|
 |
| | | Author: | brick_projects  | | Posted: | Jul 20, 2016 08:53 | | Subject: | standard deviation on price indices | | Viewed: | 200 times | | Topic: | Suggestions | | Status: | Discarded | |
|
| Please implement standard deviation.
By eliminating the (small number)% cheapest and (small number) % most expensively
priced lots, the average price index would be "more statistically significant"
in scientific terms ( less error margin ).
This is only necessary for items that have at least e.g. 10 distinct lots for
sale.
Why this suggestion ?
Yesterday i noticed one particular seller set all his prices at ( BL avg x 10000
). This caused average price index "lots for sale" to be inflated at least
tenfold.
Someone else suggested he made a mistake in currency.
If this one shopkeeper persists, unwanted "extreme" inflation may occur on several
very common items, and those lots - for sellers not paying attention when adding
to their inventory - might become impossible to sell.
|
|
|
| | | | | |
| | | | | Author: | FigBits | | Posted: | Jul 20, 2016 09:02 | | Subject: | Re: standard deviation on price indices | | Viewed: | 37 times | | Topic: | Suggestions | |
|
| In Suggestions, brick_projects writes:
| | Please implement standard deviation.
By eliminating the (small number)% cheapest and (small number) % most expensively
priced lots, the average price index would be "more statistically significant"
in scientific terms ( less error margin ).
This is only necessary for items that have at least e.g. 10 distinct lots for
sale.
Why this suggestion ?
Yesterday i noticed one particular seller set all his prices at ( BL avg x 10000
). This caused average price index "lots for sale" to be inflated at least
tenfold.
Someone else suggested he made a mistake in currency.
If this one shopkeeper persists, unwanted "extreme" inflation may occur on several
very common items, and those lots - for sellers not paying attention when adding
to their inventory - might become impossible to sell.
|
Voted no.
I do not want BL to remove any data. If you want to suggest *adding* information,
I would be fine with that. (Such as adding the median price.)
To any seller who blindly uses the current For Sale Average to set their prices,
I would have a simple suggestion: Don't.
So, for the suggestion: No. Please do not remove information which is useful
to many of us, just because a few sellers might be using that information in
a silly way (that only hurts themselves).
--
Marc.
|
|
|
| | | | | | | | | |
| | | | | | | Author: | jeslego | | Posted: | Jul 20, 2016 10:07 | | Subject: | Re: standard deviation on price indices | | Viewed: | 31 times | | Topic: | Suggestions | |
|
| In Suggestions, FigBits writes:
| | In Suggestions, brick_projects writes:
| | Please implement standard deviation.
By eliminating the (small number)% cheapest and (small number) % most expensively
priced lots, the average price index would be "more statistically significant"
in scientific terms ( less error margin ).
This is only necessary for items that have at least e.g. 10 distinct lots for
sale.
Why this suggestion ?
Yesterday i noticed one particular seller set all his prices at ( BL avg x 10000
). This caused average price index "lots for sale" to be inflated at least
tenfold.
Someone else suggested he made a mistake in currency.
If this one shopkeeper persists, unwanted "extreme" inflation may occur on several
very common items, and those lots - for sellers not paying attention when adding
to their inventory - might become impossible to sell.
|
Voted no.
I do not want BL to remove any data. If you want to suggest *adding* information,
I would be fine with that. (Such as adding the median price.)
To any seller who blindly uses the current For Sale Average to set their prices,
I would have a simple suggestion: Don't.
So, for the suggestion: No. Please do not remove information which is useful
to many of us, just because a few sellers might be using that information in
a silly way (that only hurts themselves).
--
Marc.
|
Back when I took stats, trimming extreme outliers was considered a desirable
practice. This would be particularly true if the outlier was not so much a black
swan as a data recording error or false reporting. Does anyone know if this
advice has changed?
In any case, the notion of "removing" data does not seem to fully apply here.
