You might all be getting bored with this
though the hits on the website seem to be increasingly rapidly!
I have spent a few days recently (with a break for Christmas
) gathering some back data - the ebay API only allows you to query c. 30 - 90 days of historic data automatically - but you can see historic data on their website for up to c. 2 years - so I manually added the older item numbers from relevant sales and then ran a series of queries against the API -the result was to increase data from 97 items to 681 items
NB my apologies - no pagination yet - on the task list - so for now you will get 674 results on the finished page (however most don't have photos yet, so not too slow to load)
This means that there is a lot more data in there now to start seeing patterns / trends and we can see nearly two years of ebay sales... some quick points:
I need to run some separate queries to pick up sale photos where available - in hand, but not there yet...
- the vast majority of sales are on UK ebay, of the 681 sales:
UK = 650 (95%)
USA = 25 (4%)
Australia = 4 (0.5%)
Canada = 1 (<0.5%)
Poland = 1 (<0.5%)
Because of this - I am not doing anything yet on currency exchange (too complex to try and pull in historical exchange rates) - when you see the USA as only 4% of results - and therefore the significance of the currency means an accuracy of only 1% / 2% I am currently treating $ as £
as I am starting to look at indicative trends I think this is fine... I have the raw data stored, so we can always make it mroe accurate later...
I have added an average price on the finished screen (obviously on the live page it isn't really relevant as the sales are constantly changing...) This averages price on all items on the page, whether everything, or based on a search...
So... searching on c15 for example gives you 16 items (1 not sold) with an average price of £188.95
put into a chart you get:
this would suggest an initial drop and then a gradual climb in value...
of course there are lots of factors that influence price - from condition to alligator straps / how many are sold in one time period to time of year or life cycle of the watch...
but the rough suggestion fits with perhaps a pattern of a drop in value during the watch lifetime - with a gradual climb in value afterwards... if this is a true pattern and went across lots of models then that would start to suggest that the watches are a good investment (therefore we can justify buying more
)
it will be interesting to start to plot these and see which are good / bad investments...
I hope to start automating charts over the coming weeks... now that we have a more reasonable bank of information it makes sense to start analysing it...
any thoughts / questions, do let me know
Alasdair