Predictive Pricing: Cutting Through the Noise

Forecasting is an inexact science, and the most successful forecasting models aren’t always the most complex. Successful forecasts are the ones that are the most consistent and trustworthy in their predictions, ones that help you eliminate uncertainty.

In pricing, we use a phrase called the “zone of indifference” to describe a range of price points where the buyer will not change their decision to purchase. There’s a similar concept in generating predictions. For example, you wouldn’t lose faith in a meteorologist if they’re off by a couple of degrees or if a rainstorm starts an hour earlier than predicted. But if you leave home in shorts and sandals and the snow starts to fly, you may change the channel the next time they come on the television.

In other words, you need to know how to plan for tomorrow. At DAT, we’ve been testing ways to solve this problem for months, and we recently launched a pilot program with Knight-Swift for our rate forecasting model which will be integrated into our DAT RateView product this spring along with a standalone API.

Contact us here to learn more about our rate forecasting pilot program.

Load Broker Holding Clipboard

History is the best predictor of the future

There’s been a lot of noise in the market recently about predictive models for truckload rates, and it can be difficult to separate claim from fact. As always, DAT’s goal is to create the best tool to help remove uncertainty from freight, providing accurate and actionable insights that are applicable to the problems our customers are trying to solve.

What sets the DAT forecasting model apart is that our RateView database serves as the cornerstone, the largest and most historically complete database in the market today. The current predictive model leverages a database of 88 million rate submissions and more than $172 billion in recorded transactions.

Why is this important? Because the best indicator of future rates is historical rates. .

Source data vs proxy data

Since DAT operates trucking’s largest digital marketplace, we also have direct access to supply and demand information, both historic and in real time. While other forecasting models rely on tangential data as a proxy for truckload pricing and capacity, DAT has its own systems that produce the most valuable source data in the industry. We then overlay seasonal, day of the week, holiday, and price momentum information — additional key information that bolsters our predictions.

During this pilot phase for our rate forecasting tool, we are already delivering rate projections on more than 128,000 individual lanes. Those are available in several different granularities (daily/weekly/yearly) and equipment types (van/flatbed/reefer). We don’t intend simply to create a widget or a graph overlaid on a map. The goal is to deliver something to be used, not just admired in a chart or graph.

Real-world results

One problem we’re solving for our customers is the ability to better respond to RFPs. From my time in the industry, I can attest that a lot of brokers and carriers during the bid process are simply “guestimating” future price movements based on limited historical data. Using our monthly rate forecasts, transportation professionals can make much more informed decisions related to risk and margin, while also winning more business.

You can also use the predictive tool to make short-term decisions about lane prioritization. For brokers and 3PLs, you can look to see if momentum on a lane is in your favor and prioritize finding capacity on lanes where the rate is likely to increase in the near future. For carriers, you can choose to search for loads that move your trucks away from markets with decreasing rates and toward markets where prices are poised to rise soon.

Statistical rigor

This is a tool I wish was available when I was running a pricing group. But the best part is that we’re not done innovating. The team of data scientists and analysts at DAT have built our forecasting product to allow for continuous improvement and selection of the best models.

We’re constantly back-testing our predictions, against both our own data and that of our competitors. That testing proved our forecasts to be twice as accurate compared to competitors who are forecasting their own data. Our models also correctly anticipate rate increases and decreases more often.

As we continue to partner with major players in transportation, new information becomes available. New modeling techniques are developed and validated, and we continually improve our predictions to make them the most accurate, actionable, and applicable to our customers.

< Back to posts