IMPORTANT: rolling back changes in the main site. New version is on beta.portfolio123.com

[atw]

Are the rollbacks permanent or interim?

If interim, do we know for how long?

Under the circumstances, it’s very hard to know what changes to focus on or try to adjust for, or even whether I should be working on the old or Beta site.

Thank you.

Hugh

[yuvaltaylor]

The only change that has not been rolled back is the taxonomy series. It should have a minimal impact since only people that use something like [font=courier new]Close(0,#industry) / Close(21,#industry) [/font] will be affected. We’ll try to roll it back before the Monday rebalance but cannot guarantee it.

Any update on when this change will be rolled back?

Thanks,

Daniel

Daniel -

I was mistaken about the taxonomy series. It does not affect the series parameters. The taxonomy series only affects charting and the Industry and Sector pages. It should not affect screens, simulations, or ranking systems.

[atw]

So what are best guesses at this time for issues on the rolled back site that could be still be causing some models’ sims to degrade and/or screens to return a different set of stocks than the week before when p123 initiated the changes?

Thanks very much.

Hugh

[marco]

I was mistaken about the taxonomy series. It does not affect the series parameters. The taxonomy series only affects charting and the Industry and Sector pages. It should not affect screens, simulations, or ranking systems.

Sorry folks that is not correct from Yuval. Taxonomy series do include holidays at the moment for stock pages, screens, sims, etc. You can see this by running this screen with as-of date of 2/22/2022 (around prez day holiday)

ticker("ibm")
showvar( @0,Close(0,#industry))
showvar( @1,Close(1,#industry))
showvar( @2,Close(2,#industry))
showvar( @S0,Close(0))
showvar( @S1,Close(1))
showvar( @S2,Close(2))

You will see a duplicate value (carried forward) for the industry close price variables @1 & @2, but not for the stock.

Sorry about the confusion during this time. We’re working on a solution that should please everyone

---

[judgetrade]

I am a bit confused.

Questions:

1. What do you want us to do with the beta site? Do you want us to test the changes and give feedback so you can decide which changes will be implemented in the production site?

2. Which leads to the second question: What will be permanently be rolled back in the production site?

3. What changes are in effect in the production site (as a summary, I am really sorry, I am getting lost tracking the answers).

Thank you!!!

[dnevin123]

I second Andreas’ questions.

I am also confused, and don’t trust any of my models at the moment. This means that despite P123’s heroic efforts over the weekend I am unable to trade or make any progress this week.

Marco, why not just roll back all the changes that were made instead of trying to develop a new solution on an abbreviated timeline?

Thanks,

Daniel

[marco]

We’re coming up with a solution that should please everyone. As I mentioned in the other thread here I think we have a solution that will allow us to keep the data structures with weekdays (so everything lines up) , but keep existing TA functions using bars by substituting a pre-built time-series with no holidays on the fly. Which means your systems will look exactly the same. And we’ll introduce new TA functions for weekday operations that we think should be used for multi-country strategies.

Sorry for the flip-flop. It’s a solution we did not think of before.

What this means for the current production is this
[list=1]
[]It should be used for now for rebalances and backtests
[]Taxonomy series still have weekdays. We’re going to try to roll them back today. They are accessed like this for example: Close(0,#Industry).
[]The North America universe is for testing only, things won’t align properly with TA functions, and will go though changes
[/list=1]
What this means for beta
[list=1]
[]It should be used only to test the effect of weekdays in single country and multi country
[]Any system developed on beta will have to be rewritten using weekday functions (since TA function will revert back to bars)
[]The North America universe is in beta, it will go though changes in Primary issues
[/list=1]
Again I apologize for this messy release. It’s a big refactoring with support for currency, holidays, Primary issues.

Thank You for your feedback

[test_user]

Well, this is a bit of a problem for me since I’ve converted completely to NA setup. I think I’ll be fine for a little while, but do you have any idea how long time it will take to develop/deploy the new weekday functions?

[marco]

Well, this is a bit of a problem for me since I’ve converted completely to NA setup. I think I’ll be fine for a little while, but do you have any idea how long time it will take to develop/deploy the new weekday functions?

Can you provide a list of the ones you need?

[test_user]

Well, this is a bit of a problem for me since I’ve converted completely to NA setup. I think I’ll be fine for a little while, but do you have any idea how long time it will take to develop/deploy the new weekday functions?

Can you provide a list of the ones you need?

Sure! The revelant functions are:

Mediandailytot, avgdailytot, medianvol, close, pctdev, sortino.

I think that’s all of them, I’ll dig a bit deeper and let you know if I discovery anything more.

[test_user]

Sure! The revelant functions are:

Mediandailytot, avgdailytot, medianvol, close, pctdev, sortino.

I think that’s all of them, I’ll dig a bit deeper and let you know if I discovery anything more.

I also use FHist, perhaps that too has to be updated?

[marco]

FHist uses weekly samples on the weekend so not affected by holidays

[mm123]

Hi Marco, what’s the latest on removing the “:USA” suffix following tickers? Thanks.

[marco]

mm123, you are downloading a list?

The whole entering / downloading ticker lists still has several UI issues . Sorry about that.

[hataso]

Rolling backtests are not giving correct results.

[dnevin123]

.

[mm123]

Marco, I’m not downloading, it’s what I’m seeing on the website.

---

[atw]

Hi Marco,

Thank you for posting your update in the separate thread that is locked for comment. That is why, of course, I am replying to that posting here.

Questions:

What is your best guess for the timing of p123’s “final release”?

At the time of final release will Beta replace existing production?

To put things another way, as a CompuStat subscriber focused entirely on USA, at what point can I resume testing and revising models in a stable environment?

Thank you.

Hugh

[test_user]

I prefer keeping the “:USA”, a consistent naming scheme is easier to work with.

[marco]

Thank you for posting your update in the separate thread that is locked for comment. That is why, of course, I am replying to that posting here.

Questions:

What is your best guess for the timing of p123’s “final release”?

At the time of final release will Beta replace existing production?

To put things another way, as a CompuStat subscriber focused entirely on USA, at what point can I resume testing and revising models in a stable environment?

The final release will:

• Introduce new weekday TA functions
• Internally all time series including stock prices will include weekdays but should be invisible to you if you use existing TA functions
• Tweaks to what is Primary for a region which only affects new universes and the Ind classification series (a.k.a. taxonomy series)

We should have a release by end of next week with at least the main TA functions that use weekdays.

So if you are using older universes, like USA Incl ADRs & dual listed (formerly called All Fundamentals - USA), don’t need weekdays, you should be fine right now.

Thanks

PS perhaps we should unlock the thread it and just keep it pinned