No one is suggesting it be removed from the price guide. Someone interested
in reviewing every data point is free to look at them where they are reported
in the price guide.
Frankly, the only reason I can think of for not making this change is that some
sellers will make fewer and smaller mistakes. That will prevent the rest of
us from enjoying the arbitrage.
|
|
|
| | | | | | | | | | | | | |
| | | | | | | | | Author: | FigBits | | Posted: | Jul 20, 2016 11:56 | | Subject: | Re: standard deviation on price indices | | Viewed: | 32 times | | Topic: | Suggestions | |
|
| In Suggestions, bljesprop writes:
| | In Suggestions, FigBits writes:
| | In Suggestions, brick_projects writes:
| | Please implement standard deviation.
By eliminating the (small number)% cheapest and (small number) % most expensively
priced lots, the average price index would be "more statistically significant"
in scientific terms ( less error margin ).
This is only necessary for items that have at least e.g. 10 distinct lots for
sale.
Why this suggestion ?
Yesterday i noticed one particular seller set all his prices at ( BL avg x 10000
). This caused average price index "lots for sale" to be inflated at least
tenfold.
Someone else suggested he made a mistake in currency.
If this one shopkeeper persists, unwanted "extreme" inflation may occur on several
very common items, and those lots - for sellers not paying attention when adding
to their inventory - might become impossible to sell.
|
Voted no.
I do not want BL to remove any data. If you want to suggest *adding* information,
I would be fine with that. (Such as adding the median price.)
To any seller who blindly uses the current For Sale Average to set their prices,
I would have a simple suggestion: Don't.
So, for the suggestion: No. Please do not remove information which is useful
to many of us, just because a few sellers might be using that information in
a silly way (that only hurts themselves).
--
Marc.
|
Back when I took stats, trimming extreme outliers was considered a desirable
practice. This would be particularly true if the outlier was not so much a black
swan as a data recording error or false reporting. Does anyone know if this
advice has changed?
In any case, the notion of "removing" data does not seem to fully apply here.
No one is suggesting it be removed from the price guide. Someone interested
in reviewing every data point is free to look at them where they are reported
in the price guide.
|
The information that is being suggested to be removed is the mathematical mean.
I don't want that removed. If people want to have additional information
added, such as a trimmed mean, that's fine. But not as a replacement
of what we have now.
| | Frankly, the only reason I can think of for not making this change is that some
sellers will make fewer and smaller mistakes. That will prevent the rest of
us from enjoying the arbitrage.
|
Not all prices have normal distribution. Sometimes the outliers are significant.
http://www.bricklink.com/catalogPG.asp?P=62360&colorID=11
For the last several months, the trimmed mean for this part would have shown
around $1. The actual mean (which is what currently gets displayed) was $2. If
I had some last month, I probably would have listed them at $15, which would
likely have been the highest listing at the time. And they probably would have
sold by now.
Now, in that case, the mean isn't a particularly good representation either.
But it's a better number for a seller wondering what price they should ask,
than the trimmed mean.
--
Marc.
|
|
|
| | | | | | | | | | | | | | | | | |
| | | | | | | | | | | Author: | yorbrick | | Posted: | Jul 20, 2016 12:14 | | Subject: | Re: standard deviation on price indices | | Viewed: | 21 times | | Topic: | Suggestions | |
|
| | | Not all prices have normal distribution. Sometimes the outliers are significant.
http://www.bricklink.com/catalogPG.asp?P=62360&colorID=11
For the last several months, the trimmed mean for this part would have shown
around $1. The actual mean (which is what currently gets displayed) was $2. If
I had some last month, I probably would have listed them at $15, which would
likely have been the highest listing at the time. And they probably would have
sold by now.
Now, in that case, the mean isn't a particularly good representation either.
But it's a better number for a seller wondering what price they should ask,
than the trimmed mean.
|
For parts like that that are getting rarer due to many people wanting them for
well known MOCs (similar has happened for the Bat-Pod), then having another option
to use a 1-month mean for sold parts may also make sense.
