Doing something against obviously wrong values by users [solved]

I wanted to ask if something can be done against users who obviously enter nonsense values to their coins, which then get displayed on Numista. I'm not talking about a few % off prices, I'm talking about values that are from restrikes, different coins, or just trolls.

Examples:

2002 G (1/5 of the bullion value): N#15435

1881 (less than 1/10 of the bullion value): N#20493

Yes, I just can't see how that can be done?

When the entered value is too far below the bullion value, just don't enter it in the calculation. That would at least work for precious metal coins.

Maybe even notify the user that something is wrong.

I was thinking about the general coins, also non-precious coins. For the precious, you can only check for too low values, since there is no limit upwards?

Sjoelund

I was thinking about the general coins, also non-precious coins. For the precious, you can only check for too low values, since there is no limit upwards?

Exactly. Better than nothing, and it would already help with the two examples.

There's obviously something going on with the second one.  If the frequency is 0% (0 members with that coin) then there cannot be enough to even compute a value.  That one is more of a system bug.  @Xavier 

rsirian1

There's obviously something going on with the second one.  If the frequency is 0% (0 members with that coin) then there cannot be enough to even compute a value.  That one is more of a system bug.  @Xavier 

That one got fixed (manually?) it seems like. But the general problem still persists.

I like the idea of adding a limit. In addition if we really want to make Numista’s value more accurate I guess the Users need to add the date of purchase ( at least the year) next to the purchased value. Both coins Trooper8 mentioned are gold, as a material, its value is Generally increasing within the last 100 years like all metals. I am wondering if the lower prices recorded maybe they are not wrong but the Purchases were from long ago. 

Gold coin was issued in 2002 when ounce of gold costs 300$ and you're comparing it with current value of 2300$ per ounce.

Users are putting value which they paid for a coin, not how much coin costs today. That's why prices are not correct.

The same issue we have for all recent Ukrainian coins, when only limited number of users can buy coin directly from National Bank for price x1 and on the market same day price is x3, so prices in catalogue are not correct, but when I'm adding coin to my collection, I'm putting price x1 which I've paid, not x3 which is market price.

The problems you two mentioned should be solved with the following already implemented calculation (excluding of course someone entering a coin today they purchased many years ago): https://en.numista.com/forum/topic145285.html#p1155159

But even if someone purchased that coin I mentioned 22 years ago, it would STILL be way too cheap.

We can implement something like [do not take into price calculation values enter more than N years (1-3-5 ?)] but I'm afraid it will result in disappearing many prices. 

Maybe we should show a tooltip with message that price on the web site is the lower bound not exact price?

ArsenEverlast

We can implement something like [do not take into price calculation values enter more than N years (1-3-5 ?)] but I'm afraid it will result in disappearing many prices. 

 

Maybe we should show a tooltip with message that price on the web site is the lower bound not exact price?

If you're frightened of disappearing prices, we can enter a 0 in every missing field. That's also a lower bound and helps as much as a value that is less than 1/5th of the bullion value. /s

I do not think suppressing old purchases is a good solution, as u said a lot of data will be suppressed. I think the opposite can be done, with enough data we can mathematically calibrate the expexted price of coin. its not an easy task I know, but may Be achieved (or at least try)  with enough data.

Dr_Teek

I do not think suppressing old purchases is a good solution, as u said a lot of data will be suppressed. I think the opposite can be done, with enough data we can mathematically calibrate the expexted price of coin. its not an easy task I know, but may Be achieved (or at least try)  with enough data.

I only suggest to suppress the data when it doesn't help the site. Of course, the optimal solution would be to be able to enter your purchase time and then calculating it with the system that was implemented recently. But if nobody wants to implement that, then the next best thing is to just throw these values away. Because as you can see in the examples of this post, it just makes the site worse instead of better right now.

And again, that is only for these specific cases. There are still trolls or people who are incapable of identifying their own coins, where the values HAVE to be thrown away.

If the idea of someone buying a coin at a very very low price (a bargain of a life-time perhaps) as something as “wrong” , in addition to other problems with the user-inputted values, all I can see is that the idea of values on coins is pointless, when people ask how much a coin is worth this data may be useless. that's just my 2 cents (which I paid 0.002 cents for).

edit: While personally I have found zero use for the prices users have inputted on coins, if anyone has actually found a use for this feature please let me know, i'm curious. 

Only auction results should be used for Numista value calculation as those are the only ones we can verify

Compendium

Only auction results should be used for Numista value calculation as those are the only ones we can verify

That's kind of correct, but I don't really like that approach. That would delete basically all values of small denomination or very new coins.

Ahoj.

,,Our prices for each coin must be valued according to the possibility of how much the coin can be bought for in brick and mortar numismatic stores,,

Auctions are confusing, and individual countries have different approaches to local currency.

All the silver coins of the Middle Ages and Antiquity have skyrocketed in price over the past two years.

From the beginning, our pricing concept was nonsensical.  For me, it really doesn't matter how much an American buys one cent at the bank - to me, here in the heart of Europe, one American cent in a physical numismatic shop costs much, much more - I'll show you in the link:

https://www.numismatika-ostrava.cz/1-cent/

Offer prices only: ,,buy now,, are relevant - according to the current price offer in most numismatic brick-and-mortar stores.

https://numisargo.com/?gad_source=1&gclid=CjwKCAjw4f6zBhBVEiwATEHFVnBoEtgGDTAP6x9Zmyy3S9s_Qm3KBsGDKe07tn4XlJSPKvJilEnNGBoCNIsQAvD_BwE

Etc., Etc.     These are indicative prices for me.

Ivan

Precious metal coins may have “strange” values because the strange value was correct a few years ago (THe Italian 500 Lire Caravelle went from 2.8 Euro to 8 Euro in a short time span)

GiannaReggio

Precious metal coins may have “strange” values because the strange value was correct a few years ago (THe Italian 500 Lire Caravelle went from 2.8 Euro to 8 Euro in a short time span)

Why is that so difficult to understand… I'm only talking about values that were NEVER correct. Also (again), the new system implemented in Numista takes the rising prices of precious metals into account.

I record the price I paid, plus shipping, in  “Buying value”.  That is the number I want to keep for my own records.

Recently I purchased many unidentified uncirculated low-denomination 1960s Arabic coins from a US dealer's junk box.  For most of them I got very good deals.  Am I polluting the price results by entering the prices I paid?

For low-end coins, the value can easily be 50 cents unidentified, $5 in a 2x2 flip, and $50 in a slab.  Which values are the obviously wrong ones?

esnible

Which values are the obviously wrong ones?

I was only talking about bullion values WAY under melt. How many times do I have to repeat that? It gets frustrating how many people don't want to understand that….

Another example I found. 39€ for 1oz of gold.

N#366704

Hello,

I adjusted the algorithm for price estimate calculation so that user values that are way below bullion value are discarded.

Xavier

Hello,

I adjusted the algorithm for price estimate calculation so that user values that are way below bullion value are discarded.

And what exactly is “way below bullion” now? Because ¼ apparently isn't, for example.

There was an issue with prices refresh last night. Here are the new prices that should appear tomorrow:
 

The prices are still lower than the current bullion value. I will review whether we could adjust further.

Maybe we could use a stand-in symbol like NGC (¤) that gets added for any value that is under melt.