Yup, it’s the time of year again. Black Friday is upon us, and with it, the all-important holiday season starts in earnest. The forecasts are in and predictions are out. My friend Walter Loeb predicts an exciting holiday season, with Amazon exploding again and Walmart and Target working hard to hold their own. The National Retail Federation (NRF) predicts an industry-wide increase of 3.6% over last year. I’m sure there’s lots of math behind their prediction. And most observers use the trade association’s number as the key benchmark.

Everyone loves to read forecasts and most of us trust them. In fact, every year, we observe in our merchandising benchmark data that today’s retailer places a lot of faith in retail forecasting of all sorts. And then, we also observe that those same retailers don’t really understand how those forecast models work. Usually, our commentary is somewhat caustic: how can you trust what you don’t understand?

In fact, those same retailers often cite “Reacting to deviations from sales forecasts ” as a top-three business challenge. So clearly they know that things can happen to skew their numbers, but no one really knows why.

This year’s data is no different. As you’ll see early in 2017, retailers really do believe forecasts are key to their success and as in other years, they don’t quite understand what they do.

But this year, I’m not going to take it as fact. Look at the US election that just happened: just about every poll, every forecast, every “poll of polls ” was completely wrong. How did that happen?

I’m a data junkie. I don’t always understand the algorithms behind what I read, but I trust that the people who put out their predictions or statements do. I know how RSR’s surveys work, but ask me to explain Nate Silver’s algorithms and I’ll glaze over. But I believed it. It looked complicated. He used words like algorithms, standard deviations, continual re-forecasting, and other terms that left me suitably impressed.

As we have been reminded, forecasts are not infallible. Certain things are easy to see. You don’t have to be a rocket scientist to know that opening one day early on a major holiday isn’t really going to increase total demand for the holiday season. And those non-incremental sales are expensive in payroll, light and heat. RSR knew that in 2011, and by this year, most retailers seem to have also figured it out. Many more retail chain stores are closed on Thanksgiving Day in the US than are open. This is something we applaud.

But what’s the season really going to look like? Is 3.6% accurate? And in the world of averages, is your chain going to be above, at or below that average? And what are you going to do when you realize your forecast wasn’t quite right?

At the end of the day, modern times call for nimbleness. We have to adjust. Markdowns happen. Out of stocks happen. Smart retailers will watch sales and inventory levels carefully. I’ve always wished retailers would just buy less and make hot products a scarce commodity for the holiday season, but that’s my pipedream. No one is going to leave open-to-buy dollars on the table.

So be prepared for the unexpected. We wish you all the best. Happy Thanksgiving and happy selling!