Note the following additional caution/information applies to interpretation:

These prices are for cymbals which don't have any major flaws. Major flaws are: serous keyhole, serious bell cracks (aka spiders), edge cracks, cracks along the tonal grooves, etc.

The links to the definitions and pictures for these condition categories (including pictures) will be added when I get to it.

According to Bill Maley (somebody I trust in these matters since he buys and sells a lot of them thus wouldn't be in business if he couldn't price things correctly to cover his overheads), Old Stamps get prices which are generally higher than Intermediate or New Stamps. Note that cymbals will sell for values outside the expected range, but not very often. Think of it as an 80/20 rule where 80% of sales will be in the expected range. Of course there are going to be exceptions. This might be the garage sale where you get one for $25 or it might be the top value achieved for a light Old Stamp 22" (over $4000). That doesn't make the rule of thumb incorrect, or any less useful.

My sample size is not yet big enough to do the sophisticated analysis I can do with Paiste 602s and Sound Creations which I have been tracking for a decade. And I'm still working out the best way to make my results easy for non statisticians to use. But based on the analysis I have done so far, it seems worth releasing what I've got because it already helps show what's happening. I'll be coming back and updating this page from time to time as I record new sales, and come up with improved reporting.

My analysis is based on completed sales, not on "asking" prices. A distribution for sales price is given, along with the "expected" price which is estimated from the median of the distribution. Given that actual sales figures cover a wide range, how do you know where your cymbal sits in that range? These are some of the factors which determine where a cymbal sits in the expected price range:

- Condition matters. If the cymbal is in excellent condition it will likely get more than one which is showing wear, other things being equal.
- Small edge dings, tiny bell cracks (which can happen with these given the smaller mounting hole), nicks, scratches, etc. lower the expected price, but I don't have enough data on by how much just yet.
- Higher prices may be expected for those with nice patina which haven't been polished within an inch of their life. Signatures under the bell and original paper tags, etc, may be a bonus, although obviously this factor interacts with still having patina. If a cymbal hasn't been overzealously cleaned it will tend to have both patina and the signature still under the bell.
- Certainly for 18" and bigger, "thin is in" so lighter weights tend to fetch higher prices. But note that weight ranges for Istanbul Ks aren't the same as modern cymbals ( that's the translation of a weight in grams to terms like "thin" or "medium" -- which always loses the original weight information).

In the analysis which follow we look at what production era the cymbal
come from because this is a factor in the expected price.
Bill Hartrick published the original work on trademark analysis, and you
can find his current version in
*The Gretsch Drum Book* by Rob Cook (published by
Rebeats) 2013.
In order to have large enough sample sizes for analysis I'm only dividing cymbals up
into the three major eras: Old, Intermediate, New.

If you don't know what these eras are, you can use my Which K Stamp? tool to get an overview or identify a particular cymbal. This K Stamp tool is under revision to include the earliest Turkish K Zildjians when the trademark said "Constantinople" rather than Istanbul". But in terms of the pricing analysis these are lumped in with Old Stamps for now due to low sample sizes. Don't be confused by the reuse of this K Constantinople naming in modern times. The American manufactured K Zildjian Constantinople is not part of this analysis. In fact, no K Zildjian cymbals from North America appear here. Only the Turkish ones.

The above is a plot of prices for 22" cymbals divided by Stamp (Trademark) Era. This is a box-and-whiskers plot and it may not be familiar to you. So here is a little introduction to what boxplots are and why they are a nice way to see what is going on in a set of data. In addition to the three eras of Turkish Ks, I've also included cymbals made by Roberto Spizzichino (RIP) in this analysis. Roberto's cymbals have achieved renown for their tonal qualities which are like Turkish Ks. The prices that Roberto's 22" cymbals are getting these days brings them into the same range as New Stamp Turkish Ks.

You can see that Old Stamps generally fetch higher prices than New or Intermediate stamps. In the fitting of a linear statistical model, the Intermediate Stamp price is the reference category. The expected (median) price for 22" Intermediates is $1800. That means than half sell for more than $1800 and half sell for less than $1800. We prefer the median because it is less influenced by outliers or a skewed distribution. We'll describe the other prices relative to Intermediate Stamps.

- Subtract $300 for New so $1500 for them
- Add $340 for Old so $2140 for them -- but notice that distribution goes a long way up the price scale
- Subtract $380 for Spizz so $1420 -- but beware the two oldest spizz sales I have are 2010 and 2011 and they got more expensive later so I might need to amend this analysis and remove them. The 2010 value is $775 and the 2011 value is $1050.

