The expected value of grading a card is the probability-weighted payoff of every possible grade outcome, minus the cost of grading and the value of the raw card you already hold. Each grade tier has a market price and an estimated probability, and multiplying them out gives you the average result of submitting that card many times. If the sum exceeds the raw price plus fees, the submission has positive expected value; if not, you are paying for a slab, not making an investment. Treating grading as a priceable bet rather than a ritual is the single biggest upgrade most collectors can make to their process.
Every grading submission is a bet. Most collectors place it without ever asking what the odds are.
That is strange, because grading is one of the few bets in the hobby you can actually price. The grade outcomes are enumerable. Each outcome has a market price. The entry cost is printed on the grading company's website. The only genuinely uncertain input is the probability of each grade, and even that can be estimated from public population data. Everything you need to compute an expected value is sitting in the open, and almost nobody computes it.
This piece walks through the math, the places where it bites, and the cases where the correct answer is to leave the card raw.
The formula
The expected value of submitting a card is:
EV = Σ (probability of each grade × market value at that grade) − grading cost − raw value
Read it as a sentence: average across every outcome, then subtract what you pay to play and what you already had. If the result is positive, submitting the card creates value on average. If it is negative, you are buying a plastic case and a serial number.
Two of the three inputs are easy. Grading cost is a published fee plus shipping and insurance. Raw value is the current market price of the ungraded card, which is what you are giving up: the moment a card goes into a slab with a 9 on it, the "maybe it gems" premium that raw buyers pay is gone forever.
The hard input is the probability vector, and that is where population reports earn their keep.
A worked example, fully hypothetical
Say a card books at $400 in PSA 10, $120 in PSA 9, and $60 in PSA 8 or lower. Say the raw copy sells for $90, the grading tier costs $25, and round-trip shipping adds $5. These numbers are invented for the arithmetic; do not trade on them.
Now estimate the grade odds. Suppose the population report for this card shows a distribution consistent with a 35% gem rate for well-selected copies, 50% landing at 9, and 15% at 8 or below. The expected payoff is:
| Outcome | Probability | Value | Contribution |
|---|---|---|---|
| PSA 10 | 35% | $400 | $140 |
| PSA 9 | 50% | $120 | $60 |
| PSA 8 or lower | 15% | $60 | $9 |
Probability-weighted value: $209. Subtract $30 in grading and shipping and the $90 raw value, and the expected value of the submission is +$89. That is a good bet. You will not make $89 on any single submission; you will make $310 sometimes, lose $0 to $60 other times, and average $89 if you could run the bet repeatedly.
Now change one number. Drop the gem rate to 15%, which is entirely realistic for a condition-sensitive card, and the weighted value falls to $129. After costs, the EV is +$9, which is noise. Drop the PSA 9 price to $80, which happens constantly on modern cards, and the EV goes negative. The bet did not change shape; the odds did, and the odds are the whole game.
Gem rates are not a constant
The single most common error in grading math is treating the gem rate as a hobby-wide number. It is not. It is a property of a specific card from a specific set in a specific era, and the variance is enormous.
Modern chrome and prizm-style cards come off clean production lines, get pulled into sleeves within seconds, and gem at high rates. Vintage cardboard was cut on looser tolerances, distributed in wax packs with gum, and stored in shoeboxes for decades; gem copies are condition rarities, which is precisely why vintage gems carry the premiums they do. In between sit the known problem children: sets with chronic centering issues, foil patterns that scratch in the pack, dark borders that show every touch.
The population report is your odds sheet. If a card has thousands of graded copies and only a thin slice are gem mint, the market has already run your experiment thousands of times. Take the Charizard ex from Pokemon Scarlet & Violet 151: total graded population above 100,000, with roughly 29,500 gem mint. That is a massive sample telling you the base rate before you adjust for your own copy's condition. We covered why population is the supply side of every card equation in what is card market cap, and the short version applies here too: the distribution, not the total, is the signal.
Your own copy's condition shifts you off the base rate, in either direction. But start from the base rate. Collectors who start from "my copy looks perfect" are the house's favorite customers.
The asymmetry nobody prices: the 9 trap
Here is the part of the bet that quietly kills modern submissions. On many modern cards, a PSA 9 sells for less than the raw card.
That sounds impossible until you think about what a raw buyer is paying for. A raw copy is a lottery ticket: it might be a 10. A PSA 9 is a resolved lottery ticket that lost. The grade extinguished the upside, and the market prices that resolution. So a modern card might trade at $90 raw, $400 in a 10, and $70 in a 9. Submit it, and the most likely single outcome makes you poorer than doing nothing, before you even count the fee.
Vintage cards mostly escape this trap, because a vintage 9 is itself a condition rarity and prices well above raw. The asymmetry is a modern phenomenon, born of high gem rates and huge raw supply. It means the EV of grading a modern card can be negative even when the 10 price looks spectacular, because the spectacular outcome is being subsidized by the likely one.
