A Quantitative Look at Third Party Coin Grading

[cathrine 70]

I’m primarily a collector of Victorian copper and bronze. All of my high grade coins are currently graded by either NGC, PCGS, or CGS.

I’ve read a number of articles appearing in various numismatic publications or online sources about how a certain grading company applies more or less strict standards compared with another specific grading company. As a collector of graded coins, I have my own notions of which grader applies higher or lower standards when assigning a grade. However, I haven’t seen any quantitative treatment of grading results, which are published by all of the above mentioned organizations on their websites.

To make a long story shorter, I decided to take a stab at comparing some results of two grading companies. I selected NGC and PCGS for a comparative study because both:

(1) Use a common numerical grading scale of 0 to 70,

(2) Apply some degree of judgment in assessing the most predominant color of copper and bronze coins, and

(3) Have a sufficiently large number of coins in their population census to make valid statistical comparisons.

Since I’m primarily a halfpenny collector, I decided to test my opinions about NGC and PCGS on Victorian copper halfpennies. My comparison study was restricted to the uncirculated grades, which, for copper halfpennies spans a range of grades for both services from MS61 to MS66. The three coin colors are treated both individually and collectively.

The data was collected from both websites on January 21. The NGC data set consists of 401 coins (283 brown, 115 red & brown, and 3 red), while the PCGS set contains 190 coins (112 brown, 72 red & brown, and 6 red).

Color Results

Since the coins submitted to both companies came from the same population of available coins, it’s easy to see that PCGS is more likely than NGC to consider a submitted coin as red & brown or red rather than brown and is also more likely to consider a coin to be red rather than red & brown. Looking at the percentages:

NGC: 70.6 % Brown - 29.4 % red & brown or red

PCGS: 59.0 % Brown - 41.0 % red & brown or red

NGC: 29.4 % red & brown or red - 0.8 % red (i.e.,1 coin in 37 is red)

PCGS: 41.0 % red & brown or red - 3.2 % red (i.e.,1 coin in 13 is red)

　

　

Grading Results

Looking at both distributions of coin grades for all coins in the two sets:

For NGC:

61 62 63 64 65 66

6 69 110 136 67 13

1.5% 17.2% 27.4% 33.9% 16.7% 3.2%

For PCGS:

61 62 63 64 65 66

1 20 63 76 27 3

0.5% 10.5% 33.2% 40.0% 14.2% 1.6%

The two distributions have remarkably close arithmetic mean values: NGC: 63.57 & PCGS: 63.62. However, the distribution of grades for NGC is slightly more spread out from the mean value than the PCGS distribution. The most likely grade for either grading service is a 64.

Conclusions

According to the results of this study for Victorian copper halfpennies, the two grading services are likely to give identical grading results; however, PCGS is more likely than NGC to give a more favorable color assessment.

[Prax 80]

Very useful. Where are you from Cath? Reason for my question is to find out how you managed to escape the export - re-ipmort fees payable to HMRC.

I am told FedEx can mark it a shipment for export and reimport but this is more of a hit and miss

[Rob 3,221]

That's quite an interesting comparison - thanks.

Is it possible to do a similar study for those coins submitted by named individuals as opposed to general submissions? I have a gut feeling that they tend to be more lenient with their grades if it is an 'important' collection, so for example, I have a couple of George III pattern halfpennies that were both previously in NGC slabs, one was in a 61 slab and the other a 65, but the 61 (un-named) is markedly better than the 65 (ex-Cheshire). There are other examples.

Another useful study would be the consistency of grading for the same coin, if possible. I had an ex 64 coin which I sold to someone on the PCGS forum and it came back 62 the second time round, even though it was from the same service. I know there are a couple instances where the coin has been submitted, cracked out, resubmitted, cracked out, resubmitted etc and the grading has varied by at least two grades. This is actually quite important as we all know the way slabbing afficionados think, particularly those of the registry set mentality. Buy a coin you think is undergraded and resubmit in the hope of it coming back higher. Obviously overgraded coins do not get cracked out and this has led to grade inflation with 65s commonplace today relative to say 10 years or more ago. The Peck 924 in the recent Heritage sale (lot 29616) was in a 65 slab, despite having sat quite happily in a 63 for quite a long time. This of course led to 65 pricing rather than 63 despite not being a patch on the raw P924 sold by CNG a couple of years ago. It is now unlikely to be cracked out unless bought by a person who doesn't like slabbed coins, but can only find this example to buy (and there is only one other possible candidate as far as I am aware). Selective resubmission is an obvious reason for grade creep.

I'm not looking to open the perennial can of worms that considers the merits of slabbing or not as this will doubtless be resurrected on a regular basis, but given the number of resubmissions and the fact that these coins are not normally identified in the population reports, it is an important consideration in the data available. I also collected halfpennies for a long time as a denomination and can vouch for the unreliability of populations having personally cracked out a considerable percentage of certain rarer pieces.

[cathrine 70]

I live on Hilton Head Island, South Carolina, USA., so I don't have any experience with HMRC import fees.

