Hello everyone! 👋
Today, you are diving into how to use the term structure within common analyst datasets to uncover potential Alpha signals. When you examine datasets like analyst14 and analyst15, you'll notice they exhibit term structures across various fields. For instance, if you explore anl14_mean_eps, you'll find multiple fields sharing the same prefix but differing in their time horizons, such as fp1, fp2, …, fy1, fy2, etc.
🔍 Understanding the Time Horizons:
fp1: Represents the upcoming quarter.fy1: Represents the upcoming year.
These different suffixes indicate their respective time horizons, allowing you to derive estimated growth differences across many periods.
📊 Sample Template
One potential template you can use is:
group_zscore(subtract(group_zscore(anl14_mean_eps_, industry), group_zscore(anl14_mean_eps_, industry)), industry)
This template captures the sector-normalized difference between the average estimates in Period one and Period two. Building on the previous templates, you can extend this further:
<group_compare_op_1>(<diff_op>(<group_compare_op_2>(anl14_mean_eps_, <group_2>), <group_compare_op_3>(anl14_mean_eps_, <group_3>)), <group_1>)
✨ Key Points:
- The prefix
anl14_mean_eps_is kept to ensure that comparisons are made between comparable metrics, preventing your Alpha search from devolving into random comparisons. - All operators and group data become abstract choices, each embodying the economic intuition behind the original selection. For example,
<group_compare_op_1>might initially usegroup_zscore, but other valid options could includegroup_rank, which also compares the instrument to its peers within<group_1>.
📂 More Dataset Information: The dataset includes other valuable information such as the number of estimations, standard deviation across estimates, and more.
💡 Discussion Prompt: How will you systematically utilize this additional information within your templates? Share your thoughts and tips below!
Happy research! 🚀