Small Sample Size Warning: This analysis is based on

9 intermediate

14 new

10 old

8 spizz

so it is preliminary. But some data is probably better than internet rumor, until I can collect more completed sales data. I usually try and get 25 examples of each diameter and era to get more reliable and stable estimates.

Statistical models show no sign of a change in prices over the years 2013 on, but again that could be do to small samples. But in this case the samples aren't going to get any bigger for 2013 and 2014 (unless somebody has data I haven't got).

Analysis date: 2 Dec 2015

As with 22" cymbals, there is no significant change in prices over the three years.

There is significant price variation across the three Stamp (Trademark) eras. Is the reported difference in the expected price of the 3 different stamp eras likely to be true in the total population of sales, given the strength of the effect in our sample? Yes, according to the analysis done in the R statistical system which is used for all my analysis and graphics these days.

Prediction equation:

Expected median price of 20" Intermediate = $693

Added value if New Stamp = $240

Added value if Old Stamp = $779

The smallest difference is Intermediate versus New, but this is still significant at p<0.02 (in lots of areas of research a value of p<0.05 is commonly used so this is well past that level of significance). In the 20" diameter New Stamps have a slightly higher expected price than Intermediate Stamps, with the highest expected price for Old Stamps. So the pattern is different for 22" versus 20" in terms of expected value of New versus Intermediate stamps. In both sizes Old Stamps have higher expected prices.

Residual standard error: 402.5 on 79 degrees of freedom

Multiple R-squared: 0.342, Adjusted R-squared: 0.3253

F-statistic: 20.53 on 2 and 79 DF, p-value: 6.603e-08

What do all these numbers mean? They measure the quality of fit of the regression equation. There is a very small probability that the price effect of production era (called Stamp in the modeling) is not true in the population from which this sample is drawn. But stamp era only accounts for about 33% of the variation in price (from the Adjusted R-squared value). Later on I'll fit a more complete model which includes condition and weight to see how much that improves predictability.

Analysis date: 2 Dec 2015

This analysis is based on n=41 completed sales, although 4 of these were in very bad condition and went for under $300. They have been excluded from the following analysis. In addition, data plotting and analysis showed that weight has a significant impact as well. The cymbals can be divided into those weighing 1796g and over, versus those which are much lighter. Most of the heavier ones sold for under $500. These are also excluded from the general analysis and treated separately. So there are 27 sales remaining in the main analysis. In addition you can see outliers for the Intermediate and New Stamps (one cymbal $800 and up for each). And the price distribution for Old Stamps is positively skewed (meaning it has a long tail stretching up into the higher price range).

I'm actually quite suspicious of the highest prices. I've checked back in my detailed raw data and found that in some cases eBay said these were sold at that price, but it isn't always as clear cut as it should be. Sometimes the outlier high price is a crazy high asking price and a sale didn't really take place although eBay suggests it did. In some cases (of some notorious sellers) I've actually tracked cymbals supposedly sold which then come back some months later from the same seller or an alias account for the same seller. In other cases there are a few sellers which I refer to as "super sellers" who for some reasons seem get much higher prices. The reasons may be perfectly legitimate: better ads, better merchandise, better after sales service. In other cases it is hard to be sure what is going on.

So having had a close look, I don't put much credence on the $800 New Stamp sale, the $950 Intermediate Stamp sale, or the Old Stamp sales above $1200. But I've left those points in for now. When I've got a bigger sample, I'll be in a better position to judge if they need to be trimmed out. Fortunately, using robust estimation techniques means that the few outlier values don't unduly influence the estimation functions.

The expected price for 18" old Ks is influenced by which stamp era you are looking at. Old stamps get higher prices. Their price advantage is estimated in my data as about $200 compared to Intermediate and New stamp era cymbals (p<0.05).

Prediction equation for cymbals less than 1800g:

Expected median price of 18" Intermediate = $590

Added value if New Stamp = same as Intermediate

Added value if Old Stamp = $200

Residual standard error: 213.7 on 24 degrees of freedom

Multiple R-squared: 0.2119, Adjusted R-squared: 0.1462

F-statistic: 3.226 on 2 and 24 DF, p-value: 0.05744

Note that this model only explains 15% of the variation in price. There are other factors at work. However, fitting a more sophisticated model which considers both stamp era and weight (for example) is tricky because of the relatively small sample size. Part of the problem is that lighter cymbals also happen to be the ones with an Old Stamp on them. Intermediate cymbals seem to be a bit heavier than New or Old, although there is the occasional cymbal which is over 1600g to be found in any stamp era.