Although this is where a seller that knows his stock and his business makes money.
I bought quite a few of those windshields from people listing them cheap (and
also the dino necks and fezes) because it was fairly obvious they would increase
in value.
|
|
|
| | | | | | | | | | | | | | | | | | | | | |
| | | | | | | | | | | | | Author: | FigBits | | Posted: | Jul 20, 2016 12:36 | | Subject: | Re: standard deviation on price indices | | Viewed: | 23 times | | Topic: | Suggestions | |
|
| In Suggestions, yorbrick writes:
| | | | Not all prices have normal distribution. Sometimes the outliers are significant.
http://www.bricklink.com/catalogPG.asp?P=62360&colorID=11
For the last several months, the trimmed mean for this part would have shown
around $1. The actual mean (which is what currently gets displayed) was $2. If
I had some last month, I probably would have listed them at $15, which would
likely have been the highest listing at the time. And they probably would have
sold by now.
Now, in that case, the mean isn't a particularly good representation either.
But it's a better number for a seller wondering what price they should ask,
than the trimmed mean.
|
For parts like that that are getting rarer due to many people wanting them for
well known MOCs (similar has happened for the Bat-Pod), then having another option
to use a 1-month mean for sold parts may also make sense.
|
That part is already possible now, using the "Price & Group by month" link on
the PG pages. (It's grouped by calendar month, as opposed to showing the
previous 30 days worth of sales.)
| | Although this is where a seller that knows his stock and his business makes money.
I bought quite a few of those windshields from people listing them cheap (and
also the dino necks and fezes) because it was fairly obvious they would increase
in value.
|
Right. There are some other parts where the listings don't follow a normal
distribution, which might have illustrated my point better ... but I didn't
want to draw attention to them because I am currently benefiting from that knowledge!
--
Marc.
|
|
|
| | | | | | | | | | | | | | | | | |
| | | | | | | | | | | Author: | jeslego | | Posted: | Jul 20, 2016 12:26 | | Subject: | Re: standard deviation on price indices | | Viewed: | 31 times | | Topic: | Suggestions | |
|
| In Suggestions, FigBits writes:
| | In Suggestions, bljesprop writes:
| | In Suggestions, FigBits writes:
| | In Suggestions, brick_projects writes:
| | Please implement standard deviation.
By eliminating the (small number)% cheapest and (small number) % most expensively
priced lots, the average price index would be "more statistically significant"
in scientific terms ( less error margin ).
This is only necessary for items that have at least e.g. 10 distinct lots for
sale.
Why this suggestion ?
Yesterday i noticed one particular seller set all his prices at ( BL avg x 10000
). This caused average price index "lots for sale" to be inflated at least
tenfold.
Someone else suggested he made a mistake in currency.
If this one shopkeeper persists, unwanted "extreme" inflation may occur on several
very common items, and those lots - for sellers not paying attention when adding
to their inventory - might become impossible to sell.
|
Voted no.
I do not want BL to remove any data. If you want to suggest *adding* information,
I would be fine with that. (Such as adding the median price.)
To any seller who blindly uses the current For Sale Average to set their prices,
I would have a simple suggestion: Don't.
So, for the suggestion: No. Please do not remove information which is useful
to many of us, just because a few sellers might be using that information in
a silly way (that only hurts themselves).
--
Marc.
|
Back when I took stats, trimming extreme outliers was considered a desirable
practice. This would be particularly true if the outlier was not so much a black
swan as a data recording error or false reporting. Does anyone know if this
advice has changed?
In any case, the notion of "removing" data does not seem to fully apply here.
No one is suggesting it be removed from the price guide. Someone interested
in reviewing every data point is free to look at them where they are reported
in the price guide.
|
The information that is being suggested to be removed is the mathematical mean.
