We’ve received a lot of feedback with people asking, “How does it work?” Here’s a quick explanation – we intend to make a rundown of our approach part of the HiddenLevers site itself soon.
HiddenLevers uses statistical analysis and other approaches to model the relationships between different industries and levers (economic factors). Scenarios are then created as a set of up or down changes in levers based on the event being modeled. The scenarios thus move the levers, which in turn move industries, which move individual stocks. The difficult part of the analytical process involves measuring the real-life relationship between each lever and each industry (or stock), with the general process performed as follows:
1. HiddenLevers categorizes companies by industry, using a proprietary industry list with more detail than traditional classifications (for instance, we classify companies in the solar panel business under Solar Energy, not semiconductors).
2. HiddenLevers analyzes the relationships between each industry and each lever, starting with simple hypotheses like “airlines go up when oil goes down.” The relationship between industries and levers is measured through statistical analysis like measurement of correlation, multiple regression analysis, and other techniques.
3. In performing statistical analyses, we control for fluctuations of the market as a whole, as the movement of the broader market is responsible for as much as 70% of any given stock’s movements on a single day. HiddenLevers also aggregates data from multiple stocks in an industry to derive the industry’s relationship with a lever, since looking at a single company may obscure trends in the data.
4. HiddenLevers takes its analysis a step further by performing a detailed analysis of individual companies to ensure that their lever relationships are correct. Apple (AAPL) for instance is as much a mobile phone manufacturer as it is a pc manufacturer, and this should be reflected in its lever relationships. HiddenLevers is analyzing individual companies on an ongoing basis, starting from the largest companies traded on US markets and moving down the list.
In some cases, the relationship between a lever and industry is quite obvious, as for instance with oil prices and oil exploration companies. The oil industry’s total market capitalization has grown in almost perfect lockstep with the price of its key product, and this is neither unexpected nor unknown to most investors. In other cases, relationships can be less obvious, as with lenders and interest rates. Do lenders benefit from rising interest rates if their lending margins are able to expand, or are lenders impacted as rates rise because loan volumes drop?
Since answering these questions is difficult, and the nature of measurement imprecise, HiddenLevers‘ ultimate goal is not to present a hard-and-fast numerical relationship between companies and levers. These relationships can fluctuate over time, and the key is to understand the direction and magnitude of the relationship, so that potential risks can be visualized. Remember, all the fancy Wall Street VaR (value-at-risk) models didn’t help Wall Street avoid the meltdown – the key is to use tools to be aware of risks so that we can act on them, and to understand that our models help visualize different possibilities, not predict the future.