What does PCGS use to determine the weights assigned to coins in registry sets? Does the process ne

DMWJR

PCGS has managed to boil weights down to one rather small number for each coin in each series. We used to have some raging debates on this forum between PCGS's system and NGC's weighting system. Most people felt like NGC's system is better able to weight a specific coin for true scarcity and census condition and grade between coin of the same date/mm within the same series.

I'm curious to see what current board members think, and whether people think the weighting system should be changed. I am enjoying the new changes to the registry set and it is beyond what I expected. I can only imagine that PCGS would surprise me again if they decided to change the weighting system next.

Is it one, two or a combination of all of the factors below?

true date/mm scarcity?

low vs. high mintage?

condition rarity?

collector desirability?

traditional key or semi-key vs. common date?

leothelyon

Most people felt like NGC's system is better able to weight a specific coin for true scarcity and census condition and grade between coin of the same date/mm within the same series.

WRONG! NGC SUCKS!

Leo

LincolnsRule

Pcgs's weighting on moderns is all wrong! They give the highest weights to what is the rarest in plain old BU, not what is the rarest in high grade. As much as pcgs would like people to collect pcgs graded ms60 moderns, it's not going to happen.

TomB

I believe Leo specializes in nickels, though I could be wrong, and perhaps in that arena he may have a point. However, in the US Type Set arena, NGC has a much better paradigm than PCGS. To be fair, type sets may likely be the most difficult sets to score given their nature, but the current scoring system at PCGS is wildly off the mark. Additionally, I understand that the goal of both the PCGS and NGC registries is to encourage broader collecting of slabbed coins and that type sets may force one to purchase coins they might otherwise not acquire.

The only competitive sets that I have registered are complete copper, nickel and silver type sets and, as written above, NGC is far more accurate with their weighting scheme. An example would be the relative weights given to scarce/rare types vs weights given to profoundly common types. Using the NGC system, if one were to register a 1796 quarter in F12 they would receive 4472 points. Moreover, if one were to instead register a pair of coins that might be viewed as the epitome of common for their types, a 1938-D Buffalo nickel in MS65 and a 1964 Roosevelt dime in MS65, then one would gain 164 points (155 points for the Buffalo + 9 points for the Roosevelt). The result is 27.3 times more points for the exceptionally scarce and expensive 1796 quarter in F12 (a $25,000 coin) than for the pair of incredibly common and inexpensive 1938-D Buffalo in MS65 and 1964 Roosevelt in MS65 (likely less than $50 for the pair, combined).

In contrast, if one were to register this same 1796 quarter in F12 on the PCGS registry one would gain 0.24 points in the Set Rating, which is how sets are ranked relative to one another. If this same set were to have registered instead the 1938-D Buffalo in MS65 and 1964 Roosevelt in MS65, as mentioned above, then the set rating would gain 0.29 points. That is, the addition of the $50 Buffalo nickel/Roosevelt dime paring would improve the set 1.2 times as much as the addition of the $25,000 1796 quarter. This makes no sense at all aside from trying to influence participants into purchasing product they might not otherwise obtain in an attempt to creep up in the rankings.

Perhaps other series and sets are different, though I doubt any are more accurate than the NGC model, but most reasonable folks would be able to immediately determine that there is an obvious extreme bias, to the extent of a flaw, in the weighting paradigm of type sets.

Steve

Doug,
I believe that Tom is probably right that NGC's registry more accurately reflects value. For fun I took our two 1909VDB Matte proofs. In NGC my coin (61RB) is worth 990 points. Your coin (66RD) is worth 5,922 points, just about 6 to 1 difference. In PCGS my coin gets 620 points (61+1 bonus for RB times 10 weighting) Your coin gets 690 points (66+3 bonus for RD times 10 weighting). Now the rest of the PCGS process may increase the spread, but just on this one coin the difference appears to be only about 11%. As to value my coin, according to PCGS is worth $2,500. Your coin is worth $35,000, a 14 to 1 difference.

I don't know what all this means, but you might just want to register your set over there. I did it for fun a couple of years ago and I'm still holding at #5. Their registry does not count the 1990 no S or the 1960 and 1971 doubled die error coins.
Steve

DMWJR

I haven't studied the NGC system, because I don't really have much interest in it.

Where I think the PCGS weighting system could be improved is by increasing the distance between the most common coins in a series and the most rare. I have looked at quite a few sets to check composition, and it seems that PCGS uses a 10 point scale with 10 increments for within any given series. I'm not saying all series have a coin weighted by 10, but no series I looked at has a coin over 10. Even the Barber Dime w/1894-s set shows that the 1894-s has a weight of 10, with a few other coins having 6, 7, or 8. To me, a person on the outside of the Barber dime series, there is a far bigger difference in weight between a 94-s and any other coin in the series -- not just 2 or 3 points.

The task becomes much harder when you mix proofs and mint state within a given set, and even harder for a type set. If you had a type set of common date coins in MS65, and someone else had a type set of key date coins in MS63, which would be the more important set? Shouldn't the more important set be ranked higher?

NGC's system appears to offer a broader set of weights to account for the bigger differences in importance of coins, even within the same date/mm.

Wolf359

Well vams weights need an overhaul. The proper weights for ultra rare 1882-O/S EDS coins and 1879-O VAM-28 are not reflected in the current sets, among numerous examples.

mrcommem

mrcommem

The big problem with PCGS is the weighting within the type sets. PCGS makes no distinction to the rarity of a coin within a series. For example in a gold type set you need a $3 gold piece. The same weight is given for a 1854-D three and a 1854 three. The 54-D in extremely rare with only 1100 minted while the 54 in the hundreds of thousands. What PCGS needs to do is to multiply the weighting factor used inside (for instance the $ gold pieces) with the weighting used for a $3 gold in the type set. This would add one more factor for each coin type in a type set. If I remember right the weight of a $3 1854-D is 8 within the three dollar series and multiply that weight by the weight in the type set for a three dollar gold then multiply by the grade of the coin. For instance if you have a weight of 5 for a three dollar gold in the type set the two coins would weight as follow:

1854-D = 8 X 5 XAU55=2200
1854 = 1 x 5 x AU55=275

This would give some parity to the rarity vs commonality of type coins. This only needs to be done to type sets.