New engine released

The re-architected engine that was powering beta.portfolio123.com is now the production engine. All issues that came up with Compustat should be resolved. The Factset data is still under development and will continue on the beta website to avoid disruptions.

Thank you for your feedback and the issues you reported.

More info will be added soon.

This new production engine has a problem.

Hedging does not work when hedge entry and exit rules are custom formulas.

returns “UnhandledExceptionHandler: EXCEPTION_ACCESS_VIOLATION”

https://www.portfolio123.com/port_summary.jsp?portid=1598313

This new production engine with “LEGACY” does also not work.
returns: “Error Getting Backtest Status”

Investigating . Thanks for reporting

Georg , Legacy should be ok now. You best logout the log back in. If it still does’t work click to Current, then back to Legacy

We’ll take a look at your hedging problem with the new release

Yes, LEGACY works now. Thanks.

Hedging with custom formulas works in LEGACY, but not in CURRENT server version.

Strategies using “Daily” are not working. They are only running weekly now.

I am getting “Invalid Uid in price history” on Current server version, for both Compustat and FactSet. Tested sim on different models, they are all giving this error.

Georg, the issue has been resolved pending a reload to deploy the fix. Thanks for narrowing it down for us.
Simon, what exactly are you seeing that’s not the expected behavior? I checked the two strategies you ran this morning, and they have mid-week trades.

Do any subscribers use quickratioq in their simulations or screens?

An immediate (and insufficient) comparison of quickratioq in 25 companies using today’s data revealed 3 differences between values produced by what is now the default engine w/preliminary data vs. the “legacy” option.

AABVF shows N/A in the legacy and 63.96 in the default
AAME shows N/A in the legacy and 3.45 in the default
AB shows N/A in the legacy and 1.10 in the default

It’s not possible, from the outside, to investigate this issue properly. However, these results are consistent with the previous study I ran showing approximately 12% inconsistent data values either feeding or produced by the legacy vs. Beta engines.

Although I am unable to investigate this properly, I don’t believe the Beta site was ready to be promoted to default status.

Hugh

Hugh. this is a nasty one. The new engine is always filling in missing line items with the last previous values until the new one is reported.

Normally this is not as bad as it sounds since that is what our fallback mechanism does. Missing values in preliminary reports are replaced with previous values. But in the examples you gave above a very old value is being carried forward indefinitely since the company apparently stopped reporting Current Liabilities in the interim reports.

We’ll have a fix very soon.

Thank You!

Many SIM have drastically changed. Some are better some are worse. This changed drammatically …Why?



Thank you, Marco. I know it’s not easy and I am appreciative for what p123 does. I am a fan.

Here is another issue I reported in brief the other day. However, I don’t believe a trouble ticket was created:

MScoreSGAI…
AACG is 0.15 in the Current (Beta) vs. 0.11 in the Legacy
AAMF is 0.25 in the Current (Beta) vs. N/A in the Legacy
AAWW is 1.08 in the Current (Beta) vs 1.09 in the Legacy

If these differences were caused by FactSet vs. CompuStat data, it would be one thing. But these are differences between CompuStat and CompuStat involving today’s data, no less!

Also, of great concern… I noticed the trouble ticket for WeeksIntoQ has apparently been closed for the second time. I just ran a check on 25 tickers and 40% of the values generated on the Current(Beta) site are different than the values on the Legacy site, again for today’s data. Has that issue really been fixed?

Thank you.

Hugh

I am seeing significant differences across the different possible engines for my various strategies. There is no apparent trend, sometimes the model is significantly worse sometimes better. I have attached a screen capture of my test results below. For reference, FactSet and FactSet Cons represent FactSet with and without Prelim data, respectively. I’ll start digging into the details as time permits, but as you can imagine it is hard to know where to start.

Thanks,

Daniel


I don’t understand, will this change any of my models on the regular site?

philjoe, the regular site (production) by default now points to the new engine. We are fixing a few issues so yes, your models could be affected. You can point to the “Legacy” engine from the icon on the top right of the screen and there will be no changes

dnevin123, Did you run the strategies on the legacy engine or the new one ? there are issues right now with the new engine. Also, please note the Factset will have a different taxonomy called RBICS. We hope to have it up and running next week so you can compare it with Compustat taxonomy GICS

Hugh, lets wait till the new fix is in for MScoreSGAI. Re. WeeksIntoQ there has bee a small change. It’s now calculated since the “earliest date” for a filing which could be either compustat effective date (when they process some data) or filing date of the SEC. I believe the differences are for cases where there the filing date is before compustat effective date. Like when a company files with the SEC and Compustat takes 1 week to process. We could go either way on this one.

Thank You

I ran each of the models on the follwing engines

Compustat Legacy
Compustat Current
FactSet (w/ Prelims)
FactSet Conservative (w/o Prelims)

What does engine mean in this context :S

While running the Current Compustat or FactSet engine, I receive the following error when attempting to download the result of running Ranks on any of the Ranking Systems I have tested (even the P123 provided Ranking Systems). I do not receive this error when I use the Legacy engine

Factor CompleteStmt does not exist

Thanks,

Daniel