Grading Opportunity Methodology

Grading a card costs real money and takes months, and the payoff depends on a grade nobody knows in advance. The grading tool estimates whether the trade is worth making: what each possible outcome would net you after fees, how likely each outcome is, and therefore the highest raw price at which submitting the card still makes sense. It runs on data we already hold, the published population distribution for that exact printing and our fair values for each grade, and it is careful to describe the result as the bet it actually is rather than as a margin. This page explains the model, and the places where it deliberately refuses to guess.

Maintained by the TCGBerg research team · Last updated August 20, 2026

What the Tool Answers

For one printing, the tool produces three things:

The break-even raw price. The most you could pay for the card raw and still expect grading to be worthwhile. Pay less and the submission is expected to profit; pay more and it is not. This is the headline number, because it is the one that translates directly into a decision at a listing.

The expected net. The break-even price minus what the card actually costs raw today, which is the expected surplus from making the trade.

A verdict among three routes to owning a graded copy, compared at today's prices:
  - Grade it: buy the card raw and submit it.
  - Buy it already graded: somebody else has already taken the risk and their slab is on the market cheaply enough to beat doing it yourself. This route only ever wins when there are actual slab listings to compare against.
  - Sell it raw: neither route clears zero, so grading this card destroys value. If you already own it, it is worth more as it is.

Alongside these the tool shows the full outcome ladder: each grade, its probability, what it would be worth, and what it would net.

Where the Odds Come From

The probability of each grade is the printing's own published population distribution: of every copy of this exact card that has ever been professionally slabbed, what share came back at each grade. It is a base rate drawn from real graded copies of the real card, which is why it differs so much between a card that gems easily and one with a notorious centring or print-line problem.

Three things about that base rate you should know, because they change how you read the answer.

It is not a scan of your card. The tool has never seen your copy. It is telling you how copies of this card have graded in general, not how yours will grade. Nothing in the model can distinguish your card from any other.

It is optimistic, because submissions are self-selected. People send in their better copies. The population report is a record of cards someone thought were worth the fee, not of a random sample pulled from a binder. For a card picked at random the true odds are worse than the reported distribution, and by an amount nobody can measure.

Declaring a condition shifts it. If you tell the tool the card is not near-mint, the whole distribution moves down the ladder accordingly. This is a blunt correction rather than a precise one, and it is applied honestly in one direction: a played card cannot grade better for having been described accurately.

What Each Outcome Is Actually Worth

Every grade is valued at what a sale would net, not at its headline fair value. From the fair value for that grade we subtract the marketplace fee that would apply at that price and the cost of shipping the card out. What remains is the money that would reach you.

The cost side is treated the same way. The grading cost in the model is the service fee plus shipping in both directions, not just the fee. A fee-only model quietly overstates every verdict it prints, and on the cheap cards where grading decisions are most marginal, shipping is a large share of the total.

Two deliberate conventions:

A grade that cannot cover its own selling costs contributes zero, not a negative. Nobody ships a slab at a loss; they keep it. So the floor is zero rather than a fictional negative outcome.

Grades with no fair value contribute zero too. We publish fair values for a range of grades, and outcomes outside that range, the low tail and the half grades, have no value to contribute. Rather than invent one, we count them as zero and report how much probability mass landed there, so a result resting heavily on unpriced outcomes can be read with the right amount of scepticism. This makes the estimate conservative by construction: it can understate a card, and it is built not to overstate one.

An Expected Value Is a Mean, and a Mean Describes a Lottery Badly

This is the part most grading calculators get wrong, so it is worth being blunt about.

An expected value averages over outcomes that are usually nowhere near each other. On a typical vintage card the top grade is the only outcome that makes money and every other grade loses. A single averaged number hides that completely: it reads like a margin when it is really a bet with a small chance of a large win.

So alongside the expected value the tool always shows:
  - The win probability: the chance of an outcome that does not lose money.
  - How concentrated the value is: how much of the whole expected value sits in a single grade.
  - The break-even grade: the specific grade that has to land for the submission to pay off.

