Market trends: numindex

Paradisaea

There was an interesting article in Coins Weekly recently, titled “numindex – The Stock Index for Coins”.

Article: https://coinsweekly.com/numindex-the-stock-index-for-coins/

The numismatic coin market index (numindex) featured in the article aims to present market trends in European coins. Looks to be a similar concept to the PCGS 3000 index (of PCGS-graded US coins).

What I found interesting was that the index captures the post-pandemic price correction discussed during the most recent (March ’23) Heritage "World & Ancient Coins - State of the Market" broadcast. Having felt ‘priced out’ of auctions lately, the downward trend that this index depicts is a beacon of hope!

But is anyone experiencing a price correction to the degree (~30% decline from peak) that the index suggests?

Source: https://numindex.com/

bidask

Some of the assumptions noted in Table 1 ( principles of numindex) are highly questionable in my opinion.

For example,

1) ( availability of coins)
“Therefore each coin in the index must be traded or auctioned several times a year”

This assumption alone seems to guarantee a drift downward in price since appearences are frequently per year

2) ( individual coin price )
“Each coin in the index should cost between chf 100 -chf 2000….”

Why so arbitrary?

Many great coins trade far far less frequent than several times a year and cost alot more than what was stated in the index.

How come those are not included ?

Those are the coins that seem to do well over time based on auction records.

Exbrit

Any attempt to track and analyze trends has it’s good and bad points. You have to limit the size of the sample size in order to reasonably analyze it. This has its inherit downfall, but at the same time allows some meaningful trend examination.

neildrobertson

@bidask said:
Some of the assumptions noted in Table 1 ( principles of numindex) are highly questionable in my opinion.

For example,

1) ( availability of coins)
“Therefore each coin in the index must be traded or auctioned several times a year”

This assumption alone seems to guarantee a drift downward in price since appearences are frequently per year

2) ( individual coin price )
“Each coin in the index should cost between chf 100 -chf 2000….”

Why so arbitrary?

Many great coins trade far far less frequent than several times a year and cost alot more than what was stated in the index.

How come those are not included ?

Those are the coins that seem to do well over time based on auction records.

I think the case for tracking coins under 2000 chf is that it is more representative of the bulk of collectors coins and represents most of the coins trading volume. The volume above that price point is small.

I was taught to use the square root of the number of samples to estimate Stochastic error. You really need to have a lot of data points to reduce random variability. Higher value coins with high variability can cancel out trends on lower value coins to the point where it actually weakens its stability and effectiveness as a representative index. I'm just not sure you can index high value coins and have it be meaningful at all.

We see similar approaches to how the S&P500 index is reported in 2023. The market cap of the top 6 or so companies is massive compared to the bulk, to the point that the index is telling you more about what the top 6 are doing than the other 400+. Sometime ls when the index is out of sync with other macro trends I, its because the top heavyweights are going through a very specific event that obscures the results of the bulk of the companies in the index, which are following the macro trend.

I also argue time and time again, good performance at the high end of the market is not an indication that the overall market is healthy. I don't want these sorts of exercises to paint a positive or negative picture. I want them to be accurate and be able to tell you something. It just may be that something is very specific.

John Conduitt

There has definitely been a peak and a reduction. You can see this is you analyse CNG's sales results.

I don't think the reduction was 30%. It's just a factor of a few high peaks that were followed by troughs. If you draw a line through the graph to average those out, it's less than 20%. Unless you bought a lot of coins from Feb-May 2022, you wouldn't notice.

The 30 coins are a bit odd. Why only 30? 2 Belgian, 2 French, 3 Italian, 2 Swiss, 2 German, 2 Greek, 2 Polish, and 1 of 11 other countries (including Britain). 4 coins that aren't European. I'm not sure what market this represents as it's not weighted to the heaviest trade. 3 sales of a Congolese 2 Francs isn't representative of much.

Even the British coin - a non-circulating £2 gold coin from 1887 - surely isn't representative of British coins collected. It also sometimes goes above CHF2000 - did they exclude those sales? Or include those? Either way, that would skew the results.

WCC

I don't see any kind of generic world coin price index as being meaningful. In the US, it would also need to be by TPG grade but that's not relevant hardly anywhere else.

Years ago, there was a price tracking service for Union and ZAR (maybe RSA too but don't remember) South African coinage, but it wasn't an index and it didn't last, presumably due to lack of demand (scale). It was similar to the auction data in Coin Facts.

Years later, someone else requested feedback on a scale down version of the original, but I told them it was a waste of time to track the sales of so many really low value coins. I don't know whether it ever launched.

Paradisaea

@John Conduitt said:
I don't think the reduction was 30%. It's just a factor of a few high peaks that were followed by troughs. If you draw a line through the graph to average those out, it's less than 20%. Unless you bought a lot of coins from Feb-May 2022, you wouldn't notice.

Yes, admittedly, I wasn't very scientific in my approach. In hindsight, a 20% reduction would seem a more reasonable interpretation of the index.

I like the idea of a general index, particularly for contextualising market change over time. Change can be interpreted differently depending on experience. In the limited time I have been collecting, the only major market shift I've experienced has been that associated with the pandemic. It caused me to review the feasibility of my collecting goals. As part of that process, I looked into historical market trends (that's how I stumbled onto numindex) to gauge how similar market shifts have evolved in the past (i.e., permanent level shift, temporary peak, or combination of both).

There are risks when drawing conclusions from any index, as outlined in prior posts, but the same can be said of annedotal reports of change or focusing on a single series/coin. My perspective is the more tools that permit cross-validation, the better!

Will be interesting to see how this new index evolves... I hope the 30 coins considered is the starting point, with the index expanding in scope over time.

pruebas

Rather than relying on an index, it seems to me a more accurate gauge of the market is to watch a few auctions and bidding patterns.

One auction will not generally reveal any trends, but watching a few big auctions will.

The upside is you can buy anything that falls through the cracks.

Edited to add: To me, the usefulness of the data is directly proportional to the cost to get that data.