Used E9x M3 Price Modeling ('08-'12)

[overdose]

For the past couple months I've been looking to buy an E92 M3 (new or used). I quickly found there was no single resource people could consult for E9x prices. Since I work in a data intensive field I thought it'd be fun to create a very crude formula to help members see how their offers compared with those before them.

My objective for this thread is to help other members make a smart purchasing decision for a USED E9x M3 with a simple equation. Also, at the very end of this post I'll share some interesting insights about M3post.com users that I found.

ABOUT THE DATA
There is a total of 166 data points and they're all pulled from this post...how much did you pay for your used m3? The first purchasing date goes back to December 2008 and the most recent in July 2012. I collected model years, model type, transmission type, miles at time of purchase, exterior color and of course price.

Even a small survey like this can get pretty complicated so for the sake of simplicity I excluded a bunch of variables (options/packages, negotiation skills of the individual, location of purchase, taxes, etc). I also excluded people who couldn't follow directions (new cars, older models, no transmission listed, etc). Also, for those who posted Canadian dollars I converted them to USD on 7/16/2012 without adjusting for inflation or historical currency exchange rates.

Lastly, it's important to note this was a voluntary submission from M3post members so it may not represent the entire pricing population.

THE EQUATION
If you remember basic algebra then this should be pretty simple for you. Here's the pricing equation with x being the number of miles on the car:

ESTIMATED PRICE = -0.3001x + 54146

Thus, an M3 with 20,000 miles on it would be priced at approximately $48,144. And since the standard error is 5472, a very lucky person could pay as low as $42,672. On the other hand, someone who doesn't care about negotiating (or lives in an area with a higher standard of living...or it includes all the possible packages...or etc) could pay as much as $53,616.

Of course, if you don't want to do the math you can just look it up in the image below.

To those who purchased and didn't participate in the discussion...does the equation work out?

All that being said and done, I'd like to point out that the M3post community is what makes this forum thrive. If you have a question about the price of your specific offer, go ahead and ask - don't be afraid! This formula is far from being gospel as it doesn't account for a TON of other factors.

INTERESTING TIDBITS
In my opinion, this is the best part as it gives a glimpse into the M3post community. Without further ado, out of 166 members who bought a used E9x M3...

Model Year
2008 = 64%
2009 = 21%
2010 = 6%
2011 = 9%

Model Type
E90 = 38%
E92 = 51%
E93 = 11%

Transmission Type
6MT = 52%
DCT = 48%

Exterior Color
No Color Specified = 27%
Alpine White = 20%
Jerez Black = 17%
Jet Black = 10%
Space Gray = 7%
Interlagos Blue = 5%
Silverstone = 4%
Melbourne Red = 4%
Sparkling Graphite = 4%
Dakar Yellow = 1%
LeMans Blue = 1%

I'll let you make your own conclusions on what this says about our members haha.

ACKNOWLEDGEMENTS
Thank you to the 166 members who posted their purchasing information over the past 2 years. This survey obviously would not exist if it weren't for you all.

FINAL THOUGHTS
I love doing this sort of geeky stuff so if you'd like to see another one on a larger scale that includes options, packages and whatever else just let me know (granted there is enough demand).

Finally, if you want me to narrow down an equation for super specific options (2008, E92 AW, 6MT, 12k miles) then just PM me.

[dmurray14]

Nice! I'd be curious to see it split out by model year - mostly cause I'm interested to see how they hold their value. Next up, calculate the impact the release of the e92 had on the last model year of the e46 m3s

[dagolfpro]

OD, love the data.

[DEV M3]

Great data! I'd like to see how the prices change over time... Like what were 08's going for a few years ago vs now... And project those prices into the future

[Xtopherus]

Nicely done! It would be interesting to add a third dimension to the regression (age), but that appears to make things a lot more difficult and it's difficult to reliably collect from the available data anyway.

It is rather interesting to note that each mile takes 30¢ off the value of the vehicle.

[///Machine]

Great work!! I'll be in the market for an M3 in 2013 (used or new TBD) and this is very insughtful. I probably will have to scan through all the pages in that thread anyways - focused on year, body style, options, etc. But this is a great start.

Out of curiosity, did you get a significant R2? When you include the other variables, does the R2 improves significantly? On the more detailed analysis did you look at E90/92/93 separately?

I would definitely be interested in the results of the more detailed analysis. Should I pm you or you plan on posting some time?

I wish this thread gets much more views, because it really deserves.

[trendy26]

Great Analysis! Out of curiousity, what do you do for a living?

[overdose]

Originally Posted by dmurray14

I'd be curious to see it split out by model year - mostly cause I'm interested to see how they hold their value.

