Howdy, Stranger!
It looks like you're new here. If you want to get involved, click one of these buttons!
Quick Links
Thoughts On Building An Accurate CAC Registry Weighting System.
I'm convinced that the accuracy of the current weighting systems used by the TPGS can be significantly improved making the CAC registry more accurate and fair.
Some background for those who are not familiar with the current weighting system used by PCGS.
Weights are currently defined as:
Within a set, not all coins have the same value and rarity. Sets in the PCGS Set Registry are weighted on a 1 to 10 scale with 10 being the rarest coin and 1 being the most common. While it may be true in some instances that a coin may be many times over 10 as rare as the most common coin in the set, the Registry keeps the scale simple so that all levels of collectors can compete.
The Registry weighting is done by taking three things into account:
• The overall rarity of the coin, i.e. the rarity in all grades
• The rarity in the highest 2 or 3 grades
• The price (because this is an indication of demand and importance to collectors)
This definition would be more accurate if it was modified to:
Within a set, not all coins have the same value and rarity. Sets in the PCGS Set Registry are weighted on a 1 to 10 scale with 10 being the rarest coin and 1 being the most common. While it is true in many instances that a coin may be many times over 10 as rare as the most common coin in the set, the Registry keeps the scale simple so that all levels of collectors can compete.
This simple method sacrifices accuracy by discounting the rarity of coins in an attempt to level the playing field. Unfortunately, this does nothing to help collectors compete with top sets. In this case attempting to level the playing field simply introduces errors by using inaccurate weights. In addition, this definition does not specify if the three criteria used to determine weighting are given equal weight.
For example: For some series like Liberty Seated Dollars, criteria one and two conflict with each other creating inaccuracies. Some dates require different weights for circulated and uncirculated examples. Lumping them together is not an option if you want to achieve any degree of accuracy.
The algorithm that NGC uses to calculate weights is more accurate, there is still room for improvement.
I realize that assigning weights is subjective. However, consulting with experts in each series and averaging their data and or building an algorithm should produce an accurate weighting system.
Some background for those who are not familiar with the current weighting system used by PCGS.
Weights are currently defined as:
Within a set, not all coins have the same value and rarity. Sets in the PCGS Set Registry are weighted on a 1 to 10 scale with 10 being the rarest coin and 1 being the most common. While it may be true in some instances that a coin may be many times over 10 as rare as the most common coin in the set, the Registry keeps the scale simple so that all levels of collectors can compete.
The Registry weighting is done by taking three things into account:
• The overall rarity of the coin, i.e. the rarity in all grades
• The rarity in the highest 2 or 3 grades
• The price (because this is an indication of demand and importance to collectors)
This definition would be more accurate if it was modified to:
Within a set, not all coins have the same value and rarity. Sets in the PCGS Set Registry are weighted on a 1 to 10 scale with 10 being the rarest coin and 1 being the most common. While it is true in many instances that a coin may be many times over 10 as rare as the most common coin in the set, the Registry keeps the scale simple so that all levels of collectors can compete.
This simple method sacrifices accuracy by discounting the rarity of coins in an attempt to level the playing field. Unfortunately, this does nothing to help collectors compete with top sets. In this case attempting to level the playing field simply introduces errors by using inaccurate weights. In addition, this definition does not specify if the three criteria used to determine weighting are given equal weight.
For example: For some series like Liberty Seated Dollars, criteria one and two conflict with each other creating inaccuracies. Some dates require different weights for circulated and uncirculated examples. Lumping them together is not an option if you want to achieve any degree of accuracy.
The algorithm that NGC uses to calculate weights is more accurate, there is still room for improvement.
I realize that assigning weights is subjective. However, consulting with experts in each series and averaging their data and or building an algorithm should produce an accurate weighting system.
Comments
For example, for myType Sets, I’ve chosen to always go for common dates, which allows me to afford higher grades, which is all that counts in the PCGS Registry forType sets. True, my higher grade can help a bit in the new CAC weighted system, but probably not much. So although I’ll get hurt a bit on the ranking of my various Type Sets, that’s OK, since it’s a fairer system.
With series, where everyone collects the same dates, I believe the weighting system will properly reward those with higher grades, where getting a plus or a higher whole grade will be rewarded more than the current PCGS system of just a half or whole point multiplied by the current date weight.
A fairer system for all.
Steve
1) keep it simple. A simple GPA with subtle weights are best. If it were up to me I'd weight keys 3 and semi-keys 2 and everything else 1 (or even a 5,4,3 weighting if one wants it to be more subtle).
2) What's a key/semi-key should be based on tradition tweaked a bit by modern knowledge of populations. For example in Lincoln Cents 09-s vdb, 14-d, and 26-s would be a 3 but my choices of semi keys would be something along the lines of 09-s, 14-s, 24-d, some of if not most of the other S's from 1916-1927, 1931-s, and 1954.
