Hi Preston. In real life we tend to stay on the sidelines when we don't have the ball. In modeling these three portfolios, we can sell when their are no replacements and it would increase returns but it may mess up our results. Let's say that using the GR/PE screen gives us better stocks then the control portfolio. During a market downturn
the control portfolio will have more sells because the stocks in that portfolio are weaker. Therefore, the weaker portfolio will go to cash faster than the stronger portfolio. If the downturn is a prolonged one, the weaker portfolio will look better because it was in cash when the market was down 5% and the stronger portfolio will look worse because it took a market correction of 15% before its stocks were shaken out.
Some suggestions:
-We somehow widen our base group of stocks.
-We only buy stocks when the NYSEBP is in a column of X's.
-We come up with a simple rating system instead of a screening system. For example, we now exclude stocks that don't measure up to a fixed FA standard (VL 1 & 2). IF we use RS in a column of X's and stocks above their BSL as hard and fast screens, maybe we can sort the candidates by FA such that sometimes #20 is a Value Line 2 and sometimes it is a Value Line 3. Right now are screens are hard & fast rules. We have to decide which rule to bend to insure we stay fully invested.
FA is very important because we have such small portfolios.
If we use bigger portfolios, FA will be less significant.
For example, our base could consist of the 500 companies in the S&P 500. Our screens for the GR/PE portfolio would include GR/PE, RS in a column of X's and above the BSL. If we go with 50 companies and sort them based on sector status, we will always be fully invested.
The control group will have the same P&F rules. The results will tell us how much positive impact that the GR/PE screen has when we compare the results of the two portfolios. We could also compare the results of the control group to the S&P 500 to determine the positive P&F bias.
The goal is to demonstrate the positive P&F bias in a simple well documented study similar to the RS study documented in Tom's book.
There are a lot of ways we can go on this. That's what I like about forward testing. You can always restart with fresh ideas.
If anyone else has any ideas or would like to work on this with Preston & I, please jump in. I know that a lot of you other P&F types are quite good at research. Even the loaning of a DWA portfolio (I only have 5 in the cheap seats) to help us track this stuff would be helpful. Someone could track the control group and post us when it has negative activity.
Give all this some thought Preston & all others interested.
duke60 |
