How the Grading Economics Math Works
Expected value, fees, net profit, and the GRADE / WATCH / SKIP verdict — the full math behind every grading decision on Cardboard Assets. Worked example on Charizard 1st Edition Holographic.
Every "should I grade this card?" question has the same shape. You pay to grade the card. With some probability it comes back at the gem grade and sells for a premium. With some probability it grades lower and sells for less than that. After fees, you either made money or lost it. Expected value (EV) is the mathematically correct rank-order for which cards to grade. Net profit is EV minus all the costs. This page walks the math top-to-bottom with a real example.
The EV formula
Cardboard Assets bucket per-card outcomes into three grade tiers:
- Top grade — PSA 10, BGS 9.5 / 10 / Pristine 10, CGC 10 / Pristine 10
- Grade 9 — PSA 9, BGS 9, CGC 9
- Below 9 — anything lower, treated as a single bucket because the slab premium below 9 is small relative to the gem premium
The expected-value formula:
EV = P(top) × comp_top + P(9) × comp_9 + P(below_9) × comp_below_9
Probabilities sum to 1.0. The platform pulls per-grader gem rate from public pop reports (when available) and the per-grade comp prices from hand-verified eBay sold listings (when available) or the hybrid comp pipeline. Both sources are labeled per card in the optimizer's freshness chip.
Fees + costs
Once EV is computed, subtract every cost between acquiring the raw card and depositing the sale proceeds:
| Cost line | Where it comes from | Typical value |
|---|---|---|
| Raw card cost | Your acquisition price (defaults to current raw market price) | Card-dependent |
| Grading fee | graders.json — per-grader, per-tier | $10–$1,000+ depending on tier |
| Return shipping | Platform default: $15. Real cost depends on insured value | $5–$50 |
| eBay final-value fee | 13.25% of sale price + $0.30 per transaction (toggle in optimizer header) | ~13.5% of sale |
Not included by default: state sales tax, eBay seller fees beyond final-value, PayPal/Wise transfer fees, your time. The platform's EV math is intentionally optimistic on costs — if the optimizer says SKIP, real-world friction makes it more SKIP not less.
Net profit + ROI
Net profit:
net = EV − raw − grading_fee − shipping − ebay_fee
where ebay_fee = EV × 0.1325 + 0.30 when the eBay toggle is on, else 0.
ROI:
ROI = net / raw
Annualized return (when displayed):
annualized = ROI × (365 / turnaround_days)
The platform always labels this as simple annualized because grading is a one-shot trade. Compounding would assume you take the proceeds and immediately re-grade another card; that's a portfolio strategy, not an individual flip.
GRADE / WATCH / SKIP verdict
Net profit alone is too coarse — a $50 net on a $20 raw card is a 250% ROI flip; a $50 net on a $5,000 raw card is a 1% ROI flip. The verdict combines net profit and play score (covered in play-score methodology):
- GRADE — net profit ≥ $200 or play score ≥ 80
- WATCH — net profit ≥ $50 or play score ≥ 60 (and not GRADE)
- SKIP — everything else with optimizer data
- INSUFFICIENT — comp data missing for this card; verdict deferred
The thresholds are documented in CLAUDE.md and the verdict-computation
code. They're not opinions — they're conservative defaults that match how flippers
typically rule out marginal flips (sub-$50 net usually isn't worth the time + variance).
