Imagine if 6500 trade-ins came to your workplace every single week.
You would be a rich man, right? Not quite. At the moment. you are a poorly paid security guard at this facility by the name of James. Let’s call you Bill James for now. Since it’s Bill the former security guard who is the inspiration for creating a database that will make sense out of all that data.
Click on the above database link for a moment. Then get a little background on Bill James as well. I will need your assistance to make this work.
Now consider this small foundation of numbers we have at the moment. Click away at all the information after you read this article because there is a lot that can be enjoyed by us auto enthusiasts.
What we have for the most part is a large ‘neutral’ database where no franchised dealership can tilt the inventory in favor of a particular brand. 300,000 vehicles a year will be evaluated every year under the following criteria.
- Age
- Mileage
- Mechanical condition
- Title History
- Brand
- Model
For right now, we have limited the information to a few basic questions. 
What car brands and models are usually kept for the long haul?
Which ones last over 180k and have minimal engine or transmission issues? Conversely, which ones are curbed before 120k miles specifically because they have powertrain defects?
Which cars and brands are worth a premium reputation? Which ones are rolling myths or monstrosities?
These are not easy questions and I won’t pretend that this database will furnish all the answers. We will add more data as time goes on. By the time 2015 rolls around, we’ll likely see over a million vehicles in this database which will be offered as a free service for auto enthusiasts, industry professionals and academia.
So for right now please take a visit to the database here and let us know what you think in the comments section.
Very interesting I don’t remember the quick analysis from the other day. What a piece of garbage that Freelander must be. Always wondered what the most common single year model was, ’02 Explorer makes sense.
I think Paul should hire a summer intern to dumb down the data to a few Powerpoint slides. The commentariat can think of the good questions. My contributions…
Most common kinds of repairs needed, overall and specific or a component like transmission.
Top 5 models with the least interior issues (repeat for x, y issues). Maybe show this at the 3 year old, 5 year old and 10 year old points.
Top 5 models with the most interior issues (repeat for x, y, z issues)
Correlation between residual value and a repair issue (if any). Do certain mechanical/cosmetic woes hurt the resale price more than others?
Model or brand with most improvement over time, i.e. the defect rate versus the average is improving the most
Model or brand with most degradation over time
Which brands or models “age” the worst? Which age the best? I guess that would depend on how we define age.
Model year that brand (or model) x went to hell
Correlation between mileage and certain types of repairs need
It would be interesting to focus on durability issues (e.g. Achilles heel by brand or model) because otherwise the rankings might not differ from a JD Power study for new vehicles. Freelander bad when new, bad when old.
I suppose it would be extremely challenging, but weighting results based on percentages of trade-ins that are over a particular age would be more meaningful if they were indexed against sales by year. Of course Oldsmobile, Geo, and Plymouth trade-ins are old. There haven’t been any new ones for years. I’m dubious that this somehow makes a 1998 Plymouth Neon a safer bet than a 1998 Dodge Neon. Conversely, this makes newer brands look bad. Scion is rated extremely low because they’re a newish brand even though none of their individual models are particularly bad. Same with Hummer, for that matter.
Steve – very interesting idea. #1Son is 16 will be looking for cheap wheels (as soon as he earns his Eagle Scout!). Do you picture this as becoming something analogous to TrueDelta?
@ CJ – I was also thinking that the dead brands may have too big of an effect on the average — especially in the “ratio over 18 years”. Steve, you might consider using the median instead of the average (or in addition to it). The median is much less affected by outliers like Plymouth and Olds.
Would it be possible to lump Plymouths in with Dodge and Oldsmobiles with Buick?
The quality ratings for pickup trucks were interesting. I didn’t expect to see the Chevy C/K at the top. Given the incredible rep that Toyota trucks have out here, I was expecting to see them at the top.
I wonder why #2 after the C/K is “Nissan pickup” but the Frontier and Titan are both in the negative column. Couldn’t be an average of the two. Would that be the Hardbody? If so where are the T100 and pre-Tacoma Hilux? Would those even make sense on such a chart?
“Nissan pickup” here is indeed the Hardbody. The T100 and Hilux didn’t make the list because there aren’t enough of them in the database (right now the cutoff is 40 vehicles, which the hardbody just barely exceeded).
I would remove Hardbody from the chart then because I’m not sure what it would mean to compare it to the others, unless the other data is from the same model years as the Hardbody.
To say it another way if you had the Hilux and T-100 data they might place ahead of the C/K. I just don’t know what it means to have HB #2 without sufficient sample for its competitors to be included.
If you’re going to compare you almost have to look at it by same age I think, e.g. have everything five years old.
The quality ratings currently give bonus points for vehicles that make it over 180,000 miles without any powertrain issues, while they penalize for vehicles that have them before 120,000 miles. This gives older models (ie: C/K and Nissan Hardbody) an artificial boost, since they don’t have many/any low-miles vehicles to bring down their average(which probably could be better accounted for)
If ex-daily rentals are in the data base you would have to control for that if you want to compare say Malibu to Accord, since there are next to no Accords in daily rental fleets. A Hertz Malibu would be more beat up than an off-lease Accord.
If there are no daily rentals in your data fine otherwise suggest adding a filter so someone can look at the data apples-to-apples, e.g. all non-daily rental.
Steven, I think this is awesome and I’m a big fan of Bill James as well as a huge baseball nerd. I only checked this briefly on my phone the other day and I wanna go through it in more detail, but it’s almost midnight on the East Coast and I need to be up in 5 hours so I just want to get SOME kind of reply in here before I zonk off.
The two things I saw right off the bat that I’d do differently are instead of ranking mileage on fixed numbers (>180k / <120k), find some way to come up with a rating on a sliding scale. Also, I'd have models broken out separately by year – and, ideally – even different powertrain options. If I'm trying to find data on a 1999 Honda Civic with a manual transmission, a report that incorporates data on every Honda Civic throughout history to the present day (most which came with different engines, automatic transmissions, etc.) doesn't really tell me all that much.
In other words, I love what you're doing… but I think you're doing a career VORP when you should be doing a seasonal WAR.
Please, keep us posted – and I've got it bookmarked already. Great work!
Thanks Sean. I would be happy to have you take part in the adventure.
Feel free to reach me at carselect@gmail.com