I don't want that removed. If people want to have additional information
added, such as a trimmed mean, that's fine. But not as a replacement
of what we have now.
| | Frankly, the only reason I can think of for not making this change is that some
sellers will make fewer and smaller mistakes. That will prevent the rest of
us from enjoying the arbitrage.
|
Not all prices have normal distribution. Sometimes the outliers are significant.
http://www.bricklink.com/catalogPG.asp?P=62360&colorID=11
For the last several months, the trimmed mean for this part would have shown
around $1. The actual mean (which is what currently gets displayed) was $2. If
I had some last month, I probably would have listed them at $15, which would
likely have been the highest listing at the time. And they probably would have
sold by now.
Now, in that case, the mean isn't a particularly good representation either.
But it's a better number for a seller wondering what price they should ask,
than the trimmed mean.
--
Marc.
|
As you note, your example really does not prove that one or the other is superior.
However, a measurement of the price trend might have helped to identify the
impact of the part being used in a MOC. Better still, knowing that the demand
for the part was about to change would have tipped a seller not to use the average
price heuristic. T
In the parts of the catalog I typically study, a trimmed mean would be a much
better choice as the first option. It certainly should be easy to provide both.
The problem being addressed how sellers can quickly price a large variety of
inexpensive items without getting it wrong. Price trends would enhance this
information. A warning that the amount of transactions is low (ie that attributes
other than price may have greater impact) would also be helpful (and would have
also flagged your example).
|
|
|
| | | | | | | | | | | | | | | | | | | | | |
| | | | | | | | | | | | | Author: | FigBits | | Posted: | Jul 20, 2016 12:41 | | Subject: | Re: standard deviation on price indices | | Viewed: | 38 times | | Topic: | Suggestions | |
|
| In Suggestions, bljesprop writes:
| | In Suggestions, FigBits writes:
| | In Suggestions, bljesprop writes:
| | In Suggestions, FigBits writes:
| | In Suggestions, brick_projects writes:
| | Please implement standard deviation.
By eliminating the (small number)% cheapest and (small number) % most expensively
priced lots, the average price index would be "more statistically significant"
in scientific terms ( less error margin ).
This is only necessary for items that have at least e.g. 10 distinct lots for
sale.
Why this suggestion ?
Yesterday i noticed one particular seller set all his prices at ( BL avg x 10000
). This caused average price index "lots for sale" to be inflated at least
tenfold.
Someone else suggested he made a mistake in currency.
If this one shopkeeper persists, unwanted "extreme" inflation may occur on several
very common items, and those lots - for sellers not paying attention when adding
to their inventory - might become impossible to sell.
|
Voted no.
I do not want BL to remove any data. If you want to suggest *adding* information,
I would be fine with that. (Such as adding the median price.)
To any seller who blindly uses the current For Sale Average to set their prices,
I would have a simple suggestion: Don't.
So, for the suggestion: No. Please do not remove information which is useful
to many of us, just because a few sellers might be using that information in
a silly way (that only hurts themselves).
--
Marc.
|
Back when I took stats, trimming extreme outliers was considered a desirable
practice. This would be particularly true if the outlier was not so much a black
swan as a data recording error or false reporting. Does anyone know if this
advice has changed?
In any case, the notion of "removing" data does not seem to fully apply here.
No one is suggesting it be removed from the price guide. Someone interested
in reviewing every data point is free to look at them where they are reported
in the price guide.
|
The information that is being suggested to be removed is the mathematical mean.
I don't want that removed. If people want to have additional information
added, such as a trimmed mean, that's fine. But not as a replacement
of what we have now.
| | Frankly, the only reason I can think of for not making this change is that some
sellers will make fewer and smaller mistakes. That will prevent the rest of
us from enjoying the arbitrage.
|
Not all prices have normal distribution. Sometimes the outliers are significant.
http://www.bricklink.com/catalogPG.asp?P=62360&colorID=11
For the last several months, the trimmed mean for this part would have shown
around $1. The actual mean (which is what currently gets displayed) was $2. If
I had some last month, I probably would have listed them at $15, which would
likely have been the highest listing at the time. And they probably would have
sold by now.
Now, in that case, the mean isn't a particularly good representation either.
But it's a better number for a seller wondering what price they should ask,
than the trimmed mean.
--
Marc.
|
As you note, your example really does not prove that one or the other is superior.
However, a measurement of the price trend might have helped to identify the
impact of the part being used in a MOC. Better still, knowing that the demand
for the part was about to change would have tipped a seller not to use the average
price heuristic. T
In the parts of the catalog I typically study, a trimmed mean would be a much
better choice as the first option. It certainly should be easy to provide both.
The problem being addressed how sellers can quickly price a large variety of
inexpensive items without getting it wrong. Price trends would enhance this
information. A warning that the amount of transactions is low (ie that attributes
other than price may have greater impact) would also be helpful (and would have
also flagged your example).
|
I agree with all that.
As always, more information is better than less information.
--
Marc.
|
|
|
| | | | | | | | | | | | | | | | | |
| | | | | | | | | | | Author: | jedvii | | Posted: | Jul 22, 2016 02:41 | | Subject: | Re: standard deviation on price indices | | Viewed: | 38 times | | Topic: | Suggestions | |
|
| In Suggestions, FigBits writes:
| | In Suggestions, bljesprop writes:
| | In Suggestions, FigBits writes:
| | In Suggestions, brick_projects writes:
| | Please implement standard deviation.
By eliminating the (small number)% cheapest and (small number) % most expensively
priced lots, the average price index would be "more statistically significant"
in scientific terms ( less error margin ).
This is only necessary for items that have at least e.g. 10 distinct lots for
sale.
Why this suggestion ?
Yesterday i noticed one particular seller set all his prices at ( BL avg x 10000
). This caused average price index "lots for sale" to be inflated at least
tenfold.
Someone else suggested he made a mistake in currency.
If this one shopkeeper persists, unwanted "extreme" inflation may occur on several
very common items, and those lots - for sellers not paying attention when adding
to their inventory - might become impossible to sell.
|
Voted no.
I do not want BL to remove any data. If you want to suggest *adding* information,
I would be fine with that. (Such as adding the median price.)
To any seller who blindly uses the current For Sale Average to set their prices,
I would have a simple suggestion: Don't.
So, for the suggestion: No. Please do not remove information which is useful
to many of us, just because a few sellers might be using that information in
a silly way (that only hurts themselves).
--
Marc.
|
Back when I took stats, trimming extreme outliers was considered a desirable
practice. This would be particularly true if the outlier was not so much a black
swan as a data recording error or false reporting. Does anyone know if this
advice has changed?
In any case, the notion of "removing" data does not seem to fully apply here.
No one is suggesting it be removed from the price guide. Someone interested
in reviewing every data point is free to look at them where they are reported
in the price guide.
|
The information that is being suggested to be removed is the mathematical mean.
I don't want that removed. If people want to have additional information
added, such as a trimmed mean, that's fine. But not as a replacement
of what we have now.
| | Frankly, the only reason I can think of for not making this change is that some
sellers will make fewer and smaller mistakes. That will prevent the rest of
us from enjoying the arbitrage.
|
Not all prices have normal distribution. Sometimes the outliers are significant.
http://www.bricklink.com/catalogPG.asp?P=62360&colorID=11
For the last several months, the trimmed mean for this part would have shown
around $1. The actual mean (which is what currently gets displayed) was $2. If
I had some last month, I probably would have listed them at $15, which would
likely have been the highest listing at the time. And they probably would have
sold by now.
Now, in that case, the mean isn't a particularly good representation either.
But it's a better number for a seller wondering what price they should ask,
than the trimmed mean.
