CSS Analytics has presented an interesting idea. It is interesting to me mainly because of the simplicity. The system just buys or sells the top 20 or bottom 20 S&P500 stocks based on their 5-day Rate-of-Change. Visit the blog for more specifics: Relative ROC (RROC) as a Relative Stock Selection Mechanism (Part 1).
This seemed like an opportunity for a little blog piggybacking. Below you will find the top and bottom 20 stocks as of Friday, ranked by the 5-day Rate-of-Change.
Thanks Wood. Where would I get the list of the top ten and bottom ten of S&P 500 ROC for each week ?
And to do this– you would sell all the twenty positions and swap them for the “new” 20 each week- is that right?
(Hey! where did my goofy gravatar appear from???? Sure didn’t pick that out lol)
It looks like buying the lowest ROC stocks was far more successful than shorting the top ROC stocks. Have you tried this approach or are you considering it? And if so, have you considered just buying the lowest?
thanks , Lindsay
Hi Lindsay. I’m not sure where you could get the list of the top ten and bottom ten of the S&P, ranked by ROC. Maybe IBD, or there is probably a free filter/screener program out there somewhere on the interwebs.
Yes, based on the initial study, you would close all 40 positions (if you went long the most -20 and short the most +20) at the end of the week.
I have never tried this approach, but I am always considering things 😉
At this point a ROC5 fade is not something you’d want to trade without much more tweaking. For example, do you open them on Monday and close on Friday? Is there a risk premium paid to hold over the weekend? Why did it only work well recently? These would be things I’d look at seriously, as well as a plethora of other tests.
Thanks Wood. Can you keep us posted if you decide to try it out or get more historical info on it? The results are sure interesting…
Yeah, definitely. We’ll see where DV goes with it, and what he discovers.
5 lowest RSi(2) stocks in the S&P 500 per stockfetcher traded weekly would have made -31% from 9/30/07 to present v. -45.6% for lowest 1 week return of stocks in the index. Since 2002 rsi(2) beats lowest 1 week return.
rsi2 low 5 v. S&P 500 v. 1 week return asc 5
2002 19.8% v. 22% S&P v. -42.5%
2003 66.7% v. 28% S&P v. 8 %
2004 18% v. 10% S&P v. 22.1%
2005 8% v. 8% S&P v. 20.3%
2006 28% v. 14% S&P v. 18.6%
2007 15.4% 6% 23.1%
2008 -37.7% -39% -70.3%
2009 53.55 12%? 51.2%
thomas, im going to have to disagree with those findings at least on the ROC side. i use a very expensive institutional platform for testing, and i assure you those results are incorrect. please try to determine what the problem is ……..cheers DV (CSS Analytics)
I have no way of knowing; I used stockfetcher. The test I ran is pretty simple so I don’t think it’s how I ran it; perhaps their data is wrong. Maybe it has the same problem as Zacks used to, which used the current S&P rather than as it changed over time.