It's neither. A traditional screen (top 5) with monthly hold would typically give four "possible" runs through the year. The first "run" starts the first week of a month, then the second etc. When doing backtests such as these, they are typically combined using average values to end up with an "average" which may or may not have a very significant variation based on which run you ended up with. A way around this would be to make sure you hold stocks for all runs. However, if you bought top 5 every week, you would end up holding 20 stocks. Instead, you could buy one stock every week, until you have five stocks, and then replace the oldest one with a new from the fresh pick from the latest week. This is what I am backtesting. I call it time diversification, but it has also been called instance diversification. My reason for testing it this way is that this is the way I would typically trade it, and it is less work than doing four different/independent runs and averaging them.
So, in summary, the backtest is basically buying the top pick every week, holding it for at least five weeks, and selling it when it has been held for at least five weeks. You will notice from the "main" backtest in the pdf file that the average number of stocks held is just below 5. This reflects the fact that the first week, you only hold one stock, two the second week etc. After five weeks you should always be holding at least five stocks (assuming the screen always have fresh picks available).
It's neither. A traditional
It's neither. A traditional screen (top 5) with monthly hold would typically give four "possible" runs through the year. The first "run" starts the first week of a month, then the second etc. When doing backtests such as these, they are typically combined using average values to end up with an "average" which may or may not have a very significant variation based on which run you ended up with. A way around this would be to make sure you hold stocks for all runs. However, if you bought top 5 every week, you would end up holding 20 stocks. Instead, you could buy one stock every week, until you have five stocks, and then replace the oldest one with a new from the fresh pick from the latest week. This is what I am backtesting. I call it time diversification, but it has also been called instance diversification. My reason for testing it this way is that this is the way I would typically trade it, and it is less work than doing four different/independent runs and averaging them.
So, in summary, the backtest is basically buying the top pick every week, holding it for at least five weeks, and selling it when it has been held for at least five weeks. You will notice from the "main" backtest in the pdf file that the average number of stocks held is just below 5. This reflects the fact that the first week, you only hold one stock, two the second week etc. After five weeks you should always be holding at least five stocks (assuming the screen always have fresh picks available).