corner angle | ≥ 995 | 992–994 | < 992 |
| Lowest edge / corner fray or fill | ≥ 995 | 990–994 | < 990 |
| Dimensions | 3.45–3.47 × 2.47–2.49 | — | outside |
52.0 is the buy line, where the resale premium starts. 52.5 is the visual line, past which miscentering becomes perceptible on a borderless card. Both were fitted on Gem Mint and Pristine cards, so they only gate the verdict at those two grades — at grade 9 and below, worse centering is expected, not a defect, and the tool shows the numbers without failing the card over them.
Grade 9 and below get a different verdict check: no centering line, but a band-position check instead. Every grade below Gem Mint is a fixed point window (grade 9 is 900–949, for example). Submitting a card for a TAG Score costs money, so the bar is deliberately strict — the tool takes the estimated score and asks where it falls in that window: top 5% reads CONSIDER, the next 5% (5–10%) reads MAYBE, the bottom 90% reads WALK. Not because the card is flawed, but because an ordinary example of its grade isn't worth paying to score. A real defect — a DING, a bad corner or edge, worn surface — overrides this and reads WALK regardless of where the score lands.
Edge deficit is the sum of every edge reading below 1000. It barely touches the score, but it is the best proxy for how the card was handled before it was sleeved. Reference: J7795668 carries 1, U8580361 carries 2, D5026965 carries 26, P5621262 carries 21.
This tool used to output a dollar price ceiling from a fixed multiplier table. It's gone, on purpose — the table was built from three sales of one card (a Blastoise ex 200/165), and even on that single card the relationship wasn't clean: a 964 took $565, a 974 took $750, and a 984 only took $40 more than that at $790. If the premium curve doesn't even hold steady for one card's own sales, it can't be generalized into a formula that works for every card. A cheap bulk common and a five-figure chase card do not react to the same score the same way, and no static table was ever going to capture that honestly.
Comp it yourself instead, off real sold listings, not asking prices. A "comp" is a completed sale of a directly comparable card — same card, same grade, similar centering if you can find it — not what someone is currently asking, which is aspirational by default. Pull 3–5 recent sold listings (eBay's sold/completed filter, 130point.com, or an auction house's past-results page for anything higher-end), and use the median, not the average — one outlier sale shouldn't set your number. Weight the most recent sales heaviest; card markets move.
If you can find even one sold comp of the exact same card with a published TAG Score, that single real data point is worth more than any generalized multiplier — use it directly instead of estimating a premium from scratch.
Same badge, three different real answers — this is why the verdict is never the final word:
Chase card, real comps. Grade 9, top 5% of its window. Three sold comps for scored copies of this exact card average $180 over the plain price.
Comp says: worth it — the premium clears the $105 conversion cost with plenty to spare.
Same badge, no comps yet. Grade 10, clears the buy line. Nobody's sold a high-scoring copy of this card yet — no data either way.
Comp says: maybe — pay plain-10 price and treat any score premium as upside, not the reason you bought it.
Same badge, bulk card. Grade 9, top 5% of its window. Plain copies of this card sell for $12.
Comp says: walk — even a full premium can't clear the $105 it costs to submit.
The one number that doesn't vary by card: converting an unscored slab to a DIG Report Plus costs about $105 all-in — EXPRESS submission at $59 plus shipping both ways. EXPRESS is the cheapest tier whose $1,000 insurance cap actually covers a card in the range this tool is built for; STANDARD caps out at $500. Whatever premium you estimate for a high score, it only clears the bar if it beats that $105 plus the wait. That's exactly why the verdict above is strict about grade 9 and below: an ordinary card at that grade almost never clears $105 in extra value, so the tool tells you to walk before you spend it.