How we do it
When trying to understand and model complex systems like mining operations, first-principle modeling is an effective strategy to incorporate deep domain-understanding, uncertainties and interdependencies.
When we start modeling a new asset, we break it down into a granular set of sub-processes and connect them to map the processing flow, from mining to shipping. The level of granularity is largely determined by available data and what information is critical to model the impact of.
The next step can be described as an "accounting" of published data. We rigorously go through annual reports, etc. and ingest this data into our data-model - unchanged, with original metadata. This step includes historical as well as information from news sources that describe possible future developments, such as expanding processing capacity, etc.
Finally, we chain the processes together, using historical data to calibrate their dynamics and to forecast according to assumptions about the future.
From this approach, we can not only trace back the impact of each sub-process or assumption, we can also integrate uncertainties and update the output with every new piece of information.
Questions about the methodology? Want to discuss how this applies to your portfolio?