How much did you pay for your used 996TT?
Registered User
Joined: Sep 2004
Posts: 1,740
From: San Antonio
Rep Power: 127
I got the data and did a regression to see contributors to price. Recoded some of the variables to make a model that makes sense to me. An astonishing 88% of the variance in price was accounted for by the following variables:
1) Miles - 26 cents per mile discount
2) x50 option (0/1 variable) - $4300 premium
3) Age - how old in years the car was at time of purchase (an approximation, since model year and transaction year was rounded) - $4100 decline per year old (above and beyond the mileage)
4) cab (cabriolet) - $2500 premium but not statistically significant (fewer transactions, so an unclear relationship)
5) war (warranty) - $2500 premium
6) crisis (sales after August 2007 when credit became tight) - $8800 immediate drop.
One can calculate whether they beat the average on their own purchase by applying the modeled formula to their case. For example on my purchase:
2002 X50
$45,000
March 2009 (implying an age of 7 and crisis = yes)
52000 miles
Warranty
$93039 - $.269 * 52000 - 7 * $4127 + $4292 + $2527 - $8838 = $48143
So, I beat the average by $3143.
1) Miles - 26 cents per mile discount
2) x50 option (0/1 variable) - $4300 premium
3) Age - how old in years the car was at time of purchase (an approximation, since model year and transaction year was rounded) - $4100 decline per year old (above and beyond the mileage)
4) cab (cabriolet) - $2500 premium but not statistically significant (fewer transactions, so an unclear relationship)
5) war (warranty) - $2500 premium
6) crisis (sales after August 2007 when credit became tight) - $8800 immediate drop.
One can calculate whether they beat the average on their own purchase by applying the modeled formula to their case. For example on my purchase:
2002 X50
$45,000
March 2009 (implying an age of 7 and crisis = yes)
52000 miles
Warranty
$93039 - $.269 * 52000 - 7 * $4127 + $4292 + $2527 - $8838 = $48143
So, I beat the average by $3143.
Based on this formula, I beat the average by $30k - something is amiss. Either I'm using the wrong starting point or I'm confused.....
Registered User
Joined: Mar 2009
Posts: 95
From: Los Angeles
Rep Power: 21
You have to start with the constant in the model. It's the "cons" in the picture. Then you add or subtract accordingly. It has no relationship to MSRP. And really the model won't apply to new cars (since the depreciation curve is non-linear--more in the beginning and less at year 8 or 9).
Last edited by SAVED4IT; Dec 26, 2009 at 02:39 PM.
Thread Starter
|
Registered User
Joined: Apr 2008
Posts: 472
From: Columbus, OH
Rep Power: 57
Great calculation, but where did you get your starting figure of $93,039? Your MSRP had to be way more than that - mine is over $115k and it's pretty stripped.
Based on this formula, I beat the average by $30k - something is amiss. Either I'm using the wrong starting point or I'm confused.....
Based on this formula, I beat the average by $30k - something is amiss. Either I'm using the wrong starting point or I'm confused.....
Registered User
Joined: Sep 2004
Posts: 1,740
From: San Antonio
Rep Power: 127
Gotcha - so instead of feeling like I got the steal of the century at $30k below average, I actually just did really well at $4k below average. 
However the numbers fall out, I couldn't be happier.
However the numbers fall out, I couldn't be happier.
Thanks for the data and the chart. I am currently in the market for a 996TT and the data is a valuable reference tool in determining the fair market value of the cars I am considering.
Thread Starter
|
Registered User
Joined: Apr 2008
Posts: 472
From: Columbus, OH
Rep Power: 57
Good luck with your search!
TK
Registered User
Joined: Dec 2009
Posts: 266
From: SW Missouri
Rep Power: 42
73,900 (original sticker 145k+)
6800 miles
2005 Turbo S Cab in Midnite Blue
3yr warranty since it was out on years
drove it home on Dec 4th 2009
link to original ad
http://porschedallas.parkplacetexas....ition_id=10426
6800 miles
2005 Turbo S Cab in Midnite Blue
3yr warranty since it was out on years
drove it home on Dec 4th 2009
link to original ad
http://porschedallas.parkplacetexas....ition_id=10426
Last edited by mtcraigco; Dec 26, 2009 at 05:10 PM.
Registered User
Joined: Dec 2009
Posts: 266
From: SW Missouri
Rep Power: 42
That is frigging UNCANNY!
But the market is math I suppose and math can be used to predict anything if you have the variables eh?
Registered User
Joined: Aug 2008
Posts: 158
From: La cresenta
Rep Power: 26
Well with any luck I will be adding my info to the list Looking Now fro the right 996TT. If anyone see's a great deal PM me heeltoe914@aol.com
Thread Starter
|
Registered User
Joined: Apr 2008
Posts: 472
From: Columbus, OH
Rep Power: 57
Well with any luck I will be adding my info to the list Looking Now fro the right 996TT. If anyone see's a great deal PM me heeltoe914@aol.com
https://www.6speedonline.com/forums/...iles-ohio.html
Good luck with your search, TK
Registered User
Joined: Feb 2007
Posts: 389
From: san antonio, tx
Rep Power: 35
I got the data and did a regression to see contributors to price. Recoded some of the variables to make a model that makes sense to me. An astonishing 83% of the variance in price was accounted for by the following variables (although it's a small data set, I know):
1) Miles - 26 cents per mile discount
2) x50 option (0/1 variable) - $4300 premium
3) Age - how old in years the car was at time of purchase (an approximation, since model year and transaction year was rounded) - $4100 decline per year old (above and beyond the mileage)
4) cab (cabriolet) - $2500 premium but not statistically significant (fewer transactions, so an unclear relationship)
5) war (warranty) - $2500 premium
6) crisis (sales after August 2007 when credit became tight) - $8800 immediate drop.
One can calculate whether they beat the average on their own purchase by applying the modeled formula to their case. For example on my purchase:
2002 X50
$45,000
March 2009 (implying an age of 7 and crisis = yes)
52000 miles
Warranty
$93039 - $.269 * 52000 - 7 * $4127 + $4292 + $2527 - $8838 = $48143
So, I beat the average by $3143.
1) Miles - 26 cents per mile discount
2) x50 option (0/1 variable) - $4300 premium
3) Age - how old in years the car was at time of purchase (an approximation, since model year and transaction year was rounded) - $4100 decline per year old (above and beyond the mileage)
4) cab (cabriolet) - $2500 premium but not statistically significant (fewer transactions, so an unclear relationship)
5) war (warranty) - $2500 premium
6) crisis (sales after August 2007 when credit became tight) - $8800 immediate drop.
One can calculate whether they beat the average on their own purchase by applying the modeled formula to their case. For example on my purchase:
2002 X50
$45,000
March 2009 (implying an age of 7 and crisis = yes)
52000 miles
Warranty
$93039 - $.269 * 52000 - 7 * $4127 + $4292 + $2527 - $8838 = $48143
So, I beat the average by $3143.
Once I get the data, I'll also try to include quadratic effects of age and miles.
Last edited by PeterK; Dec 28, 2009 at 06:15 PM.
Registered User
Joined: Mar 2009
Posts: 95
From: Los Angeles
Rep Power: 21
I didn't bother looking for higher-order relationships or interactions, since the R2 was so high. If I was doing it professionally I would have explored those things just in case. Good thoughts.
Registered User
Joined: Feb 2007
Posts: 389
From: san antonio, tx
Rep Power: 35
The high R2 values could perhaps be artefactual because of collinearity among the independent variables. One way to test this would be by using an orthogonalization. But this would, of course, have made the model very unfriendly.