3) No bonuses for top pops or pop 1s.
4) Separate sets with no bonuses for designations (RD, FS, FBL etc) and sets with bonuses for such would be nice. Weightings could be different. For example in the non-FS Jefferson nickel set a 53-s would be a 1 but in the FS one it would be 3. Since weightings are different it may not make sense to also have bonuses different. I'd like to see just 1 point across the board for a FS, FBL etc bonus.
5) Very important is the choice of what goes in a basic set vs a with major varieties set. The worst thing pcgs ever did (in terms of registry modifications) was remove the 76 Ike types from the Ike Basic set. These are a pretty drastic obvious design difference, more than a mere variety. I don't know who talked them into doing that. To me it's like not having a 1909 vdb and plain in the basic Lincoln set. My feelings are if the mint intentionally changed the design of the coin then it should be a basic set. My basic sets would (and do in my own collection) include 1960, 70-s, 82 Lincoln sd/ld/brz/zinc vars, 1939 nickel rev types, 1934 quarter motto types, 1976 ike types, 1979 wr/nr SBA (that's my WR picture!).
6) Another thing pcgs did that annoyed a lot of people was suddenly added the 1958 DDO Lincoln to the major varieties set. While it's major in that the doubling is obvious or that it's an important and rare coin, it's not major to the 100s of collectors that can't ever come close to affording it. Aesthetically it'd have been nice to not have it be required, and for all the other collectors to show set completion at 100% (and a bonus still could be given for it). Just some common sense could go a long way.
However, for LSD's and TD's it's clear that the algorithm used by NGC for weighting is superior to the current weighting system used by PCGS. Despite providing detailed data to PCGS they have refused to make any modifications to their current system. Their response continues to be perhaps in the future.
The most important goal of an accurate and fair weighting system should be representing the rarity of each date in a set. This can be determined by combining data from population reports and consulting with experts in specific series. For example: John Albanese, Doug Winter, Sheridan Downey, Rick Snow, etc. There are other factors that could be considered but I strongly disagree with your statement "What's a key/semi-key should be based on tradition tweaked a bit by modern knowledge of populations.
Simple is good but adopting subtle weights of 1, 2, and 3, will not allow for accurate weighting of common versus very rare dates.
I agree with your point regard the 1958 DDO penny. Although the 1870-S Seated dollar is not a variety it is required by PCGS to reach 100% completion. NGC does not require it. PCGS estimates a mintage of 15, with only 12 accounted for, and only 3 examples approved by CAC. Therefore, for the CAC registry only 3 people could ever achieve 100% completion if it is included.
The situation I want to avoid is a pop 1 ms68 45-d Lincoln accounting for some huge score for being a pop 1 while a 25-s in 65rd is some lower score since it's a pop 13 (i think) and a 14-d in 65rd even lower as it's a pop 40 or more (if I remember correctly). I feel like 09-s vdb and 14-d still should have that edge for coolness factor, and it evens out the scoring a bit so those with the money for pop 1s can't just easily run away with it.
Heck, if you want one extreme why not just add the price guide value of each set and rank it in order by whose is the most expensive.
"There has already been a reckoning of sorts [regarding high prices for some modern coins in very high grades]. Many of their prices have dropped, some considerably. Submissions and populations have increased for those coins. I believe much of the overvaluation has been due to a misguided registry score. We'll correct that with our registry score, at least for CAC coins. I'll reference the 1932-D quarter, which as kids we knew as the King of Washington Quarters. PCGS and NGC have graded 119 pieces in MS 65. The PCGS value is $11,000. Now, let's look at a 1962-D quarter. PCGS and NGC show 10 graded in MS67, no MS68s. Let's say a single MS68 was made today, CAC or not. My guess is that at auction it could bring $30,000, which is about triple the value of a MS65 1932-D. We will have a different scoring formula which uses an average grade price.
Now clearly, the 1962-D in MS68 is a great coin, even a MS67 is $3,000, but in MS66 it's $70, in MS63 $10, in EF melt. Our scoring would be slanted towards rarity, and by using the average price in many grades we'll have a much better picture of how valuable a coin is. It doesn't make sense that the finest known 1931-S dime is worth more than the finest known 1916-D dime. Take the 1877 Indian cent in MS65 Red, a really neat coin. It's worth about one-fourth of the MS68 Red 1902 cent that brought $120,000 at auction. In VF the 1877 is worth $1,000, the 1902 $3.
I don't believe many modern coins are bringing a lot of money because the buyers necessarily enjoy them more. I think the buyers do it because the registry is telling them this is they get credit. These registry buyers are super competitive, not thinking as much as value as they are of winning the registry. I'd rather have them chasing rarity; it makes more sense numismatically. As to your question about a price reckoning, I don't think buyers of those coins are going to miss a meal if one of their coins drops a lot of value as they're using discretionary money."
The Rosen Numismatic Advisory vol. 47 no. 5
Steve