Rupert Pontin, director of valuations at Cazana.com, looks at a completely new way of data-driven forecasting, one which doesn’t involve any human intervention and provides real-time data to the industry.
The traditional model for vehicle forecasting has always been built around a combination of both human opinion and data-driven statistics.
Organisations collect market data from a previous month and have a team that works with anecdotal information from auction houses and market contacts.
Valuation data editors then review information and change what they see according to historic data, attempting to predict what could happen in the coming months.
This has been the model for many years and has played an important role helping manufacturers, dealers, remarketing companies, gateway providers and finance companies understand residual value risk and the changing prices of vehicles.
Where this approach falls down is both on accuracy and retail relevance.
Once you start to pass too much data through too many humans and editors, it is difficult to paint an accurate picture of the market without some bias creeping in.
Data edited by individuals may show a swing in results because an editor has changed their mind and not because there is a swing in the market.
If you are a finance company taking a risk on an asset, or a dealership buying a car at auction, you will be selling that into the trade or retail environment.
As a result, you need to understand how the market is behaving and how prices are changing right now, to be able to fetch the best price.
If market data has been edited and manipulated, it is no longer 100% accurate and could end up leaving you with stock assets for longer than you would have liked, based on a level of subjective contemplation.
Working from retail data
Instead, most dealers are making concerted efforts to work from retail data downwards and quickly moving ahead of where prices and valuations have traditionally come from, heralding a new approach.
Cazana is the first vehicle valuation engine to correctly value vehicle condition and specification, to help its client’s price vehicles more accurately and with greater certainty – made possible through its innovative science-based business model.
Whilst many businesses are working on historic data, the vehicle market is seeking out real-time analytics which can give an up-to-date reflection of retail market trends.
To do this requires an altogether different business model; one built around science and data, as opposed to editing.
This new approach is to collect vehicle data 24 hours a day, seven days a week with a team of data scientists and machine learning experts constantly building new models that interpret what the retail market is doing.
The entire process is automated, with a team of specialists working to translate, rather than edit the information, for market consumption.
Quality over quantity is becoming the norm and the speed at which data is being acquired and processed is critical to giving accurate, real-time market insight.
Cazana reviews in excess of 15,000 individual websites every day, collecting vast amounts of data to process, analyse and create insights from.
Over the course of a year this amounts to more than 10 million data points and that is before considering advancing technological enhancements which are facilitating further increases in data.
The market has moved ahead of edited data and is looking for easily accessible, real-time market insight, using nothing but a vehicle registration number.
Just as technology is making vehicles easier to buy, it should be making them easier to value and sell. With a new, data-driven approach to valuations, the trade is taking advantage of better forecasting tools, returning more accurate prices for vehicles, selling stock more quickly, and making higher profit margins.