Worked example: Charizard 1st Edition Holographic
The canonical demo card on the platform. Run the same math the optimizer runs:
| Input | Value | Source |
|---|---|---|
| Raw market price | $5,500 | Hand-verified vintage WOTC market comp |
| PSA 10 comp price | $850,000 | SCI auction comp + verified eBay sold listings |
| PSA 9 comp price | $81,000 | Recent eBay sold |
| PSA <9 comp price | $5,500 | Equivalent to raw on partial grades |
| PSA 10 gem rate | 2.35% | GemRate (125 PSA 10s / 5,317 graded) |
| PSA 9 rate | 13.7% | GemRate |
| Below 9 rate | 83.9% | GemRate |
| PSA grading fee | $9,999 | PSA Premium 10 tier ($1M declared value) |
| Shipping | $15 | Insured return shipping |
| eBay FVF | 13.25% + $0.30 | eBay trading card category |
EV calculation:
EV = 0.0235 × $850,000 + 0.137 × $81,000 + 0.839 × $5,500
EV = $19,975 + $11,097 + $4,615 = $35,687
Net after fees:
Net = $35,687 − $5,500 − $9,999 − $15 − ($35,687 × 0.1325 + $0.30)
Net = $35,687 − $20,243 = $15,444
The platform's current snapshot shows the slightly more favorable
+$18,261 headline because it uses fresh GemRate probabilities + the
full-precision comp prices. Run the live math on the
Charizard 1st Ed optimizer page — same formula, current numbers.
For context, if your raw Charizard happens to land a PSA 10, the deterministic best outcome is:
$850,000 − $5,500 − $9,999 − $15 − ($850,000 × 0.1325 + $0.30) = $721,861
A $722K upside on a 2.35% probability is what makes this card the platform's headline GRADE verdict. Most cards don't have this risk-reward shape.
What the math doesn't include
Honest about its limits:
- Variance. The EV is the average outcome across many submissions. Any single submission can be far above or below. Don't grade a card you can't afford to have come back at the median outcome.
- Crack-and-resub. Some flippers grade a card, send the 9 back for review, get a 10. Real-world gem rates effectively rise after multiple submissions. The platform uses single-submission rates because that's the honest baseline.
- Market timing. Comp prices change. A card flagged GRADE today might be WATCH in three months if the PSA 10 comp drops 30%. The optimizer reads from the daily comp pipeline; pages refresh nightly.
- Personal preference. Some collectors get non-monetary value from the opening experience or from owning the raw card. The math doesn't price intangibles.
- Tax. Capital gains on a flip is taxable income. Federal short-term cap gains are at your marginal rate. The math is pre-tax.
Frequently asked questions
- What is 'expected value' in grading economics?
- Expected value (EV) is the probability-weighted average sale price across possible grade outcomes. EV = P(grade 10) × comp_10 + P(grade 9) × comp_9 + P(below 9) × comp_below_9. It's the mathematically correct answer to 'how much would I average if I graded this card many times.' Any individual flip might be far above or below.
- What is the GRADE / WATCH / SKIP verdict?
- GRADE = net profit ≥ $200 OR play score ≥ 80 — the math clearly favors grading. WATCH = net profit ≥ $50 OR play score ≥ 60 — marginal; depends on your specific copy or market timing. SKIP = everything else — sealed beats grading at these comps and fees. The thresholds are documented in CLAUDE.md and the play-score methodology page.
- What fees does the optimizer subtract?
- Raw card cost (what you paid to acquire), grading fee (per-grader, per-tier from graders.json), return shipping ($15 default), and eBay sales fee (13.25% of sale price + $0.30 transaction). All fees are toggleable. The 'Selling on eBay' toggle in the optimizer header swaps the eBay fee in/out.
- How accurate are the comp prices?
- Per-grade comp prices come from one of: hand-verified eBay sold listings (manual_verified — highest trust), eBay disappearance-detected inferred sales (inferred_sold_hybrid — the hybrid comp pipeline writes these nightly with a confidence score), or manual seed values (manual_seed — labeled BETA verifying). Every card surfaces its comp source in the freshness chip.
- Why is annualized return labeled 'simple' not 'compound'?
- A grading flip is a single-cycle trade — you grade once, sell once, done. Simple annualization (ROI × 365 / turnaround_days) is the correct framing because there's no compounding. Calling it CAGR would imply reinvestment which doesn't happen on individual grading flips. CAGR would apply to a portfolio strategy of repeatedly grading and reinvesting.