IMPORTANT: Multi-Country Functionality

[Jrinne]

All,

So this goes to the one important question in my mind: Will my port have anything to do with my sim?

Even out-of-sample this is in question. I think we have all been left trying to decide this for ourselves.

I will say that recently it has looked like there may not be much of a problem comparing my ports on auto to my sims and I have posted this. I am just looking for solid evidence (one way or the other). My only point is that I do not know and it will be a while before I can know this with a lot of certainty. I will be starting this process again, it looks like.

Now my port and sim should not be expected to look similar because of this change. My sim is expected to change. My sim has, in fact, changed.

People have been quick to dismiss this as a potential problem. But isn’t that the root of everyones concern above? There are a lot of posts above. Maybe you can explain how this is a separate concern that is completely unrelated.

I guess I can start a port on auto again and see if the sim is adequately similar the sim again, when the beta goes live.

Maybe some of you think you already know the answer. I then just ask you: What is your main concern in your posts above? Why not just rerun the sim and if it is still good why are you worried?

There are differences between the sims and ports. It is only a question of whether the differences are large enough to warrant concern.

P123 has access to a lot of data we do not have: like the data in the designer models. How similar the ports are to the sims could be answered quickly with that data. P123 could consider this as part of a quality-control measure.

Maybe investigate PROACTIVELY when a sim and port diverge by a large amount.

Or maybe we should just revisit this question regularly without considering the big picture. Like it is all brand-new again. Like we have not asked this very basic question before in multiple different contexts. [color=firebrick]Maybe continue to let people notice a big loss in their ports and report it then, like happened with the Close(-1) question and the problem with made-up opening prices.[/color]

I am all for P123 moving to international stocks and finding the best way to implement that. Please do not misunderstand my larger concern–which many above seem to share (again).

The big question is:[color=firebrick] Will my live port have anything to do with my sim after this is implemented.[/color] It has always been a good question.

Best,

Jim

[sraby]

Great change, thank you. I am Canadian and I have a linked account with IB. You mentionned thatthe base currency in IB has to be US. Anything else I should be aware trading a strategy that buy / sell US and CAN stocks?

Thank you

[geov]

On the Beta site there is no option to set the Benchmark?

[yuvaltaylor]

On the beta site, I found that when I run an etf port , place XES in the restricted buy list, the restriction is ignored, and the restrict entry is deleted for successive runs…

We’ll look into this and let you know what’s going on.

[yuvaltaylor]

All,

So this goes to the one important question in my mind: Will my port have anything to do with my sim?

Even out-of-sample this is in question. I think we have all been left trying to decide this for ourselves.

I will say that recently it has looked like there may not be much of a problem comparing my ports on auto to my sims and I have posted this. I am just looking for solid evidence (one way or the other). My only point is that I do not know and it will be a while before I can know this with a lot of certainty. I will be starting this process again, it looks like.

Now my port and sim should not be expected to look similar because of this change. My sim is expected to change. My sim has, in fact, changed.

People have been quick to dismiss this as a potential problem. But isn’t that the root of everyones concern above? There are a lot of posts above. Maybe you can explain how this is a separate concern that is completely unrelated.

I guess I can start a port on auto again and see if the sim is adequately similar the sim again, when the beta goes live.

Maybe some of you think you already know the answer. I then just ask you: What is your main concern in your posts above? Why not just rerun the sim and if it is still good why are you worried?

There are differences between the sims and ports. It is only a question of whether the differences are large enough to warrant concern.

P123 has access to a lot of data we do not have: like the data in the designer models. How similar the ports are to the sims could be answered quickly with that data. P123 could consider this as part of a quality-control measure.

Maybe investigate PROACTIVELY when a sim and port diverge by a large amount.

Or maybe we should just revisit this question regularly without considering the big picture. Like it is all brand-new again. Like we have not asked this very basic question before in multiple different contexts. Maybe continue to let people notice a big loss in their ports and report it then, like happened with the Close(-1) question and the problem with made-up opening prices.

I am all for P123 moving to international stocks and finding the best way to implement that. Please do not misunderstand my larger concern–which many above seem to share (again).

The big question is: Will my live port have anything to do with my sim after this is implemented. It has always been a good question.

