Buy the coin, not the holder
What 65,000 Morgan dollars actually show: where PCGS and NGC differ, where they surprisingly don't, and why two coins in identical holders may not be equivalent.
Every dealer has seen it
Two coins have the same date, the same mint, the same grading service and the same grade on the holder. Put them next to each other and one is clearly the coin you would rather own.
"Buy the coin, not the holder" may be one of the oldest pieces of advice in numismatics. But how often does it actually matter?
We recently analyzed 65,092 classic Morgan dollars using the computer-vision grading system developed for CoinAiLyzer. Without judging whether PCGS or NGC was "right," we used the same grading instrument as a common ruler across all of the coins. The results put some numbers behind things dealers have debated for decades.
Same grade on the slab does not always mean same quality
We looked at groups of coins with the same date, mint, grading service and stated grade. Within these groups the coins differed slightly in quality. All grades are bands, so this is expected.
But the quality difference was more than we'd expect just given each grading band. In the commercial Mint State range from MS-63 through MS-66, two randomly selected coins from the same date, mint, service and grade differed by at least one modeled Sheldon point roughly 13% to 27% of the time, depending on grade.
Some of the observed spread could come from photography and measurement variation instead of coin difference, but even taking that difference into account, there was more variability than expected.
We were also able to examine a limited set of repeat photographs of the same physical coins. The variation produced by re-imaging the same coin was substantially smaller, on average, than the differences between separate coins carrying the same grade—showing that measurement variation was not the whole story. In other words, the spread among same-grade coins is not simply an artifact of taking another photograph.
In short, it's not uncommon for a coin to actually be a grade less than or higher than the grade on the slab. We did find that it's extremely uncommon, from between 1 in 205 to 1 in 25, for two coins in one group to differ by 2 or more Sheldon grades. Meaning that while there is some variability it is not extreme.
A slab narrows the quality range considerably. It does not eliminate the value of choosing the right coin within that range.
Is PCGS tougher than NGC? Sometimes. But not across the board.
We also tested one of the most common claims in the Morgan market: that PCGS simply grades tougher than NGC. The data do not support such a simple rule.
The biggest and most repeatable differences occurred around the upper-AU and low-Mint-State boundary. In our primary population, PCGS AU-55 Morgans scored about 0.73 Sheldon points higher on average than NGC AU-55 coins carrying the same stated grade. At AU-58, the difference was about 0.79 points.
By MS-62, the difference essentially disappeared. From MS-62 through MS-66, PCGS and NGC populations were remarkably close, almost indistinguishable, under the common grading ruler.
That means the statement "PCGS Morgans are better for the grade than NGC Morgans" is much too broad. A separate held-out group of another 27,272 Morgan dollars broadly confirmed the same pattern. It also found substantial separation near the upper-AU/low-MS boundary, followed by convergence beginning around MS-62.
The relationship between PCGS and NGC changes depending on where you are on the grading scale.
The AU-58/MS-60 area deserves attention
Why would the largest service differences cluster around AU-58 and low Mint State? We don't know yet. There are several plausible explanations.
The services could apply somewhat different standards at this difficult boundary. Dealers could choose different coins to submit to each service. Strong NGC coins could migrate into PCGS holders through crossovers or crackouts. Historical changes in grading could contribute. More than one of these could be happening simultaneously.
My hunch, not fully backed by the evidence, is that this is not a difference in standards, but is a difference in who sends what where. Our study measures the coins we have in slabs today; we don't know how those coins got where they are.
This poses some interesting ideas on possible profitable crossover strategies. The study does not prove that a particular crossover strategy is profitable. But we have it tagged for a future study.
In commercial Mint State, the individual coin matters more than many dealers might expect
Does an MS-65 Morgan mean substantially different things depending on the date and mint?
Morgan collectors know that some issues are famously weakly struck, unusually attractive or difficult to find with clean surfaces. So it would be reasonable to expect the visual center of an MS-65 population to move considerably from one issue to another.
Among the well-populated issues we could test, however, those average shifts were fairly modest. So modest that I feel the data does not support the idea that different dates and mints are graded differently. At MS-65, 38 date/mint issues had enough PCGS and NGC coins for comparison. The standard deviation of the issue averages was only 0.16 Sheldon points for both services.
Compare that with the variation among individual coins inside the exact same date, mint, service and grade group, which was roughly 0.5 to 0.7 points through much of commercial Mint State.
Put more simply: MS-65s within a grade/year/mint vary much more than the average MS-65 of different grade/year/mints.
That doesn't mean date and mint don't matter. Rare dates frequently did not have enough coins to qualify for this analysis, and we did not independently measure strike quality. But for the well-populated commercial Mint State issues we examined, the data did not show wildly different visual standards from issue to issue.
The much larger variation was between individual coins. Again: buy the coin.
What should a dealer take from this?
- Don't treat same-grade slabs as completely interchangeable inventory. The holder tells you a great deal. But even when date, mint, service and grade are identical, there can be meaningful quality differences between individual coins. Good coin selection still matters.
- Don't use one simple PCGS-versus-NGC adjustment across the Morgan grading scale. The relationship between the services changes depending on where you are on the scale.
- Pay particular attention around the AU-58/MS-60 boundary. That is where our two large populations consistently showed the greatest service separation. We don't yet know exactly why, but it is clearly a different part of the market from MS-63 through MS-66.
- Don't assume every visible difference between two certified coins means somebody made a grading mistake. Professional grading places coins into useful quality bands. Those bands still contain variation. That is exactly why dealers and experienced collectors continue to care about eye appeal, surfaces, luster, strike and the overall look of the individual coin.
The data don't undermine professional grading. If anything, they help show what the grade does and does not tell you.
How we tested it
The main analysis included 65,092 classic Morgan dollars dated 1878 through 1904 and 1921, all sourced from dealer imagery. Each coin was evaluated using the same computer-vision grading system underlying CoinAiLyzer. The system was used as a common measuring instrument, not as an assertion of the coin's objectively correct grade.
The grading model was not told whether a coin was in a PCGS or NGC holder, and the surrounding holder, labels and background were removed from the image before grading. The analysis was also structured so that the model evaluating a particular physical coin had not been trained on that same coin.
We then ran numerous checks involving coin sides, dealer sources, date and mint composition, an earlier generation of the grading system, and a separate held-out population of 27,272 additional Morgan dollars.
The full research paper contains the statistical methodology, limitations and supporting analyses for anyone who wants to dig into them. The dealer takeaway is considerably simpler:
The holder grade is valuable information. But it isn't the whole coin.