4 ms·

Hi HN. I really want to buy a Golf GTE, but was worried that they were still depreciating quite heavily. Looking around the web, I couldn't see a way of trackin

Hi HN. I really want to buy a Golf GTE, but was worried that they were still depreciating quite heavily. Looking around the web, I couldn't see a way of tracking the price of a used car over time. I'd bookmarked the search page on AutoTrader, and regularly went back to it to see if they were any cheaper. Low and behold, the cars were depreciating, but it was a drag having to go back to the site each time. I was too lazy to put the prices and dates in a spreadsheet. Writing this little app seemed like way less work (it wasn't, but whatever). I thought it might be useful to other people, so I bought a domain and here we are. Would love feedback! Would this work in the US? What are the large car sites in the US?

vmateixeira 10y ago
Nice app, good work! It would be nice if the price/mileage graphs could have a label informing what each axis means as they can have similar values at times.
mewwts 10y ago
Very cool tool. I work on similar stuff, and I wrote this blog post last year. http://bytes.schibsted.com/price-car-data/ http://bytes.schibsted.com/price-car-data/ With time most models will see an exponential decay, according to my analysis. But some will, of course, decay faster than others.
- adamqureshi 10y ago
  I wanna build a tool like this for my marketplace idea. http://onlyusedtesla.com/ http://onlyusedtesla.com/ i have 47 listings and growing weekly. Would love to offer some useful metrics to my users.
vmarsy 10y ago
Awesome, I was thinking about the same thing, as during some car shopping I noticed the 2014 Volkswagen CC, I liked it but I was wondering why its price dropped so low compared to other vehicles from 2014 who had a similar starting price. The website I use to look up prices is kbb.com
- coredog64 10y ago
  At one point in time KBB would sell their data. While it may have been correct, format-wise it was hot garbage. Packed fields, arbitrary format changes from MY to MY, and inconsistencies in accessory codes.
breckenedge 10y ago
AutoTrader is popular in the US for cars, and they aggregate some dealer's listings. This would work in the US, but prices for different types vary widely based on region. Using a Metropolitan Statistical Area would help. I once wrote something sorta like this for motorcycles, which are easier to compare due to fewer model options. My goal was to identify what brands or models held their value well (Hondas and Harleys mostly) and to automatically alert when one fell below the curve by X%. I never got much out of it, but it was a fun project.

vmateixeira 10y ago

Nice app, good work! It would be nice if the price/mileage graphs could have a label informing what each axis means as they can have similar values at times.

mewwts 10y ago

Very cool tool. I work on similar stuff, and I wrote this blog post last year. http://bytes.schibsted.com/price-car-data/ http://bytes.schibsted.com/price-car-data/ With time most models will see an exponential decay, according to my analysis. But some will, of course, decay faster than others.

adamqureshi 10y ago
I wanna build a tool like this for my marketplace idea. http://onlyusedtesla.com/ http://onlyusedtesla.com/ i have 47 listings and growing weekly. Would love to offer some useful metrics to my users.

vmarsy 10y ago

Awesome, I was thinking about the same thing, as during some car shopping I noticed the 2014 Volkswagen CC, I liked it but I was wondering why its price dropped so low compared to other vehicles from 2014 who had a similar starting price. The website I use to look up prices is kbb.com

coredog64 10y ago
At one point in time KBB would sell their data. While it may have been correct, format-wise it was hot garbage. Packed fields, arbitrary format changes from MY to MY, and inconsistencies in accessory codes.

breckenedge 10y ago

AutoTrader is popular in the US for cars, and they aggregate some dealer's listings. This would work in the US, but prices for different types vary widely based on region. Using a Metropolitan Statistical Area would help. I once wrote something sorta like this for motorcycles, which are easier to compare due to fewer model options. My goal was to identify what brands or models held their value well (Hondas and Harleys mostly) and to automatically alert when one fell below the curve by X%. I never got much out of it, but it was a fun project.