Hello!
To use the forum, login or register above.
(If you are already logged in above, please click here)
Better formula to calculate "current average"
would be (Mean + Median) divided by 2
This would **help** reduce the impact of errors in pricing (of which there many) and give a better representation of changing prices of past averages vs. current averages
This would **help** reduce the impact of errors in pricing (of which there many) and give a better representation of changing prices of past averages vs. current averages
This discussion has been closed.
Powered by Vanilla
Comments
Some stores I've noticed, for example, have ridiculous prices on many items compared to the vast majority and they throw off the true average - and may cause everyone who updates their prices regularly based on the averages to artificially inflate market value. Other stores seem to not catch up to prices and occasionally under-price themselves dramatically on some things. Both these scenarios skew the true average.
For example, let's say 10 stores have 100 of an item and all of them have it selling for $0.15 and 1 of the stores has 1,000 of it and listed for $3.00. Ridiculous, right? Nobody will buy that part from that store. Well, some of the stores (say, 5) who might update their prices blindly every 1-2 months (not spot-checking but taking the average) will end up listing their part for $0.44. Now you have 4 stores at $0.15, 5 stores at $0.44 and 1 store at $3.00. The next month, 3 of the 5 stores do another blind update based on the average and their prices get set to $0.58. Now the listings look like:
$0.15
$0.15
$0.15
$0.15
$0.44
$0.44
$0.58
$0.58
$0.58
$3.00
Then the stores who spot-check instead of take the average take another look and feel like their prices are under market. They tick their listings up to a little under $0.44 to stay competitive and the new prices look like:
$0.31
$0.35
$0.43
$0.43
$0.44
$0.44
$0.58
$0.58
$0.58
$3.00
So then the store who had it at $3.00 drops the price to $0.30, is the lowest seller, and corners the market in selling his 1,000 pieces for double what he would have 2 months prior - just for instigating artificial price manipulation.
Does this really happen? Maybe not a lot but I do think it does. So the average should throw out the extremes on either end and then do the mean of the remaining and use an algorithm that accounts for past recent sales, if any.
http://www.brickowl.com/catalog/lego-white-gear-rack-4-3743
There's no way pricing like that is reliable on any level, it is skewed both on the low end (for stores who rarely update their pricing) and skewed on the high end which causes gradual artificial uptick. Which just results in few stores making sales on any of the items while some stores, when they do sell, get it sold at artificially inflated pricing. When people make their buying decisions based on a combination of price, shipping for total order and count of available pieces from a store - in the part I linked to the lowest-priced stores will almost never move the item and the mid-tier ones with more variety will be able to sell the item at inflated prices. I understand this particular part may be more on the rare side but it's a great example of artificial price manipulation that happens when just 1 store lists things for ridiculous amounts and just sits on that price for months.
As far as i can see you see it sometimes happen on the other site but the advantage is that you have the raw data so you can get rid of the outliners etc etc and make a good estimate. Athough this alot of manual labor !
on that why does anyone buy this part? I know it's OOP but...
http://www.brickowl.com/catalog/lego-black-windscreen-3-x-6-x-1-curved-62360
it comes from a sub-par Ninjago set, of which I bought "quite a few" at a good price, all of a sudden a run on it - so I take a look at my price
and yes I have sold at that price!
@pwpeter Where is the raw data? In a quick and easy to read form - I don't access the API because I haven't a clue how to use it - I have to rely on "averages" as a start point. As mostly a seller, I need to catch OOP parts and price accordingly, not only for my own benefit but also to catch other sellers buying out my rare parts and reselling at a higher price - which happened to that example before I caught it!
"Current average" is the only way to achieve this, and that is useless "as-is" due to mostly mistakes with the decimal point (I think)
Hence my suggestion of Mean + Median ÷ 2
@JayB-SoCal
I'm not seeing the problem...
For starters, selling prices can be set 'individually' (by each seller), 'averages are just what they are: 'averages', regardless whether they are 'selling' averages or 'listed for sale' averages, and in no way 'all of them' are 'errors', some are very deliberate (I know as I apply such on 'some' items).
This debate is an endless debate, and therefor completely useless: how much do you take off? One will say top 3 (lowest and highest), another will say 5, another will say 10. So what's the point ????
Particulary as this is a 'free market', any deliberate 'force change' of any data could/would/should be considered 'influencing' the market flow, and as far as I know that is 'punishable' by law, so fat chance any site (regardless whether BO or BL) would/should/could play such 'game'.
@leopard37 is also quite correct: however and 'how hard' one would try to 'influence' pricedata, there would still be some form of error, as 'purchase' or 'listing' prices are not just related to the items themselves, they are in straight relation to many other factors...
I'll give you an example:
Recently I sold a stickersheet at 2x average, so I investigated the 'why'.