Best,

Jim

If you’ve been running a live strategy over the last, say, six months, and then you run a simulated strategy with the same parameters after the change on Tuesday night, you shouldn’t expect them to match because of the changes I outlined in my initial post. These changes will cause the simulation to choose different stocks than the live strategy did.

In general, trying to get simulated strategies to match live strategies after the fact is difficult because FactSet data is mostly but not completely point-in-time and because of the changes Portfolio123 makes from time to time to keep everything running smoothly.

[marco]

On the beta site, I found that when I run an etf port , place XES in the restricted buy list, the restriction is ignored, and the restrict entry is deleted for successive runs…

It’s a bug. Thanks

[yuvaltaylor]

On the Beta site there is no option to set the Benchmark?

In live strategies and simulations that’s now in the “Universe & Ranking” tab (it was moved because the benchmark is much more dependent on the universe than on the general settings). In screens it’s on all the tabs. In ranking systems it’s where it always was, in the Performance tab.

[Jrinne]

If you’ve been running a live strategy over the last, say, six months, and then you run a simulated strategy with the same parameters after the change on Tuesday night, you shouldn’t expect them to match because of the changes I outlined in my initial post. These changes will cause the simulation to choose different stocks than the live strategy did.

In general, trying to get simulated strategies to match live strategies after the fact is difficult because FactSet data is mostly but not completely point-in-time and because of the changes Portfolio123 makes from time to time to keep everything running smoothly.

Yuval,

Correct. And I am not actually over-concerned about this but it is something I constantly monitor to the best of my (limited) ability to do so.

FWIW, the new sim on the beta site is about the same. But also no longer has much of an advantage over the benchmark with regard to drawdowns for reasons I will never fully understand, I think.

Thank you for your reply.

Best

Jim

[yuvaltaylor]

On the beta version I’m seeing master limited partnerships in my screens despite having the rule !Universe(MasterLP). Not sure if this is a bug or if there is a different way we need to filter out MLPs now?

This bug has now been fixed. Thanks again for bringing it to our attention.

[sglinski]

When I create the rank performance table for 3/21/22 (Using “ETF Rotation - Conservative” P123 ranker) on my universe called “Sectors,”
for both beta and current platforms, several tickers switch positions in the Ticker column (more so toward the bottom), even though the rank score columns are identical! I believe the settings are identical for both platforms:
https://beta.portfolio123.com/rank_curr.jsp?asof=3%2F21%2F2022&rankMethodOverride=2&univ_list=-113084&tickers=&pitMethodUid=1&level=1&getnames=0&rankid=88102&Get=Get
versus
https://www.portfolio123.com/rank_curr.jsp?asof=3%2F21%2F2022&rankMethodOverride=2&univ_list=-113084&tickers=&pitMethodUid=1&level=1&getnames=0&rankid=88102&Get=Get

[trendyist]

On the beta version I’m seeing master limited partnerships in my screens despite having the rule !Universe(MasterLP). Not sure if this is a bug or if there is a different way we need to filter out MLPs now?

This bug has now been fixed. Thanks again for bringing it to our attention.

Perfect glad that’s all sorted!

[test_user]

I’ve tested my models a bit on the beta server - things look fine. I see a small reduction in performance - but my models are somewhat fitted to the data so this is to be expected. It’s very exciting to finally have the ability to compare & rank US and Canadian stocks. I can now combine my small cap models and utility models for Canada and the US, which (based on the beta server) should increase total performance and simplify running the live models.

Happy so far!

One more question: The list of supported main currencies is not very long - how are you planning on dealing with currencies beyond the list (like NOK,SEK,DKK)?

[yuvaltaylor]

When I create the rank performance table for 3/21/22 (Using “ETF Rotation - Conservative” P123 ranker) on my universe called “Sectors,”
for both beta and current platforms, several tickers switch positions in the Ticker column (more so toward the bottom), even though the rank score columns are identical! I believe the settings are identical for both platforms:
https://beta.portfolio123.com/rank_curr.jsp?asof=3%2F21%2F2022&rankMethodOverride=2&univ_list=-113084&tickers=&pitMethodUid=1&level=1&getnames=0&rankid=88102&Get=Get
versus
https://www.portfolio123.com/rank_curr.jsp?asof=3%2F21%2F2022&rankMethodOverride=2&univ_list=-113084&tickers=&pitMethodUid=1&level=1&getnames=0&rankid=88102&Get=Get

You would expect to see some differences since close(120) and SMA(21) refer to different dates in the beta site than in production.

