cbzzoom

So I wrote a program to extract the prices from cars.com

These are asking prices. They are only reflective of actual sale prices in the way that asking prices ever reflect final sale prices.

I listed only cars from the last 30 days to avoid cars that have been sitting for sale forever (though sellers can easily fool this by taking the car off and relisting it).

This listing is for manual 6spd coupes only.

I excluded GT3, GT2, and Turbo, but this sometimes fails if the listing does not describe the car correctly.

By engine size :



2006 : median= 49800.00 mean= 49037.67 sdev= 5559.33
2007 : median= 51990.00 mean= 59733.09 sdev= 16373.28
2008 : median= 59988.00 mean= 58150.83 sdev= 4548.13
2009 : median= 69588.00 mean= 70271.00 sdev= 12145.57



2006 : median= 56900.00 mean= 55829.26 sdev= 4804.42
2007 : median= 61991.00 mean= 61161.85 sdev= 5069.27
2008 : median= 67981.00 mean= 68558.00 sdev= 4532.27
2009 : median= 79880.00 mean= 79492.58 sdev= 6171.92

cbzzoom

BTW those crazy high price spikes on the 3.6 are in fact turbos. The 3.8's are much easier to extract correctly.

Fortunately the vast majority of the listings on cars.com seem to correctly identify the type of car in the description or title. There appear to only be 2 or 3 bad listings out of the whole set.

Autotrader data is not nearly so clean.

One thing that I found quite surprising already just from glancing at the data is that price falloff with mileage is much lower than expected.

Galvano

I was told there would be no math. Just kidding,very nice work. I'd be curious to see the drop off in price over time as the new model is introduced. I have a feeling that the price shifts won't be uniform over the years or with varying mileage. Buying opportunities will be there for those who are patient.

winkingchef

Mmm data tasty.
Yum.
Thank you.

Spectro28

+1

And +1 on the surprising lack of influence of mileage on asking price.

Nice work, a script that can be run periodically, I presume, which warrants a "Sticky" thread.

pewter82

06 S cars ? seems high at 56k for a listing price

winkingchef

He's not factoring out Cabrios and Targas and options which can be up to 10k more.

Romulus90

I hope this isn't OT by too far.. How much did 996 prices drop when the 997 was introduced, if anyone recalls?


Thanks for the data.. very interesting

cbzzoom

Cabrios are filtered but you're right Targas, 4S, x51 , etc. are not.

I would really be interesting to be able to compare actual sale prices to asking prices, but I don't know how to get that data. Maybe from DMV sales tax records you could figure out actual sale records? I wonder if that info is public record...

Spectro28

Hi all --

Charles was gracious and shared the raw data, and I made a couple of charts of it.

Have a look at this analysis of the price distributions for the different models, by mileage, colored for model-year:


and this more simple analysis, without separating cars by mileage (these are really all pretty low mileage cars):


(ok, I'm really just working to better learn the nuances of a data visualization environment I like, so hence the dorky analysis).

Cheers!
Marc

autobahnA8

cool! thanks.