Overview
A challenge that many investors face is the ability to fairly and accurately compare manager products. The existing peer universes available through 3rd party databases tend to lump too many unlike managers in to one large peer group. As a result, decisions are often un-informed. Assembling manager peer groups along traditional dimensions of style and capitalization often leads to inappropriate comparisons.
Aapryl can create peer universes that can be categorized across 12 factor groupings or clusters that incorporate the following dimensions:
- High vs Low Quality
- Large vs Small Cap
- Style, identified as Value, Blend and Growth.
To group managers, Aapryl modifies the K-Means Clustering statistical technique whereby each manager’s factor exposures measured against each cluster’s factor exposures and then assigned to the cluster which minimizes the distance between the manager and cluster’s factor loadings.
How it works
START A CLUSTER
Clustering allows you to create more precise, accurate peer groups given their factor exposures.
SELECT A UNIVERSE
Select an entire universe, single or multiple Manager Products.
ANALYZE CLUSTERS
View and further drill down to create more relevant, precise peer groups for a group of Manager Products.
Videos
No items
White Papers
Purpose: The Clustering Module’s purpose is to properly classify funds or managers into categories. This is important to users because:
- Third party databases do not necessarily do a good job of classifying managers.
- Proper classification creates better peer groups useful for manager searches and comparisons.
- AAPRYL’s classifications are an essential building block to many of the calculations in the system including Skill Analysis and Skill Screening.
Glossary of Terms:
- K Means Clustering- Statistical method of grouping managers together based on multiple factors in which the average exposure to factors is calculated for all constituents of a universe. Groups or clusters are then formed that minimize the cumulative distance of each constituent from a group’s average.
- Quality- One of the primary classifications of managers in AAPRYL. Managers are separated into High Quality and Low Quality groupings based on their exposure to commonly used factors such as ROE, Earnings Stability, leverage, dividend yield and momentum.
- Hypothetical Beta Portfolio- Established sample portfolios used by AAPRYL as measuring stick to group managers. The sample portfolios used are preloaded into the system and are created by the AAPRYL team using proprietary methodology.
- Style- One of the primary classifications of managers in AAPRYL. Managers are separated into Value, Growth and Blend groupings based on their exposure to commonly used factors.
Description of Methodology: AAPRYL use K-means clustering to classify managers into one of the following six categories: Low Quality Value, Low Quality Blend, Low Quality Growth, High Quality Value, High Quality Blend, and High Quality Growth. The categories are used to create peer groups using the following methodology:
- A regression is run for each manager that calculates exposure to commonly used factors.
- Each manager’s exposures are compared to the exposures of hypothetical portfolios built by the AAPRYL team.
- K-Means clustering is used to group the managers into the categories described so that difference between a manager’s average factor exposure and a group’s average factor exposure is minimized.
Information Provided: Once categorized, AAPRYL is able to provide users with charts and graphs that contain an abundance of useful information which includes the following:
- Peer Groups Composition- Users can see all of the managers included in each of the 6 peer groups defined.
- Long-Term Classification- Users can see which group a particular manager has been categorized in based on the manager’s exposure over time.
- Short-Term Classification- Users can see which group a particular manager has been categorized in based on current exposures.
- Historical Classification- Users can see how a particular manager would have been classified at various points in a manager’s history.
Tooltips
No items