Originally Posted by DEV M3

I'd like to see how the prices change over time... Like what were 08's going for a few years ago vs now...

Ask and you shall receive! As you may have guessed from the image below, the data is a little choppy so I'd take the downward trend with a super fine grain of salt (and also because they are averages). But if you really need a hard data point I'd look only at MY2008 since that is 64% of all observations.

[overdose]

Thank you for all the props so far - I hope you guys enjoy it as much as I loved analyzing it!

Originally Posted by Xtopherus

It would be interesting to add a third dimension to the regression (age), but that appears to make things a lot more difficult and it's difficult to reliably collect from the available data anyway.

Yes, that would be very interesting and if there's enough demand for it maybe we'll do it despite the inherent difficulties.

Originally Posted by ///Machine

Out of curiosity, did you get a significant R2? When you include the other variables, does the R2 improves significantly? On the more detailed analysis did you look at E90/92/93 separately?

I would definitely be interested in the results of the more detailed analysis. Should I pm you or you plan on posting some time?

Very good question but no I did not - R2 was 0.374. In spite of this, I shared it anyways because the model predicted nearly 40% of the prices using just two variables so I thought that was "good enough" for a non-scientific study.

RE: Detailed analysis - I definitely plan on posting more over time depending on the demand. Also, since the data set is so small I can only slice and dice it so many ways. Was there something you were interested in?

Originally Posted by trendy26

Out of curiousity, what do you do for a living?

I work in the Interactive Marketing department for Macys.com. I attempt to find trends in our data and see how users behave on our website to make them buy more stuff

[SamS]

Incredible work!!

[Z K]

So based on your findings can you predict what the price of the car will fall to over time? I assume the delta in pricing will decrease as the car gets over eventually settling on a plateau as the car gets older.

[dmurray14]

Originally Posted by overdose

Ask and you shall receive! As you may have guessed from the image below, the data is a little choppy so I'd take the downward trend with a super fine grain of salt (and also because they are averages). But if you really need a hard data point I'd look only at MY2008 since that is 64% of all observations.

Awesome! Thank you!!

[PrimoM3]

Check out the big brain on Brad... Nice work though, seriously.

[Lay-Z M3]

Originally Posted by overdose

I work in the Interactive Marketing department for Macys.com. I attempt to find trends in our data and see how users behave on our website to make them buy more stuff

Well I'm convinced (esp as you are based in San Fran), that you are Pete (really based on Paul DePodesta) who created the record breaking 20 straight winning team for GM Billy Beane at the Oakland A's 10 years ago!!! lol

You are our Moneyball!!!

[m3buckeye]

Nice work, son.

[chrisjrobi]

nice work. Thanks for putting this up.

[cmpe]

Thanks for this.

Can you share the spreadsheet as well?

[overdose]

Originally Posted by Lay-Z M3

Well I'm convinced (esp as you are based in San Fran), that you are Pete (really based on Paul DePodesta) who created the record breaking 20 straight winning team for GM Billy Beane at the Oakland A's 10 years ago!!! lol

You are our Moneyball!!!

LOL - thank you for the compliment but that might be an over statement.

Originally Posted by cmpe

Thanks for this.

Can you share the spreadsheet as well?

Absolutely - PM me your email and I'll forward you the raw data.

[overdose]

Just wanted to share with everyone a recent exchange I had with a fellow M3post member...this was about one week ago.

trex858: Really cool study you compiled! I am looking at buying a used 2009 and was wondering if you can run the numbers for me? 2009, Space gray, DCT, 37k miles.

overdose: Thanks for your PM and comment! I'd be glad to run the numbers but I'm obligated to let you know that the sample set for MY2009 is very small - only 35. Again, take it with a grain of salt and use it as a benchmark.

That being said, the equation for your specific request is...
Y = -0.2706x + 56024 with a standard error of 4409.

That means your M3 with 37k miles would estimate to around $46,012. A really good deal would be at $41,603 and a very expensive one would be $50,421 or higher.

trex858: Awesome, thank you. The dealer wants $46,998 and $42,000 is exactly what i want to pay for it so from what i can see your numbers are on target based on the current market.

[jonbig12]

I've been lurking for a while and had to register when I saw this just to say thanks! No more "Is this a good deal" threads for me. I was considering doing something similar, but would never have been able to do this good of a job. Kudos.

[dreamspeed]

Nice work! And it's probably too much work but any chance on seeing this data split by location? I'm looking into buying a used E92 soon and would be interested in seeing the average cost in the Los Angeles area. Or if a nearby state has substantially lower costs it might be worth it to check it out.

[mdosu]

sticky this!

thanks, you need to somehow normalize option costs