If a "statistically significant" amount of valid census data is available, a similar assessment can be made for grading and color. However, it must be kept in mind that statistical results do not "prove" the hypotheses drawn from them; they merely indicate, more or less strongly, such conclusions. Grading bias for a particular subset of coins withdrawn from the general population can be detected. However, the cause of the bias cannot be established. Was the bias induced by the grading skills of the submitter? Is the bias a product of a relationship between the submitter and the grading service? Is the bias caused by a different group of graders being employed by the grading service? Conclusions to such questions cannot be established from the type of data being analyzed in my study.

[azda 982]

  On 1/23/2015 at 6:45 AM, Prax said:

Very useful. Where are you from Cath? Reason for my question is to find out how you managed to escape the export - re-ipmort fees payable to HMRC.

I am told FedEx can mark it a shipment for export and reimport but this is more of a hit and miss

There i's an NGC office here in Munich, they take care of all the handling and paperwork when coins are sent to the US of A for grading. I assume that they can state that the coin is already pre owned and no tax should be owing when sending out To NGC. The flipside of the coin is that you can also send a coin to PCGS in Paris but their prices are a little steep for my taste. Around 45 a coin more than 50 years old i believe.

I could join NGC at their top level service for $275 And get 5 free gradings thereafter. $275 = 244 Euros but you have to pay courier service There and back with PCGS as they won't accept normal post, so You could in theory add another 30 euros on top with PCGS

Edited January 23, 2015 by azda

[Chingford 166]

I noticed at the latest Goldberg sale you can now get 4th Party Grading, WINGS as they are known will give you a little gold sticker and tell you if you coin has been correctly 3rd party graded for a small fee.

Edited January 23, 2015 by Chingford

[azda 982]

  On 1/23/2015 at 7:51 PM, Chingford said:

I noticed at the latest Goldberg sale you can now get 4th Party Grading, WINGS as they are known will give you a little gold sticker and tell you if you coin has been correctly 3rd party graded for a small fee.

And for another £50 i'll tell them IF the 4th and 3rd party graders have got it correct

Where does this shite stop lmfao

[Peter 615]

It's all a bit WITH WINGS.

[TomGoodheart 434]

I suppose my only concern would be that this assumes similar coins are submitted to both companies. But considering that on a US forum you quite often see people ask which TPGS is 'best' for a particular type of coin, it's possible that those submitting have a preference.

And if some of those submitters that have more expertise in British halfpennies chose one company disproportionately over another then the sample will not be equivalent...

As Rob says, really you need the same coin submitted to both companies. But as he also says, the same coin resubmitted will not always achieve the same grade so ... um, I'm not sure what I was trying to say now.

LOL

[VickySilver 883]

I agree with Rob that the consignor is all important when it comes to grade, even and despite the protestations of blind grading. One really interesting example is when it comes to matte proof grading, both the
"Standard year" 1902 and the much rarer mattes of years like 1926, 1927, 1937, 1951 and 1953.

The '02s were nearly all rubbed by mint workers with some sort of cloth evidently and show hairlines to smaller or greater extent. How much is deducted for these varies tremendously by PCGS and NGC in my opinion. I have also seen in the later much scarcer bits some ridiculous grading, both high and low.

Another example is with 19th C proof gold and silver where the buyer needs to really look at the coins - I have seen lowly 62s clearly better than 64s (see for example, 1853 proof florins and half crowns, or even the 1862 & 1864 proof half crowns).

Caveat emptor!

[cathrine 70]

I've repeated the preceding process for the NGC and PCGS graded Victorian young head bronze halfpennies. Again, the uncirculated census for both grading services ranges from MS61 to MS66.

The data was also collected from both service's websites on January 21. The NGC data set consists of 606 coins (173 brown, 371 red & brown, and 62 red), while the PCGS set contains 293 coins (59 brown, 179 red & brown, and 55 red).

Color Results

The results show that PCGS is again more likely than NGC to consider a submitted coin as red & brown or red rather than brown and is also more likely to consider a coin to be red rather than red & brown. Looking at the percentages:

NGC: 28.6 % Brown - 71.4 % red & brown or red (7 coins in 10 are red & brown or red)

PCGS: 20.1 % Brown - 79.9 % red & brown or red (8 coins in 10 are red & brown or red)

NGC: 71.4 % red & brown or red - 10.2 % red (1 coin in 7 is red)

PCGS: 79.9 % red & brown or red - 18.8 % red (1 coin in 4 is red)

　

　

Grading Results

Looking at the distribution of coin grades for all coins in the two sets:

For NGC:

61 62 63 64 65 66

9 60 137 220 160 20

1.5% 9.9% 22.6% 36.3% 26.4% 3.3%

For PCGS:

61 62 63 64 65 66

0 20 63 149 54 7

0.0% 6.8% 21.5% 50.9% 18.4% 2.4%

Both distributions have remarkably close arithmetic mean values: NGC: 63.86 & PCGS: 63.88. Again, the distribution of grades for NGC is more spread out from the mean value than the PCGS distribution. The most likely grade for either grading service is a 64.

Conclusions

According to the results of this study for Victorian young head bronze halfpennies, the two grading services are likely to give identical grading results; however, PCGS is more likely than NGC to give a more favorable color assessment. This is the same conclusion previously reached for the Victorian copper halfpennies.