@Oliver. Great points. Agree with all of them and will be looking at ADRs.
I have been suspicious enough about the estimates data that I have spent the weekend looking again at Zacks. Yesterday I said to myself: “self, they really do not have backtest results that are as good as P123.” This morning, I said to myself two things. First, their Zacks Rank is guaranteed to be PIT unless there are accounting issues or they are dishonest. This is because the Zacks Rank data is generated in house. Second, I note that their less than spectacular backtest results are in line with my out-of-sample results.
The thing that is really causing me to look for answers is that my out-of-sample results are NOT EVEN CLOSE TO BACKTESTED RESULTS OVER THE SAME TIME PERIOD. I have looked at every trade for one port (and started on the others) and it is not due to slippage. Comparison of port to sim is difficult because I am still pretty new so I have made changes in my ports. BTW my ports have lots of trades and few degrees of freedom. The difference between sim and port is looking statistically significant. Still, I am making money and have no reason to complain. However, if I can find a way to do better…
Is this still planned to be fixed? It still shows 119 compnies with Country(“CAN”). This will again change the sim results for anybody using Prussell 3000 universe.
They are not the “Russell” universes, rather our own approximate versions. Excluding companies domiciled in Canada seems like a good reason to rebuild them, which we can, however it’s a balancing act between being closer to the Russells and causing backtests to change.
We also need to investigate which countries are ok to have in the Prussells. Pretty sure that Bermuda (and few others) domiciles should be ok
Just released the first Canadian R2G on P123 platform, let’s see how it goes ! It is a 20 stocks large caps with high liquidity low turnover low drawdown portfolio.
If you are interested in the highest liquidity /w no hedge version, Canadian Portfolio Managers lurking here please PM me.
I would have liked XIU… Marco any plans to make it available in the ETF list any soon? I’m going to use SPY temporarily.
This is a good question. .UN are investment trusts and I don’t think there is a way to distinguish these. I suggest using the following rule:
DivPSDays(365,0,#Regular,#PayDate,TRUE)<12
This will exclude all companies paying monthly dividends. This will eliminate the majority of .UN’s. But it will also eliminate other companies, perhaps some of them former .UN’s as the tax laws changed a few years ago and many investment trusts converted to normal companies.
Even with the above rule, there are a handful of .UN’s that won’t be filtered out. Try screening for: DivPSDays(365,0,#Regular,#PayDate,TRUE)=4 and you will still see some of these. You can get rid of these by adding them to a global restriction list or inlist.
Should we should just throw out .UN companies our of the main “All Fundamentals - CANADA”? We will create a universe only of investmet trusts that you will be able to use to screen them out, but maybe its better to separate them completely?
I wouldn’t throw them out of the main universe but perhaps a Universe filter such as Universe(UN) would be a good idea. A separate universe would also be nice but not absolutely necessary.
Geov asked a good question: how are the gain and losses of a US hedge instrument handled inside a Canadian portfolio? Are they converted in Canadian dollars every time or the US dollar is mixed with the Canadian funds assuming a 1:1 rate? This brings again why a Canadian hedge is mandatory.
In addition to these Income trusts, gonna need a reliable way to identify companies that might be PFIC’s for foreign investors. Possible flags could be very hi levels of ‘passive income’…and ‘passive assets’ not sure if Compustat lets you guys screen for this?
One weird thing that you might notice in backtests is .NV, .SV, .MV, .RV suffixes after some of the ticker symbols. This happened 9 or 10 years ago for a year or two. The suffixes described the voting characteristics of the shares. For example, .NV meant nonvoting. They weren’t very popular and didn’t last too long. But they might live on in the Compustat historical data.
! It is a 20 stocks large caps with high liquidity low turnover low drawdown portfolio.