Comparing used car deals – or for that matter figuring out what is the right price to sell a car – is both tedious and difficult. Why tedious? Because there is no “advice” provided by anyone – in the specific sense that sites list a ton of used car deals without giving any priority listing of what’s good and not-so good. So as used car buyers, we end up sifting through tens and sometimes hundreds of car listings to figure out our shortlist.
And that’s a difficult job because we are always comparing apples and oranges. A Hyundai i20 Magna 1.4 CRDi 2014 with 78,000 km (red color) or a Hyundai i20 Sportz 1.4 CRDi 2015 with 87,000 km (red color) – both listed at Rs 230,000?
This one is within the same brand, and model. When we are comparing between brands and models, it’s really difficult keep all the factors in our head – and try to choose. What we then end up doing is ask around for advice, which will always a hit-and-miss.
What if we had a way that allows us to easily compare used car deals – across models, brand, different age, different mileage, petrol or diesel or CNG? A metric that factored in all the important variables across all major used car listing sites – original car price, listed used car price, mileage, year of make, brand, model, petrol or diesel – and then gave us a “True Price”.
So with all variables accounted for, all the cars had a True Price value associated with it. Compare the True Price of the car with its listed price on the site – and that tells you whether it is a good deal or not. If the listed price is lower than the True Price, that’s a great deal. If higher, than not so much.
That’s exactly what the Data Science team at Cartoq has done. We collected over 29,000 data points from all the major used car listing sites, about the listed price, brand, model, mileage etc. And then we trained this big data set on different machine-learning (ML) algorithms to predict the used car price. What we eventually got was an optimized ML model that predicted the true price of a used car, which had the least amount of variance with our test data. Without getting technical, our algorithm “learnt” how much each factor (year, mileage, brand, model, car type, fuel type and a few others) affect the depreciation of a car.
With a large data set of 29,000 used cars to learn from, we were able to build an accurate equation to predict the used car price, or the TRUE PRICE as we call it.
An important point: the TRUE PRICE is not necessarily the recommended price to buy or sell. BUT SORTING the used car listing on the basis of their TRUE PRICE will help identify which deal is better than the other. So the TRUE PRICE helps identify the BEST DEALS, and thus will help the buyers and sellers to choose and negotiate the final price better.
Using the Cartoq TRUE PRICE Index, we will be publishing articles for the best used car deals going around. We aggregate the used car listings from all major sites, and update this list periodically. From this list, we will share the best deals with you daily – which will address a variety of buyer and seller requirements, including budget, car type, brand, mileage and more.
What are the best deals to buy a used car if your budget is up to Rs 3 lakh? What if your budget is Rs 5 lakh but you want cars with less than Rs 5 lakh mileage?
For those who want to sell their cars, expect articles such as: If you have Swift ZDi 2016, with mileage between 40,000 km and 50,000 km, what is a good price for your car? What if your mileage is only 25,000 km?
Basis the Cartoq TRUE PRICE index, our readers will now have an independent, objective source of assessing whether they are getting a good price for selling or buying a car. Check out these Used Car best deals articles.