A 36% win rate with most of the value in one grade is a bet, not a margin, and any tool that prints the expected value on its own is describing it dishonestly.

A second precision, about the percentage. There are two ratios you could quote and only one of them is a return. Dividing the expected profit by the raw price alone gives a number that explodes as the card gets cheaper, because the grading cost is subtracted in the top of the fraction and missing from the bottom: a card costing a dollar with a fifty-dollar grading bill can read in the thousands of percent, which is arithmetically true and completely useless. We rank and display the return on total cost, which includes the grading bill in the denominator, so the figure means what a return normally means and a card agrees with its own individual grade rows.

When the Tool Declines to Answer

The model rests on the population distribution, so where that distribution is not informative the tool says so instead of producing precise-looking output.

Below a minimum total population, the distribution is noise. A card with a handful of graded copies could show a 50% gem rate purely because one of two submissions came back clean. The tool reports insufficient population rather than a number.

Where fair values are missing across much of the ladder, the result is reported with its unpriced share visible, because a conservative estimate built mostly from zeros is a weak estimate even when the arithmetic is sound.

Where there are no slab listings, the buy-graded route simply cannot win, because there is nothing to price it against. That is an absence of data, not evidence that grading is the better option.

Limitations

What this tool is not:
  - It is not an assessment of your card. It is a base rate over other people's copies. Centring, edges, surface, and print defects on the copy in your hand are exactly the variables that decide your grade, and the model cannot see any of them.
  - The odds are optimistic. Population reports reflect self-selected submissions, so a random raw copy grades worse on average than the distribution suggests.
  - Prices are today's prices. Grading takes months. The values are current fair values and current slab asks, and the market can move a long way before your card comes back.
  - Fees are modelled, not quoted. Grading service tiers, bulk rates, promotions, and marketplace fee schedules all change, and your actual costs may differ from the ones modelled.
  - It covers PSA. The outcome distribution and the graded values both come from PSA data, so the answer is about a PSA submission.
  - It is not financial advice. It is a model estimate to inform a decision that stays yours.

Frequently asked

Does the tool look at a photo of my card?

No. It has never seen your copy. It tells you how copies of this printing have graded historically, which is a base rate over other people's cards. Your grade depends on the centring, edges, surface, and print quality of the specific card in your hand, and none of that is an input here.

Why are the odds described as optimistic?

Because population reports only count cards somebody chose to submit, and people submit their better copies. The distribution is a record of self-selected cards, not a random sample. A card pulled at random from a binder will do worse than the reported numbers on average, by an amount nobody can measure.

Why is the expected value not the headline number?

Because it averages outcomes that are nothing like each other. On many cards the top grade is the only one that makes money. Presenting one averaged figure makes a bet look like a margin, so the tool always shows the win probability, how much of the value sits in a single grade, and which grade has to land.

Why does the tool show a return on total cost rather than a return on the card price?

Because dividing profit by the card price alone leaves the grading bill out of the denominator, which makes the ratio explode as the card gets cheaper. A cheap card with an ordinary grading bill can read in the thousands of percent that way.Including the grading cost gives a figure that means what a return normally means, and that agrees with the individual grade rows underneath it.

What does the break-even price mean exactly?

It is the most you could pay for the card raw and still expect the submission to be worth making, after the grading service fee, shipping in both directions, and the marketplace fee on the eventual sale. Below it the trade is expected to profit; above it, it is not.

Why does it sometimes tell me to buy an already-graded copy?

Because sometimes that is cheaper. If somebody is selling a slab for less than what it would cost you to buy the raw card and grade it yourself, they have taken the risk and the timing hit and you have not. That comparison needs live slab listings, so where there are none this route never wins by default.

Why does it say there is not enough population data?

Because below a certain number of graded copies the distribution is noise rather than a base rate. A card with a handful of slabs can show a very high or very low gem rate on the strength of one or two submissions. We would rather decline than dress that up as precision.

Are grading costs and turnaround times included?

Costs are modelled, including shipping both ways rather than the service fee alone. Turnaround time is not modelled as a cost, but it is a real risk: values are current, and the market can move considerably in the months a submission takes to come back.