The "Selectable Input Smoothing" option in Global Attribution allows Eagle’s performance engine to consume a set of unsmoothed contribution or attribution effects from performance tables and smooth them with Industry-standard smoothing algorithms. The feature is designed to work at the total report level for stored performance models.
This feature is useful in several ways:
- The user can compute non-Brinson attribution effects using Eagle's Performance Calculation function and commit them to the performance tables. The sum of single period effects usually reconciles to the arithmetic excess return. However, the multi-period sum of effects is not usually equal to the linked excess returns. The difference is usually called residual. Once the effects are smoothed with a smoothing algorithm like Carino 99, the residuals would disappear.
- A client can use an in-house attribution system to break down excess returns into effects and store them in Eagle's performance db. With this feature, the client can use Eagle’s smoothing algorithms to ensure that the computed effects reconcile to the excess returns.
- If a client has pre-computed unsmoothed attribution effects from a legacy system, it can be loaded directly into performance tables and smoothed with Eagle’s smoothing algorithms.
Input Data
The inputs to be smoothed will be either committed or uploaded into the performance tables. The following table summarizes the inputs needed for contribution and attribution smoothing.
Stored Model | |
---|---|
Contribution
| Attribution
|
The unsmoothed effects are Inputs. The performance fields pointing to these unsmoothed effects are called Input fields. Returns at the total level are needed for each single period to compute K factors in Carino 99. The performance fields pointing to these returns are called control fields.
Smoothing ensures that the sum of effects is equal to the
- Linked fund or index return at Total level for Contribution.
- Linked excess returns at Total level for Attribution.
To configure the Global Attribution Group for Selectable Input Smoothing:
- In the Performance desk, click General Reporting in the left navigation.
You see the Performance Center with the General Reporting workspace. - Expand the Fields folder, expand the Custom folder, and click Global Attribution Group.
You see a list of Global Attribution Groups. - Right-click your mouse and select New from the pop-up menu. Or click New on the toolbar.
You see the Creating Global Attribution Group dialog box. - Click Selectable Input Smoothing and click Next.
You see the Creating Global Attribution Group dialog box. The Options tab display. In the Options tab, enter a name for the field, select the SIS Options and Field Map field and define the field periodicity. For more information, see Selectable Input Smoothing Options .
- Click Next.
- Select Options & Field Map from the Custom folder and click New.
You see the Options & Field Map dialog. - On the Options tab, choose the attributes required for smoothing. Only Carino 99 Arithmetic is currently supported. In the Attribution field, you can select either Contribution or Attribution. If you select Absolute (Contribution) in the Attribution dropdown, only the portfolio return would be available. If you select Relative (Attribution), both portfolio and benchmark returns would be available. Click Next.
- In the Input Fields tab, you can choose up to thirty unsmoothed effects. Click Next.
In the Control Fields tab choose the fund and index return. Because smoothing is supported only at the total level, the fund/index weights are not included.
If you selected Absolute (Contribution) in the Attribution field in the Options & Field Maps field, only the portfolio return is available. If you selected Relative (Attribution), both portfolio and benchmark returns would be made available- Click Finish.
After you have made your sections in the Options and Field Map dialog, all the chosen Inputs (Returns and Unsmoothed effects), and Outputs (Smoothed effects) display in the Effects tab of the Global Attribution Group dialog.
Add Comment