The cheapest seller had 'several', yet their minimum was about $50> probably a big 'no go' for the buyer, they just wanted to buy 'that', other sellers where in between 'that one' and 'me', but some where located 'abroad' (higher shipping and/or potential import taxes). then the 'combination': My buyer bought another few items, was I a 'perfect match' for the buyer? Dunno, as I don't know the starting point from the buyers perspective, was it that sheet listed at 'average' and the 'high rated sheet allowed him to reach minimum, or was it that sheet rated x2, getting along a few things at 'average', to make up for my 'minimum', was it my low shipping cost compared to others, was it the 'total picture' of the purchase that made the buyer chose for me, was it my 'availability' was it 'match' to the buyers 'whishlist'? Was it the S&H that made a difference? Was it the location (buyer wanting it 'fast')? Was it my 'credebility' (feedbacks) as a seller that 'helped' the buyer choose?
All of these (and many more) very different elements, have buyers 'decide' on how much they are willing to spend on 'something', who are we, or why would any site feel itself 'obligated' to determine they need to 'skip off' a bunch of 'excess' prices..
It is just what it is: 'averages', and those 'low' and 'high' excesses are part of the pricing history, so no reason to change 'anything' (and again, it would be against many economical laws)
One would only be able to inflate 'listed for sale', not 'sold' that way and 99% of sellers use '6 mth sold average' as guideline for listing stuff, rarely ' averages listed' as they would end up in the higher prices and not be 'selected' by buyers > they would shoot themselves in the foot.
So even if a few sellers would do so deliberatly, averaged out against hundreds on thousands of sellers and their sales it is simply 'neglectable' and will hardly make a difference in 'averages sold'.
Besides, do you really think those sellers would have the time to check the thousands of lot's in their store on frequent bases just to figure out how they can influence the market? That's hilarious.
No, there are just stores who sell at higher prices either some items, either their entire store, if it works for them, fine, but against the hundreds of other sales, it is not going to make a difference.
Example:
Black Brick 3001 new.
On BL: 193000 sold in 6 months
Lowest price sold 0.01, highest 0.95 , average 0.16
On BO: 9200 sold in 6 months (quite a difference in numbers)
Lowest price sold 0.05, highest 0.63, average 0.17 (but also averaged 0.16 if you take BO's 2 years ).
So even when you have low or high excesses, very high quantities sold or much lower quantities sold compared to it, the 'average' over a longer period simply hardly changes, because those few excesses are neglectable compared to the hundreds or thousands of sales of more 'averaged' prices.
As BO also provides data on 1 year and 2 years on top of the '6 month' average also proves the price is pretty stable (keep in mind the numbers are rounded to 2 digit's so 0.164 would be 0.16 and 0.167 would be 0.17, allthough the difference is only 0.003, after the 'rounding' it's 0.01) and cannot be influenced by a few excesses.
Yes there is 1 seller with an 'outraged' price, but I doubt it is with the intention to manipulate the market, his whole store is 'expensive' and as far as I can tell he's not 'playing' the game you're claiming be played.
As a matter of fact, he's listing is not doing a thing: check the sold averages:
Over 2 years (I'm seeing it in Euro) it's 0.34, over 1 year it's 0.39, last 6 months it's 0.38, so hardly any influence, as the 'sold' price is pretty stable, and has even 'dropped' the past six months.
If I take the BL data then it's comparable (but quantity sold 10x that much): 0.35 on average for used and 'listed for sale' average 0.41 (including that outraged price). Now if you check his listing there, it was listed on november 2013, if you check the production data on that item, you will see there is a bit of a 'gap' between 2010 and 2013, so the part was fairly rare in 2013, but then rereleased in higher quantity in 2016. So while he was aiming further 'rareness' as of 2013, the availability has 'risen' since then, so he's pricing is outdated, which proves he's not even thinking of 'adjusting' his price to manipulate the market, it's just his way of pricing at moment of listing and never looking back... And clearly it doesn't make a difference on BO (dunno when he started syncing), as besides influencing the average of 'listed' there is no influence on the 'sold', neither on the newly (used) listed 'for sale', and you gave the key reason for that: Brickstock, as yes a lot of sellers use Brickstock to price, but Brickstock uses BL data (higher quantities sold, so even beter data), not BO data. As there are so many sellers using it and 'syncing', the price remains pretty stable on 'listed for sale'. OK, a slight difference on averages sold, but that is caused by other reasons, which are pointless to include.
In general prices tend to drift to the bottom, and although that might feel good for buyers it's not good long term. So having things selling at higher prices too helps in keeping the market viable. In the long run I think that is good for everybody.
Yes, I know one store at least that seem to price well above everybody else. I wonder how they sell anything but maybe their local market is such that it works. Good for them. Overall I worry more about stores having 70% off sales than if somebody is a bit above average in price. At 70% off you most likely sell at a loss and it contributes to the drive to the bottom.
There was no misunderstanding in what you said earlier (at least not from me) :)
Prices are pretty stable, when I look at old data on basic parts it's still about the same.
Yep, there are some stores with very high prices that manage to survive, offcourse their orders generate high returns, so they have less orders but more benefit, I'm not sure it's gonna keep working...
Huge sales (with starting point on 6 months average) are 'temporary' (short term) influences, yes those sellers would be selling near break even or even at loss ( but remember they already made benefit on other items), but it is up to you to determine whether you wanna play along with it, or just think twice and do *your* thing... In the long run, you'll be better off ;)