[judgetrade]

After the site is live the difference is still substantial, the below strat has an about 12% less performance with the new implementation →

please investigate:

https://www.portfolio123.com/port_summary.jsp?portid=1676805 (Sim)
https://www.portfolio123.com/port_summary.jsp?portid=1673115 (Live Port)

Live Port picked: BRT, IMOS (ADR!!! → Taiwan), PANL, REPX, SOI
Sim Port picked: all the above besides IMOS

Thank you!!!

Best Regards

Andreas

[atw]

It’s hard to express how much frustration a change like this can cause.

Based on this p123 update I’m getting about 75% correlation with the results I was getting previously.

In other words, the same screens I used yesterday to produce 16 stocks, not produces a list of 12 stocks that are the same and 4 stocks that are different.

How can that be?

Worse, I HAVE NO IDEA WHERE TO BEGIN IN TERMS OF SORTING THROUGH THE CHANGES.

Having spent hours developing the models I use not only to trade for myself, but to trade for other people, I have no idea what to do now except to ask…

Is it possible to continue running screens, sims, etc., based on the way things used to be while trying to accommodate the changes?

Hugh

[atw]

It’s hard to express how much frustration a change like this can cause.

Based on this p123 update I’m getting about 75% correlation with the results I was getting previously.

In other words, the same screens I used yesterday to produce 16 stocks, not produces a list of 12 stocks that are the same and 4 stocks that are different.

How can that be?

Worse, I HAVE NO IDEA WHERE TO BEGIN IN TERMS OF SORTING THROUGH THE CHANGES.

Having spent hours developing the models I use not only to trade for myself, but to trade for other people, I have no idea what to do now except to ask…

Is it possible to continue running screens, sims, etc., based on the way things used to be while trying to accommodate the changes?

Hugh

[judgetrade]

I second that, extremely frustrated, I invested 1000s and 1000s of hours in my models, just to see that 5-10% (depending of the system) of alpha is gone!

Have a look at that presentation, and you see what I am talking about
https://1drv.ms/p/s!AsT65yHImWIXhJBzZiGlLeHcDzGCbw?e=JaZf11 → P123, please investigate, I allow you to have a deep look at all models to find out how I can readapt (or to find improvements on the P123 logic).

This stuff is not new, it has OOS of at least > 1 Year, e.g., extremely valuable!

I could now trade models with simple Olikea ranking (which are not hit) and would generate the same performance (things I did 10 Years ago, before all the hours invested!).

I stand where Hugh is standing, I have absolutely no idea where to start and readapt the models to get back original alpha.
The transfer of the old data vendor to the new data vendor: models got hit, others got better. I understood that there are more small stocks with fundamental data, I had hints.
But this is a disaster! I have no idea where to start.

Also, I really dislike how this went all along:

a) P123: Do you want international Data?
b) We: Yes -->But would we have known the impact would we have said yes too, I certainly not. Why can you just not let the universes divided like they have been between USA and Canada?
EVERYTHING was fine! This is such a bad design decision
c) Then we have about a week (!!!) notice that this is coming live with no chance to adapt in time and not chance to say, hey we do not want this change to happen until the performance hits are sorted out!

Guys, there are people using P123 professionally, I live of the trading of P123!!!

I ask P123 desperately to please undo the changes until we have found out how we can readapt (or p123 to deliver changes that have less impact) and I ask P123 to help the users to do that!

Please think about the user you have more than users you will win over European data!

[sglinski]

Agreed!! Please Roll Back Until We Can Reconcile Performance Differences!! 1 Week is completely inadquate!

[michal1234]

I think the best solution would be to have the old version of portfolio and the new one running simultaneously with real time data for some period of time.
I like the new changes very much and I am waiting for Europe and Asia update, however some time for investigating the changes is needed.

[dnevin123]

Guys rolling out this change this quickly is ridiculous. I was still in the process of reviewing the beta site changes. One week is an absurdly short period of time. Please don’t treat your user base this way.