Turnaround time: the bet settles later than you place it
Grading is not an instant bet. Between submission and return, weeks or months pass, and the market does not wait.
You price the submission at today's grade values, but you settle it at the values prevailing when the slab comes back. If you submitted during a hype cycle, you are statistically likely to receive your card after the peak, because hype cycles are exactly when queues are longest. The price you ran the math on is gone, and so is everyone else's: hype submissions return in waves.
There is no clean hedge for this. The honest adjustment is to haircut your grade values by your own estimate of trend risk over the turnaround window, and to be most skeptical of submissions whose EV only works at a price that printed last week.
The dilution kicker: you are betting against your own bet
The subtlest cost of a submission is that it changes the market it is betting on. Every card you grade adds one to the population, and so does every card the thousands of collectors running the same math grade alongside you.
When a card's price spikes, raw copies flow into the grading queue, and months later the population report jumps. Supply arrives precisely when the math looked best, which is precisely when the math stops working. A card whose gem population doubles has effectively run a share issuance against everyone who already held a slab. This divergence between population growth and price is measurable, and it is exactly what the DRIFT dashboard tracks weekly across soccer, Pokemon, and MTG. For the full mechanics, read the DRIFT population divergence explainer.
For the EV formula, dilution means your grade values are not static inputs. If you can see a grading wave coming in the population data, mark your 10 price down before you submit, not after.
When not to grade
The math produces clear fold conditions. Do not submit when:
- The PSA 9 trades below raw and your gem odds are not exceptional. This is the standard state of low-value modern cards. The slab subtracts value. - The grading fee is a large fraction of the gem price. A card worth $60 in a 10 cannot mathematically carry a $25 fee plus shipping at realistic gem rates. - The EV only works at a spike price. If the bet requires last week's comp to hold through a three-month queue, you are betting on the trend, not the grade. - The population is visibly inflating. A pop report climbing fast is the market telling you your payoff table is stale. - You would not buy the card raw at today's price. Grading does not rescue a position you no longer believe in; it just adds fees and time.
None of this says the card is bad. It says the submission is bad, which is a different claim. Plenty of cards are best owned raw, and plenty of cards are best bought already slabbed, letting someone else eat the grading variance.
This is just asset math, applied honestly
The EV framing is not a trick for grading. It is the same discipline that turns a price into a market cap: enumerate the quantities, attach the probabilities and prices, and let the arithmetic say what the asset is. A card's market cap is its population times its prices, summed across grades; we walk through that in what is card market cap, and the grading decision is just the marginal version of the same equation, asked one slab at a time.
That is how GrailRank approaches all of it. The live market cap rankings recompute daily from confirmed sales and population reports, so you can see the payoff table for any card we track before you ship it anywhere. The weekly digest flags the population moves that change the math while your card sits in the queue.
To be explicit: this is measurement, not advice, and certainly not a system for winning. Grading is a bet with a computable expected value, and most submissions lose because nobody computed it. Compute it. Then decide.
Frequently Asked Questions
Is grading cards worth it?
Sometimes, and the only honest answer is the math. Grading is worth it when the probability-weighted value across all likely grade outcomes exceeds the raw card's value plus the grading fee and shipping. For high-value cards with strong gem odds, the answer is usually yes. For modern bulk where a PSA 9 sells near or below raw, the answer is usually no. Run the expected value calculation before every submission rather than relying on habit.
How do I calculate the expected value of grading a card?
Estimate the probability of each grade outcome, multiply each probability by the market price at that grade, and sum the results. Then subtract the grading fee, shipping, and the current raw value of the card. A positive result means the submission is profitable on average; a negative result means you are paying for the slab. Grade probabilities can be estimated from the card's existing population report distribution, adjusted for your own honest assessment of condition.
What percentage of cards get a PSA 10?
There is no single percentage, and any source that quotes one is misleading you. Gem rates vary heavily by era, set, and even print run within a set. Modern chrome cards from clean production lines gem at far higher rates than vintage cardboard that survived decades of handling, and condition-sensitive sets with known centering or surface issues sit somewhere in between. The right move is to check the population report distribution for the specific card you are considering, not a hobby-wide average.
Can grading a card lose money?
Yes, and it happens constantly. On many modern cards a PSA 9 sells for less than the raw asking price, because buyers of raw copies are paying for gem potential that a 9 has already extinguished. Add the grading fee and a months-long queue during which the market can fall, and a submission can turn a profitable raw position into a certified loss. Negative expected value submissions are the norm for low-value modern cards, not the exception.
Should I grade cards before or after prices spike?
The uncomfortable truth is that the queue decides for you. Cards submitted during a spike come back after the spike, because turnaround time means you are pricing the bet at today's market and settling it at a future one. Submissions made before hype, when fees are low and queues are short, capture spikes; submissions made during hype usually arrive into the correction, alongside everyone else's, which also grows the population at exactly the wrong moment.