Go here for a list of free quantitative trading models.
Today I’m going to show you a mediocre quantitative trading model. I will show you the problem with this model, and how I will make it much better in a later post.
Today’s model is a pure fundamental trading strategy (i.e. only uses economic data). As you probably know, Industrial Production is SOMETIMES is a leading indicator for the economy. And since the economy and stock market move in the same direction in the long run, Industrial Production is also SOMETIMES a leading indicator for the stock market.
This model is very simple. It only has 2 lines:
- Buy and hold SSO (S&P 500 2x leveraged ETF) if the the year-over-year change in Industrial Production is 0% or more (i.e. above 0).
- Be 100% cash if the year-over-year change in Industrial Production is below 0%.
This means that you SELL when Industrial Production falls below 0%, and BUY when Industrial Production rises back to 0%.
Here’s the data in Excel. It includes the BUY and SELL dates
This chart demonstrates the year-over-year change in Industrial Production. You can get the data from FRED.
This strategy yields an average of 9% from 1950-present. This is clearly worse than buying and holding SSO from 1950-present. A simple buy and hold would yield 13.3%!
*This also means that if you use $SPY (no leverage), this model would underperform a simple buy-and-hold strategy.
What’s wrong with this model? (The secret)
This chart demonstrates the model’s P&L on historical trades. Notice the many large losses from the late-1960s to early-1980s.
Look at the most egregious example in 1974. This model told us that “Industrial Production is ok, so the economy is ok, so you should BUY stocks” all the way until late-1974 when clearly the economy was deteriorating from 1973-1974!
So what gives?
The answer is very simple. You need to know what goes into the Industrial Production data.
Industrial Production data comes from 3 major industry groups:
Find this data here
The problem is that Mining & Utilities take into account commodity (eg oil) PRICES. Here’s a sample scenario:
- Manufacturing is showing zero growth.
- Oil prices (and commodities) are going up, while volume is mostly flat.
- Mining & Utilities activity will seem like it’s going up when in reality, it’s only the PRICES that are going up.
- So Industrial Production will seem like it’s growing when in reality, the only thing that’s increasing are commodity prices.
That’s why it’s better to ONLY focus on the Manufacturing component in Industrial Production.
For example, if you strip out Mining & Utilities, Industrial Production would have tanked in 1969-1970, 1973-1974, 1980-1981, and 2007. These were all periods in which the economy deteriorated but oil (and commodity prices) went up.
The Manufacturing component is the true representation of “Industrial Production”.
Future version of this model
We’ll use the Manufacturing component of Industrial Production in the next version of this model.
- Buy and hold when the year-over-year change in the Manufacturing component is at least 0%.
- Sell and stay in cash when the year-over-year change in the Manufacturing component is below 0%.