Global resin prices have been volatile lately, mainly due to factors related to COVID-19. But historically, volatility is the rule, not the exception. It is embedded in prices, fueled by various factors centered on the prices of raw materials and upstream raw materials, according to MLT Analysis. So why do existing price forecasts tend to be relatively linear despite historical price volatility? This question intrigued the founders of MLT Analytics, who questioned whether there was a rational way to introduce volatility into resin price predictions as seen in historical data.
The answer is twofold, according to the company:
- Develop a series of hypothetical commodity price scenarios based on megatrends, such as peak oil, vehicle electrification, proliferation of natural gas electricity and regulatory developments.
- Then use machine learning and artificial intelligence to identify the main influencers of resin price. Once identified, these influencers can be correlated with historical prices to generate real world forecasts.
“The long-term forecast for oil and gas prices, from the US Energy Information Administration, for example, lacks volatility. They rise or fall, depending on the scenario, in a relatively linear fashion. What we do is introduce volatility into our forecast based on past commodity trends and assumptions about future market developments, such as peak oil. This, in turn, introduces volatility into resin price predictions, â€said Stephen Moore, co-founder and CEO of MLT Analytics. “We analyze multiple raw materials and use machine learning to explore their correlations with resins by type and region or country where they are sold,†added Moore, who is also a longtime contributor to Plastics Today.
|
MLT Analysis
|
| The overlap of the historical red line and the blue “back-cast†line indicates the validity of the modeling. |
While plastics are the starting point for MLT Analytics, prices for any type of product, including non-plastic materials, can be theoretically predicted once historical prices are correlated with data from major influencers.
Once the forecasting model is in place, the latest historical data is fed into the model as soon as it becomes available, allowing for further refinement of the forecast. In addition, the “back-castingâ€, as indicated in the graph by the part of the blue line overlapping the historical red line, is a way of checking the validity of the forecast. A tight overlap of historical and historical data is proof that the modeling works from a statistical point of view.
“Unless the data you feed into the forecast model makes sense from an industry and economic perspective, your forecast, unfortunately, will look like an unlikely outcome,†Moore warned. “This is where the decades of industry expertise that our team has accumulated become of ultimate importance. “