Portfolio performance now includes accrued dividends

Moving forward, the computation behind performance for live portfolios, live books, and simulated portfolios will account for accrued (unpaid) dividends, decreasing apparent volatility and improving the accuracy of both the performance curve and various risk measurements.

Great, thanks Aaron, good to know.

Hi Aaron,

Sorry - I need to check that I fully understand what this means:

I hold stock xyz on the ex-dividend date. The payable date is in 1 month.

  • Till now, the computation engine would only account for the dividend (when calculating port returns etc) after the payable date
  • From now onward, for the period of 1 month (in my example) between ex-div date and payable date, it will assume the dividend has been already “earned” when calculating all the metrics

Am I right?
If yes:

  1. Is it industry standard?
  2. Can I double check that this “accrual” is not considered cash available when re-balancing next. i.e. cash is only available for rebalance after the payable date
  3. Finally, can I check the TotalReturn function: the dividend is included only after the payable date or at the ex-div date? (I assume this function has not been modified. It is just to know)

Many thanks,

Jerome

Excellent questions, Jerome. I was wondering about these same issues. Some more questions:

  1. What is the effect of this change?
  2. Will it cause ‘perturbances’ in our past work?
  3. What was the compelling reason that justified changing the data upon which all of our sims and portfolios rely?
  4. How will this affect our sims/portfolios and everything else we have built in the past?

For me, unless there is a “pound-the-table” good reason to change the internals of the data - no matter how benign it is assumed to be - why upset the apple cart? I didn’t see anyone demanding that accrued dividends be included in the data, did any one else?

Perhaps I am a little jaded by the turbulence that has recently rocked the performance of my team’s work in the name of a massive revision of the rebalance/reconstitute module (a feature for which I have yet to identify a useful purpose). It will be interesting to see, a year from now, how many members are actually using “Formula Weight” from the Rebalance section - or for that matter, “Accrued Dividends” in a formula.

Chris

Chris,

I compared the out-of-sample performance of two of my Designer Models with their simulations that I ran today and found no significant differences in performance. These models have been out-of-sample for about 34 and 20 months. I think most well designed models will be similarly unaffected by the dividend changes.

However, I would also like P123 to explain the changes in more detail.

Walter

I am grateful to P123 when they change the data or software to make simulations and backtests more accurate fixing issues I am even not aware of.
I am more grateful if it ruins one of my models because it clarifies the fuzzy border between model design and graphical art.

Sorry guys, can we keep this thread focused on the accrued dividends change so P123 can explain on Monday exactly what this means.

I am also interested in the robustness conversation but better suited for the other thread “DEVASTATING DISCREPANCIES”.

Many thanks,

J

bump. Thank you

Before the release, daily performance was somewhat distorted by the dividends. At ex date performance was dropping by the same amount of the dividend, while on pay date it recovered the same amount.

As you can see in the image below, there is a characteristic pattern between ex and pay date. This effect is particularly relevant for low risk portfolios with sizable dividends, such as the one in the example (which is basically short term corporate bonds), while for stock portfolios the effect was much smaller in comparison to their price volatility, but still present. Daily statistics were also affected.
We solved the issue by introducing the dividend accruals in performance calculation.

If you have any other doubt or you need more technical details, P123 team is available for further disclosures.


Thank you Riccardo

Hi Aaron,

should the inclusion of accrued dividends increase or lower the portfolio performance? my sims gave me a mixed bag.

SH

Unpaid dividends are considered the same as cash exclusively for computing total return. In the simple case of dealing with only one accrued dividend, the performance will track with the old approach before the ex date and after the pay date. One should be aware that backtests pay all accrued dividends at the end date, regardless of the pay date (it did the same thing before this revision), so the ‘General Info’ might not exactly match what you’re seeing on the curve.
Bottom line: you should see neither increase nor decrease due to this revision.