--
Marc.
|
holy cow, what caused the sudden rise in the price of that part?
|
|
|
| | | | | |
| | | | | Author: | yorbrick | | Posted: | Jul 20, 2016 09:40 | | Subject: | Re: standard deviation on price indices | | Viewed: | 32 times | | Topic: | Suggestions | |
|
| | | If this one shopkeeper persists, unwanted "extreme" inflation may occur on several
very common items, and those lots - for sellers not paying attention when adding
to their inventory - might become impossible to sell.
|
There is the real problem here.
|
|
| | | | | |
| | | | | Author: | jancg | | Posted: | Jul 20, 2016 09:55 | | Subject: | Re: standard deviation on price indices | | Viewed: | 36 times | | Topic: | Suggestions | |
|
| Voted no.
Same argument as others: don't hide any information. If you want a smarter
pricing routine, build it yourself using the BL API.
|
|
| | | | | | | | | |
| | | | | | | Author: | ToddMyers | | Posted: | Jul 20, 2016 09:57 | | Subject: | (Cancelled) | | Viewed: | 37 times | | Topic: | Suggestions | |
|
| | (Cancelled) |
|
| | | | | | | | | | | | | |
| | | | | | | | | Author: | jancg | | Posted: | Jul 20, 2016 10:09 | | Subject: | Re: standard deviation on price indices | | Viewed: | 28 times | | Topic: | Suggestions | |
|
| I apologize. Naturally, most sellers don't use the BL API.
I probably would vote yes if the suggestion was to ADD 'mean price' and/or
'median price' and/or 'mean qty price' and/or 'median qty
price'. Perhaps even with settings on your account to switch either on/off.
|
|
| | | | | | | | | | | | | |
| | | | | | | | | Author: | Leftoverbricks | | Posted: | Jul 20, 2016 11:06 | | Subject: | Re: standard deviation on price indices | | Viewed: | 30 times | | Topic: | Suggestions | |
|
| In Suggestions, ToddMyers writes:
| | In Suggestions, jancg writes:
| | If you want a smarter pricing routine, build it yourself using the BL API.
|
Yes, because every seller here naturally has the skills to do this.
What a terrible option.
|
Great remark!
|
|
| | | | | |
| | | | | Author: | garbanzo | | Posted: | Jul 20, 2016 10:09 | | Subject: | Re: standard deviation on price indices | | Viewed: | 29 times | | Topic: | Suggestions | |
|
| In Suggestions, brick_projects writes:
| | Please implement standard deviation.
By eliminating the (small number)% cheapest and (small number) % most expensively
priced lots, the average price index would be "more statistically significant"
in scientific terms ( less error margin ).
This is only necessary for items that have at least e.g. 10 distinct lots for
sale.
Why this suggestion ?
Yesterday i noticed one particular seller set all his prices at ( BL avg x 10000
). This caused average price index "lots for sale" to be inflated at least
tenfold.
Someone else suggested he made a mistake in currency.
If this one shopkeeper persists, unwanted "extreme" inflation may occur on several
very common items, and those lots - for sellers not paying attention when adding
to their inventory - might become impossible to sell.
|
I don't understand the downvotes. Implementing this doesn't necessarily
mean replacing current data.
I voted yes, this makes sense. I would also like to see 6-month mode and median
in addition to mean.
|
|
|
| | | | | | | | | |
| | | | | | | Author: | jancg | | Posted: | Jul 20, 2016 10:13 | | Subject: | Re: standard deviation on price indices | | Viewed: | 29 times | | Topic: | Suggestions | |
|
| Perhaps a down vote is too harsh, but several downvoters want to stress not to
REPLACE the exact 'avg price' and 'avg qty price'.
Leaving out the most extreme samples is still a sound strategy when doing statistics.
|
|
| | | | | | | | | |
| | | | | | | Author: | yorbrick | | Posted: | Jul 20, 2016 11:16 | | Subject: | Re: standard deviation on price indices | | Viewed: | 27 times | | Topic: | Suggestions | |
|
| | | I don't understand the downvotes. Implementing this doesn't necessarily
mean replacing current data.
