Every grading ROI article you have read runs the same con: raw price versus PSA 10 price, as if your submission is guaranteed to gem. Here is the honest version. For every card with a PSA population above 500 in our recorded registry data, we know the real gem rate, the real 9 rate, and verified sold medians for both grades. That is enough to compute what a submission is actually worth before you mail it, with the failure cases priced in.
Net expected value per submission, $20 fee, registry-average outcomes
| Card | Raw | Gem rate | PSA 10 / PSA 9 | Net EV |
|---|---|---|---|---|
| Charizard, SWSH Black Star Promos | $581 | 55.5% (pop 8,977) | $3,027 / $526 | +$1,311 |
| Umbreon VMAX, Evolving Skies | $2,329 | 69.5% (pop 30,777) | $4,300 / $2,113 | +$1,271 |
| Giratina V, Lost Origin | $824 | 40.6% (pop 23,817) | $3,225 / $860 | +$948 |
| Rayquaza VMAX, Evolving Skies | $1,250 | 68.1% (pop 21,935) | $2,626 / $1,275 | +$905 |
| Blastoise & Piplup GX, Cosmic Eclipse | $494 | 32.6% (pop 4,857) | $3,050 / $508 | +$800 |
| Mega Charizard X ex, Phantasmal Flames | $673 | 59.8% (pop 49,604) | $2,001 / $758 | +$796 |
| Charizard VMAX, Champion's Path | $235 | 69.6% (pop 29,800) | $330 / $180 | +$31 |
| Hisuian Sneasler V, Astral Radiance | $38 | 49.7% (pop 2,238) | $103 / $30 | +$8 |
All nineteen are positive, and that should worry you
Every one of the nineteen cards we can fully measure shows positive expected value. Did we just prove grading always pays? No. We proved the registry is a graveyard of decided questions. A card only accumulates a 500-plus population because thousands of people already did this math and it worked; the cards where grading loses money never build a population big enough to enter our study. The list above is not a menu of guaranteed wins. It is the survivor pool, and the interesting information is the spread inside it.
That spread is wide. Giratina V breaks even if just 3.1% of submissions gem, against an actual gem rate of 40.6%; that is a thirteen-fold safety margin. Champion's Path Charizard VMAX needs 46% gems and gets 69.6%, a margin of 1.5x, thin enough that one bad centering batch erases it. Same verdict, completely different risk.
Your copy is not a registry average
The gem rate we use is the population's, built from every submission ever made, including the ones from people who grade anything that fits in a penny sleeve. If your copy is pack-fresh under a loupe, your true odds beat the table. If it has a fingernail line on the holo, they are far worse, and the 15% haircut we apply to non-9 outcomes will meet you personally. The table prices a strategy, not your card. The strategy's lesson is unambiguous though: the margin of safety lives in the break-even gem rate, so check that column before the EV column.
All 19 cards holding: a recorded PSA population of 500-plus with grade splits (Moonstone daily registry snapshots), current verified PSA 10 and PSA 9 sold medians with three-plus sales each, and a live TCGplayer base-printing market price from our daily sync, as of August 26, 2026. EV = gem_rate x PSA10_median + nine_rate x PSA9_median + remaining_rate x 85% of raw, minus raw cost and a $20 grading fee; the 85% factor prices sub-9 outcomes conservatively at a modest discount to raw. Net EV as shown; break-even gem rate solves the same equation for zero. All inputs are recorded market data; the EV arithmetic is ours and is disclosed in full. Not financial advice.