The interaction of weight and stamp era confounds model fitting until we have more sales to work with. On the meantime, fitting a linear model with weight alone shows that lighter cymbals fetch higher prices (p<0.01), and a model with both weight and stamp era shows the same pattern.

Weight alone does better as a price predictor than stamp era alone. Adjusted R-squared is around 27% (up from 15%). The combination of weight and stamp era has an adjusted R-squared of 34%.

In summary:

18" cymbals which are in poor condition go for $300 (give or take). 18" cymbals which are in the 1800g up weight range sell in the range $400 - $500. If an 18" cymbal is in good condition and weighs less than 1800g then you use the estimating equation given above.

Analysis date: 2 Dec 2015

Here are the weights for 16" Turkish made K Zildjians I've got in my database (in grams):

1011

1043

1043 (int)

1048

1100

1100 (old)

1154

1155 (old)

1162 <- median weight for "lighter" ones

1192

1198 (new)

1203

1225

1270

1275 (new)

1280 <- overall median

===== natural place for a break in the distribution (above/below 1400g)

1426

1437

1445

1445

1468

1500 <- median weight for "heavier" ones

1510

1545

1545

1654

Analysis date: 2 Dec 2015

Based on 8 hi hat pairs the median expected value for 15" Hi Hats is $660, which is the same as the median expected value for the 14" hi hats (see below).

In addition, a band pair with wooden handles went for $575 and an Orchestral pair with leather handles went for $810. There were also also 8 single 15" cymbals sold, and the median price was $325.

Image: 14" K Hi Hat Price by Era

There are two obvious outliers where a high price ($1000 and above) was achieved. It looks like there are some differences, but the statistical test shows no significant difference in price between the different stamp eras. So for now the best choice of a prediction equation is

Expected median price of 14" Hi Hats = $658

with half selling between $537 and $775.

But it's early days and small samples.

The distribution of top versus bottom weights looks like this:

Which shows that the pairs for sale are matched top and bottom (versus the mid 1960s New Beat style with a heavier bottom. Except for one unusual pairing which stands out as a light top and the heaviest bottom. Note that when these Turkish made K Zildjian hats are sold they are sometimes referred to as "factory matched pairs" but that may or may not be the case. Matched in the sense of similar weights, yes. Matched at the factory in Istanbul? Matched by Gretsch when they arrived in the USA? Only your hairdresser knows for sure.

There are only 9 sales for single 14" cymbals, with a median price of $202.50 and half selling from between $200 and $225. It is looking like the expected price a single is less than half the expect price for a pair. But again it is early days on sample size and data collection.

Analysis date: 3 Dec 2015

There are only 2 13" single cymbals, both Old Stamps. They sold for $107 and $158. There was also a pair of band cymbals with wooden handles which sold for $285, but had a tonal groove crack at the bell bow transition which means one wasn't playable.

Analysis date: 3 Dec 2015

Here are some quotes from 2008:

"Old Ks are also going up in value fast. I've seen nice old stamp 22s get above $3k lately."

"My friend is a vintage drum dealer and he stashes the Ks, because they are going up in value so fast right now that it doesn't pay to sell them. If you can wait a month, it might be another 50 or 100 bucks."

"The old K market is unpredictable. In the 90s, a 22 was 300 bucks tops. Now its 10 times that for a really nice old stamp 22. In the early 2000s there was a 'boom' and then a 'bust' around 2004. Now the values have gotten out of control again and are skyrocketing."

and again in November 2015 the same beliefs seem to hold on eBay:

"These cymbals are highly desirable and are selling for huge sums here on Ebay."This 2015 quote comes from the sale of an Old Stamp 16" 1011g which sold for $270 with 7 bids. In terms of condition it had one tiny crack at the bell hole (coloquially known as spider cracks or just spiders). So is that a "huge sum"? How does it compare with similar sales? Compared to similar sales in 2014 and 2015 this one sold for under the expected median value of $350.

Clearly these beliefs aren't consistent with the current evidence. If 22" Turkish made Ks were getting "above $3k" in 2008 and they are selling for under this in 2013-2015, then clearly they can't have been going up "50 or 100 bucks per month" in the interim.