I voted yes, this makes sense. I would also like to see 6-month mode and median
in addition to mean.
|
I voted no, as the first line of the suggestion is different to the rest.
Calculating the standard deviation is fine (first line of the suggestion).
The rest of it, I am against. They want data eliminated from the calculations
of the mean.
|
|
| | | | | |
| | | | | Author: | ToriHada | | Posted: | Jul 20, 2016 10:52 | | Subject: | Re: standard deviation on price indices | | Viewed: | 42 times | | Topic: | Suggestions | |
|
| I voted yes. This suggestion has been made many times before, including by myself.
There is no reason why users cannot have the OPTION to exclude extreme data points
when using the Price Guide. Just like we now have the option to exclude incomplete
sets from the Price Guide. If someone does not like this option, then don't
use it. But don't deny it to others who would find it useful.
The idea that more information is better is not necessarily true. Not if you
include "information" that is misleading, untrue or a mistake. Furthermore,
it has been long established in statistics that removing extreme end-point anomalies
can make the data and averaging more accurate and statistically relevant.
This OPTION can be implemented as a percentage as the OP suggests, or in a fixed
manner where the highest and lowest one or two listings or data endpoints are
excluded when determining the average.
But in addition to this option, I would also include the further OPTION for members
to select MEDIAN price data as opposed to the mean or average.
Thor
|
|
|
| | | | | |
| | | | | Author: | Made_In_Bricks | | Posted: | Jul 20, 2016 11:19 | | Subject: | Re: standard deviation on price indices | | Viewed: | 35 times | | Topic: | Suggestions | |
|
| voted no - better suggestion, allow users to program how they want the price
guide displayed via their own settings, if users could define the rules they
want everyone would be satisfied with their concerns over the price guide.
Go back and look at my thread on price guide manipulation. I pointed out that
any jerk can manipulate this to where it causes more work for others (and actually
caused it to try and prove a point), basically one can sabotage other sellers
here by listing crazy prices in common parts, they could further manipulate it
by setting up a second account and then buy such listings and drive people nuts.
|
|
| | | | | | | | | |
| | | | | | | Author: | FigBits | | Posted: | Jul 20, 2016 12:03 | | Subject: | Re: standard deviation on price indices | | Viewed: | 31 times | | Topic: | Suggestions | |
|
| In Suggestions, BrickItYourself writes:
| | voted no - better suggestion, allow users to program how they want the price
guide displayed via their own settings, if users could define the rules they
want everyone would be satisfied with their concerns over the price guide.
|
Yes, this.
Changing the way that the current averages are calculated is a terrible idea.
Adding options for users to see the data in whatever way they find useful is
a great idea.
--
Marc.
|
|
| | | | | |
| | | | | Author: | wildchicken13 | | Posted: | Nov 17, 2020 01:17 | | Subject: | Re: standard deviation on price indices | | Viewed: | 38 times | | Topic: | Suggestions | |
|
| In Suggestions, brick_projects writes:
| | Please implement standard deviation.
By eliminating the (small number)% cheapest and (small number) % most expensively
priced lots, the average price index would be "more statistically significant"
in scientific terms ( less error margin ).
This is only necessary for items that have at least e.g. 10 distinct lots for
sale.
Why this suggestion ?
Yesterday i noticed one particular seller set all his prices at ( BL avg x 10000
). This caused average price index "lots for sale" to be inflated at least
tenfold.
Someone else suggested he made a mistake in currency.
If this one shopkeeper persists, unwanted "extreme" inflation may occur on several
very common items, and those lots - for sellers not paying attention when adding
to their inventory - might become impossible to sell.
|
Rather than excluding data points from the calculation of the mean, why not add
a median to the price guide? The median is much more resilient to outliers like
the example you gave. Both the mean and the median provide useful information,
especially when they are different.
|
|
|
|
|
|
Belgium
Canada, Ontario
USA, Washington
United Kingdom, England
Netherlands, Noord-Holland
