Building and Selecting Features
On financial data, the choice of features often matters more than the choice of model. The questions about them are practical: "propose features for this target", "how many features would you try?" and "how did you choose which ones to keep?". The last question matters most, because the usual way of choosing features makes the result look better than it is.
Constructing a feature
A good feature has four properties.
- It is point in time. It uses only information available at the moment the decision would have been made, including publication delays for fundamentals and revisions to economic data. See look-ahead and survivorship bias.
- It is stationary enough to learn from. A price level has no fixed mean, and a return does, so use changes, returns and ratios rather than levels.
- It is comparable across assets and time. Scale by volatility, or rank across the universe. The same value then means the same thing for a quiet utility and a volatile technology stock.
- It has a reason. An economic story, such as slow diffusion of information, risk compensation or a structural flow, makes a feature more likely to survive out of sample.
The rest of this lesson is for subscribers
Unlock every lesson in Machine Learning for Quantitative Research, and every other premium course.
Subscribe to continueTest your knowledge
Questions are only available to subscribers.
Keep reading Machine Learning for Quantitative Research
27 lessons in this course, and every other premium course, on one subscription.
- Every lesson in every course, with the worked examples and interactive simulators
- Graded questions on every lesson, with explanations for the wrong answers as well as the right one
- The trainers, timed assessments and brainteaser library that go with them