There was this event called the Global Financial Crisis (GFC) which had an impact. There were certain influential people (Bill Hartrick who created the timeline for Turkish made K Zildjians among them) who expected that prices for Turkish made K Zildjians would go on rising at 20% per year despite the GFC. In 2007 Bill wrote:

"I bought my first old stamp 22" K in 1999 for $650. Today we see one up for auction that will reach $3000 or maybe even $4000. If you don't believe that, then just stick around until the end of the auction."This 20% appreciation didn't continue. The past is not always the best guide to the future. Using one sale as a starting point and one sale as an end point and drawing a straight line between them is not really a big enough sample for successful extrapolation. You are more likely to end up with something like this (thanks to xkcd , the internet comic for science geeks):"1999 was 8 years ago. And the difference between what I paid then vs. what this item is worth if we assume a $3000 value today... correcting for inflationary affects means that the actual appreciation would be around 20%."

"I didn't do the calculations that we arrived at when we first analyzed this...someone else did. I provided the data and he did the figuring. He was an accountant or something like that, so he was able to give us the most detailed explanation including the inflation factor. And he showed us in detail just how he arrived at his result. His result agrees pretty closely with what I've shown you here. So I still stand by that figure of about 20% per year."

"By the way...did you see what the stock market did yesterday? It dropped about 360 points. You know what my old K's did? They went up."

It may seem too easy to poke fun at historical predictions of 20% appreciation which didn't come to pass. However, there is a serious side to doing this. One contributing factor to overestimation (well documented in the general case -- this isn't just about cymbals) is media coverage. The effect of media coverage on price bubbles is discussed in The Wisdom of Crowds. In this case I'd say that the "media environment" includes asking prices on quality web sites like Steve Maxwell, Classic Vintage Drums, etc. Plus the media environment also includes all the online forums where there are Turkish K "experts" expressing opinions (but no data) on how fast prices are rising. Here is an article about just how far out people are at estimating something getting lots of media attention.

Another thing which leads to overestimation of prices is that some people think that you look on eBay (or the premier sites like Steve Maxwell and Classic Vintage Drums) for a couple of listings and get your idea of price from what people are asking. That isn't a sound way to do it. Asking isn't getting. You really need to look at completed sales. And you need to look at a number of completed sales to get any robust pricing information. People tend to focus on individual values, rather than the full distribution. Again, this is a general phenomenon not restricted to Turkish Ks or even cymbals. Actual selling prices are quite variable, and two Turkish Ks which are pretty much interchangeable (same diameter, stamp era, weight class, condition) might sell a day apart and one gets $600 and the other gets $1200. So yes, the median price might be $730, but the range includes much higher prices and much lower ones. And yes, your cymbal might sell for $1200, but that doesn't mean equivalent cymbals all sell for $1200.

Another reason people overestimate the expected prices of Turkish Ks is what is called confirmation bias. In a nutshell, if you don't record all the sales (or a representative sample) then you will selectively remember the more extreme values because they stick in your memory. For those who believe that Turkish Ks are going up in value all the time, they tend to "confirm" this belief system by remembering the highest prices. This is a well known phenomenon, and it is why collection of proper statistics is valuable to society. Now you could just claim that I'm suffering from "confirmation bias" myself since I'm a statistician and thus I remember the cases where the public beliefs about things are at variance with what is actually happening. But unless you are able to front up with lots of properly recorded data which shows my equations are poorly estimated, you are just adding more "evidence free" opinion into what I'm hoping will be an increasingly evidence based discussion.

Another contributing factor to people overestimating gains (and indirectly prices) is failure to include the cost of buying and selling when considering price. Sellers who sell on eBay pay 10% percent of the sales value plus the 10% of the domestic shipping cost. Usually the buyer pays the shipping cost, but the seller still pays eBay percentage of the shipping cost. So we are over 10% in buying and selling costs, and that's before PayPal starts taking their cut. When a cymbal is later sold for a higher amount, it is important to subtract the buying and selling costs from perceived "price increases". If Turkish made K Zildjian cymbals are being sold via other channels the fees are lower. But when they are sold through eBay unless you get maybe 15% more than you paid for the cymbal (including shipping) you haven't made any profit. eBay and PayPal ate your profit.

The final issue I'll mention is related to care in naming. The common nickname for Turkish made K Zildjian cymbals is "old Ks". But these come in different stamp eras and have significantly different expected prices. I believe part of what has happened to create misleading price expectations is people not being careful to distinguish "old Ks" from "Old Stamp Ks". Given what we know about the price difference between Old Stamp Turkish made K Zildjians (higher) and New and Intermediate Stamp prices, every time estimates of value are given you should ask yourself: was the person giving this figure careful to distinguish which cymbals they are talking about? And what is their sample size?

or back to the introduction page

text first published 28 Nov 2015 3:17 PM

text last updated 25 Mar 2017 4:02 PM

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.