Why Google Fails For Accurate Short-Term Metal Price Forecasts

by on

Have you ever gone to Google to look for opinions of where prices might go for a particular material, component, or assembly to better time a purchase order? Have you ever placed a forward buy because you thought prices might increase in the next weeks or months – but then after executing a forward buy, prices went the opposite direction?

It’s because of those types of risks that manufacturing companies typically seek two types of price forecasts: long-term forecasts and short-term forecasts. Long-term forecasts tend to help companies during the annual budgeting process, or when negotiating longer-term sales contracts with key customers. Short-term forecasts help companies with the tactical buying needed to meet actual customer demand.

What are the key differences between the two when it comes to creating accurate metal price forecasts?

Long-term forecasts tend to predict prices for a period of at least one year. These types of forecasts do not rely upon statistics and therefore remain subjective. But they help buying organizations obtain industry insights to better understand where prices may go over time. We at MetalMiner believe that long-term prices hinge upon the dynamics of supply and demand. The only way to really better understand long-term price trends involves studying the specific market’s underlying market dynamics.

On the other hand, short-term forecasts predict where prices will go in the next four, six, eight or possibly even 12 weeks.

Get the latest 2014 aluminum, steel and copper price forecasts at our upcoming live event:

commodity procurement EDGE conference banner

Macroeconomic variables such as PMI, GDP, inflation or employment rates do not explain short-term price movements. Other variables such as inventory levels, capacity utilization or the cost of fuel might not show statistical correlation with the price of a particular metal and therefore should not make up any statistical model or regression, even if they might make perfect sense to you. In the commodities markets, more than 70% of daily movements appear speculative and do not represent actual supply and demand. In this manner, we believe that fundamental analysis simply fails to forecast commodities in the short-term.

So How to Forecast Commodities in the Short Term?

The answer comes down to developing and deploying statistics using objective criteria. Typing “Where is the ___ price headed?” into a Google search to better understand a particular market’s supply/demand picture will not provide the buyer with any insight into short-term price trends (unless you land on a MetalMiner page or article, perhaps!).

Instead, buying organizations should focus on objective analyses only for short-term purchases. By analyzing historical patterns, statistically testing correlations among and between variables, and testing the performance of models over time, accurate short-term forecasting can work. We know that forecasting commodities remains challenging. Fluctuations in probability will happen from time to time, but by relying upon statistics, buyers will have a better chance to make the right decisions. Statistical short-term forecasts (with good back-tests) provide buyers with the ability to better time purchases, gauge expected volatility and consistently generate cost savings or cost avoidance over time rather than relying upon subjective opinions and best guesses as to when prices might rise or fall.

To learn more about MetalMiner’s new statistical forecasting capability, please complete this form and a member of our team will get in touch with you!

Raul de Frutos Tinoco is MetalMiner’s lead forecasting analyst.


Leave a Comment

Your email address will not be published. Required fields are marked *