Today’s market is volatile and difficult – success in it requires having a good idea of where prices are going. 

In today’s rapidly changing market environment it’s more important than ever to make fact-based decisions when it comes to buying and selling. Oftentimes the difference between making a profit and taking a huge loss is decided by pennies, so precision matters in determining where a market is going. Many companies project prices on key commodities using only current trends as a basis for their assumptions, which can be a very risky way to run a commodities business.

A forecast and a projection are both ways to do it, but they’re not the same thing, and the difference is important.

So, are you using a Projection or a Forecast?
First, a definition: A PROJECTION is defined as an estimate of future possibilities based on a current trend, while a FORECAST is a calculation of a future event based on data analysis. What makes commodity price forecasting so complex is that routine buy/sell decisions are made for points in time well into the future, and the further into the future you’re forecasting, the greater the difficulty. To do a reasonable job of forecasting commodity prices two to six months out requires a robust understanding of what factors are moving markets, and that’s what we’re going to discuss here.

Part of good forecasting is understanding drivers (i.e., not relying on a black box)

Being able to transparently see what’s driving a forecast inspires confidence in that forecast. The alternatives of course are a “Black Box” forecast, in which an answer is delivered without explanation, or historical averages, which is really the most basic of forecasts. 

And to make your forecasting even better, it’s best to use multiple approaches at once, to be more confident in your answer. 

Best practice in forecasting is to provide users with multiple independent forecasts, each based on unique market drivers. For beef and pork items best practice is to use forecasts built on 1) the ratio of the cut to carcass value, inflated by the CME forward curve; 2) an econometric forecast based on supply and demand variables built by data scientists and industry experts; and 3) machine learning algorithms based upon historical trends, seasonality, and machine-selected driver variables.

Lastly, game planning scenarios in the market adds significant context around what might happen in the coming months: 

• Maybe you know of a large, market impacting event and would like to quantify its impact. 

• Perhaps you would like to see the impact of labor shortages or supply disruption. 

Best practices is to allow users to interact with the forecast and all the variables driving it to draw immediate insight. 

Finally, once the forecast is modeled out to your satisfaction, best practices is to then help our customers script out buy/sell options in real time, using the forecast to find the optimal timing and deal structure.  Identifying the risks surrounding various pricing decisions is also important.