In Downloaded Simulation Chart Data Portfolio Value is Changing On Trading Holidays

Research
1

Guys just noticed that when I download the Chart results from a recently run simulation that the Portfolio Value is changing on the now visible Trading Holidays.

[url=https://www.portfolio123.com/port_summary.jsp?portid=1677836]https://www.portfolio123.com/port_summary.jsp?portid=1677836[/url]

See the simulation above and take a look at the change versus the previous day in 100 USD Ret column. I have looked at the trading holidays in 2022/2021 and found the following percent changes in Portfolio Value on Trading Holidays

2/21/2022 = 0.031%
1/17/2022 = -0.401%
9/6/2021 = -0.519%
5/31/2021 = 0.359%
2/15/2021 = 0.645%
1/18/2021 = -0.374%

Some of the dates appear to be behaving correctly and just copy the value from the previous day. For Example,

12/24/2021
11/25/2021
7/5/2021
4/2/2021
1/1/2021

I did verify that these dates don’t appear in the downloaded Chart Results from a simulation that was run prior to the recent changes.

-Daniel

2

Guys are you taking a look at this?

It seems bad… don’t see how we can have any confidence in simulation results with this type of behavior

I’m seeing it in everyone of my new simulations, but the changes in portfolio value are completely random from day to day and simulation to simulation.

What am I missing?

3

Yes we noticed. I don’t think it affects the overall performance since money or extra shares can’t just be created out of thin air. But I don’t know for sure at the moment. Thanks

4

Ok… I just did the math for two of my simulations that run from 2/18/2010 to 2/18/2022 and the cumulative performance on Trading Holidays is -5.97% and -3.99% respectively. (I manually created a list of trading holidays, so might have gotten a couple wrong).

5

ok, but everyday the portfolio value is calculated from holdings*price + cash. So I think these are transient errors

6

Hopefully that is the case, but seeing changes on those days implies some kind of off by one error to me (at least that is what I would worried about). It surely means that the calculated percent change for the next day is incorrect, but that should be an easy and not too consequential fix.

7

What seems to me to be happening is that transactions that would have taken place on the Tuesday after the holiday are being logged as taking place on the holiday. If that’s the cause, then, as Marco said, these are transient errors, and don’t affect any of the performance numbers except those for the holiday and the day after. My evidence is that there are transactions on the day after every single one of the affected holidays and no transactions on the day after the unaffected holidays.

8

Yuval, that makes some sense. Is this a behavior that can be corrected, so we can make sure there is nothing else funny going on?

(Note: it looks like Dividend Payments and Stock Splits might cause this as well)

-Daniel

9

It’s in the top priority list of things to look at.

10

Any progress on this issue? I would like to see it fixed before I feel comfortable proceeding with the work I started before all these changes were implemented.

Thanks,

Daniel

11

Hi Daniel,
The fix is being tested now and we hope to get it out by the end of this week if no issues are found in testing.

12

Thanks!

13

Hello Dnevin123,
Just an update, we’re still working the issue and the earliest it will be released is mid next week.
We’re sorry for the inconvenience and thanks for your patience.

14

Looks like the Change in Portfolio Value on Trading Holidays is still occurring for simulations run on the Beta site.

-Daniel

15

Hi Daniel,

Quick update: We’re still working and testing the issue, but it seems like we’re hitting a slight delay for the fix.
Apologies for the inconvenience. We still consider this a high priority issue and are working to get it out as soon as we possibly can.
Thanks for your patience.

16

Hi Daniel,

We have released the fix for this. Thanks for your patience.
Please let me know if you are still running into the issue.

17

Thanks Hem, it looks like it is fixed to me.

-Daniel