The Style Drift Score
The Style Drift Score measures the change in style over time.
Below, the Asset Allocation Graphs on the left show an area graph of the combination of the indexes that have the lowest tracking error to the product selected over a ten-year rolling period. The Style Maps on the right show the location of the products’ style within a Small to Large and Value to Growth Style Box. The larger points represent more recent periods.
The top two charts show Product A and the bottom two show Product B. We can see from the top left chart, the orange or Russell 2000 Value consistently has the most area coverage throughout the time period for Product A. We can see from the corresponding map on the right that Product A’s points are consistently in the same location in the Russell 2000 Value corner. Conversely, for Product B we see colors scattered throughout the time period in the bottom left chart. The map on the right shows Product B has clearly shifted styles by starting in the Small Value corner, moving to Large Value, then to Large Growth. While it is easy to see these results visually, it would be useful to quantify this information. This is what the Style Drift Score does.
The Style Drift Score:
The Style Drift Score measures the variability of the style over time. It does so by measuring the volatility of the coefficients or "weighting" to each index in the style benchmark. A Zephyr Style Drift Score of 0 indicates perfect style consistency. The larger the Style Drift Score, the more inconsistent or “drift” there is of the style of the product. The calculation is below:
*One main advantage to looking at the Style Drift Score is that it encompasses in a single value all the information you see in the graphs above. It is also more accurate than other measures of style drift.
Style Drift Scores for Products A and B above:
Product A: 9.49%
Product B: 53.24%
Other Metrics that attempt to measure Style Drift:
- Tracking Error relative to a single Market Benchmark: This can be a reasonable measure of style consistency. When the tracking error to a single benchmark is low, this generally means the style of the product is consistent with the market benchmark. However, if it’s high, the style is unknown.
- Style Benchmark Turnover: This does not show changes of style over time so it fails to show if one product moves between styles or moves a lot within one style. Below are pictures of these scenarios where both Style Turnover values are around 44%.
If you are interested in a deeper understanding of Style Drift and the Style Drift Score, please see this white paper from our website here
and/or the calculation from our website here
For a broader discussion on Style Analysis, you can find that here
. Research articles on the math behind Returns-Based Style Analysis are here, Part 1
and Part 2
More information about this topic: