There seems to be an error in the “save log”. As attached, the log only shows American shares, even though the screen backtest itself is in Norwegian (ExchCountry(“nor”)) and appears to be correct, but the log shows something completely different?
And what dose this symbol mean besides the ticker symbol “^”
Hello Whycliffes.
The log file displayed is unrelated to the backtest you ran. This is only happening on the Beta site, so there could be some code that is incorrectly looking at the Production site and returning the wrong log file. I will pass this issue onto the developers.
The ^ in tickers means that it is an inactive ticker. For example, NWK^12 became inactive in 2012 but was active in the 2000-2001 period covered in that log file you posted…
Thank you so much, Dangp!
I was also curious what the ratingsystem does with a node that it cannot find any information on in the European stock data?
Assume I want to test the R&D effect in Europe (IsNA (RandDTTM, RandDA) / MktCap), but there is insufficient information in the dataset. How would p123 handle that node? Will it show an error? Will it give it zero weight?
Insufficient information results in a value of N/A. N/A handling is controlled by the “Ranking Method” tab on the right of your ranking system. You have several choices. Percentile NAs negative will give all your N/As the lowest rank. Percentile NAs neutral will give all your N/As a middling rank. Normal Distribution does something different, but we don’t advise using it.
I don’t think this is the proper handling. Opincttm(0,qtr,ZERONA) and Opincttm(1,qtr,ZERONA) should both show the six month number divided by 2. Otherwise, those factors over mktcap or ev or anything won’t be comparable to us and can numbers. That makes having all stocks in a universe problematic.
This is a great point. Having an all stock (EU, US, and CAN) isn’t even a prerequisite for this to be an issue. Since some European countries require quarterly reporting, even a full EU universe would be comparing apples on oranges.
In the current form, I don’t even think there would be a way to get consistent values for quarterly vs semiannual companies with the FALLBACK option. While OpcIncTTM(0,qtr,FALLBACK)/2 would get you quarterly values for semiannual reporting, it’d get you half the correct value for quarterly reporters.
We are introducing semiannual functions that should help with these and similar cases.
In the meantime, you can use an Eval function to determine if a stock is semiannual or not, and then treat it accordingly. The Eval function is Eval(InterimMonths(0)=6, . . .) If that’s true, then the stock reports semiannually.
I’m not sure what you meant by OpIncTTM(0,Qtr). Either you’re looking for TTM or for Qtr. If you meant OpInc(0,Qtr) and OpInc(1,Qtr), you are going to get numbers that are not comparable between semiannual and quarterly companies, of which there are plenty of both in Europe, as Dave points out. And Dave is right, fallback options won’t help. You really need to use the Eval function to get numbers that make sense.
Here are a few examples from my own ranking systems.
eval(interimmonths(0)=3,1,2)*loopsum("epsexclxor(ctr,qtr) > epsexclxor(ctr+4,qtr)",12,0)This counts the number of times the quarterly EPS is larger than the same quarter the previous year over the last 12 quarters. For semiannual companies, the number of quarters that could possibly fulfill this would be 6. So I double the count if it’s a semiannual company.
eval(interimmonths(0)=3,1,0.5)*(opercashflq+cashfrinvestq)/asttotqThis ranks companies on the free cash flow ROA of their most recent reported quarter, using the sum of operating and investment cash flow as an alternative to the conventional free cash flow measure. For semiannual companies, I use only half of this sum so that it’s comparable to quarterly.
To follow up on this, I was able to set up a Margin account Roth IRA in IB and seems like I have the ability to submit international orders. For initial set up, I had to submit a request to access each individual country when submitting a stock from a new country for the first time and permission was granted almost instantaneously.
I get some strange results using beta vs. the original p123 page
I tried to see if it is possible to replicate Alpha Quant performance. The strategy of Alpha, and the one I test on p123 is:
I looked at the return it gave in the Screen backtest in the p123 original page, and the beta page. I’m getting completely different results and can’t figure out why:
I see that the universes can contain somewhat different numbers of shares, but it should not be able to produce such different results. What am I not seeing?
I get some strange results using beta vs. the original p123 page
I looked at the return it gave in the Screen backtest in the p123 original page, and the beta page. I’m getting completely different results and can’t figure out why:
I see that the universes can contain somewhat different numbers of shares, but it should not be able to produce such different results. What am I not seeing?
I have run a number of different tests, with different unviers, ranking system, rules, and I always get different results in Screen backtest.
The starting universes shown in your image are very different. ‘United States (all listed stocks)’ contains ADRs and ‘United States Primary’ does not. If you want to use ‘United States Primary’ in Beta, then use the ‘United States’ universe in the Production site.
I altso did test it against a lot of different universes, but still getting different results.
But here is my last test with "‘United States Primary’ in Beta, then use the ‘United States’ "
Beta site: 3,812.47% 33.89% -34.12% 1.81 2.71 16.73% 0.76 0.57 0.85 21.86%
Normal site: 816.11% 19.28% -38.45% 1.10 1.52 16.88% 0.91 0.83 1.03 6.26%
Can you please post the URLs of your screens?
Main site:https://www.portfolio123.com/app/screen/summary/263636
Beta site: https://beta.portfolio123.com/app/screen/summary/263636
I’m very happy about the beta site so far, it’s easy to make a universe of the markets I need, and everything just works.
The only issue I’ve had so far is: Stolt-Nielsen Ltd. (SNI:NOR) has a forward dividend yield of 43,5 %, while most other sources say it should be roughly 4,5 %. I can’t find any information about a special dividend, so I wonder if something is wrong. Perhaps the issue is that the company is domiciled in Bermuda?
test_user, what happens when you rund identical screen backtests one the beta site, and det main site? Do you get the same results?
I’ll take a look - waiting for the servers to load!
FactSet gives us two metrics IAD (Indicated Annual Dividend) and the actual Yield for current data. They should be equivalent but using IAD gives 43% yield. Definitely seems fx related . 1 USD= 0.1 NOK. We think we are doing everything right. Might be a FactSet issue. Still investigating
Thanks
NOTE: historically FactSet only gives us Yield together with the filings. It’s not great since companies usually announce changes in the middle of a period, but it’s all we got.
After testing a bit I see some very small differences, but nothing big or surprising. I’ll test a bit more and report if I see anything strange.
Thanks - This is the only company where I’ve found an issue, everything else seems fine so far.
These two screens look completely different to me. All the rules are different. The universe is different. The ranking system is different.