Kevin_in_GA
4,599 posts
msg #80668
- Ignore Kevin_in_GA
modified |
10/5/2009 7:58:42 PM
Ever since I first read TRO's thread on his CROCK POT filter I have been intrigued by its potential. While originally designed to look at 1% weekly gains, and a second version that compares various percentage gains over the past year, I thought I might take a crack at modifying his original code to my specific goal - to find a set of stocks that will deliver 3% a week consistently across recent and historical time frames.
This was inspired also by a modification that Kevin McNulty (kmcninvest) had discussed that shoots for 5% gain a week over the past 13 weeks. This led me to think of viewing the timeframes like an exponential moving average, where the most recent performance is weighted most heavily - by being represented in all timeframes - and having the older data present but less important to the current stock ranking.
Here is my current version (comments welcome):
The stocks are sorted on their weighted average, with the highest scores being at the top. Looking at the top 5-10 picks, you would have hit your 3% a week goal every week for the past 13 weeks, and about 95% of the time over the last 6 months (admittedly, the last six months have been awfully good, so this success rate should be tempered by that fact).
I have never been able to deliver consistent results in any other approach I've tried, and it has the added advantage that your money is not constantly in play (usually less than 50% of the time, often you're in only for a few hours before hitting your target for the week).
Obviously, this can be modified to find short candidates - I'll post a shorting version in a bit. I also want to backtest this manually (since each week the stock selections might be different) to see how well it really did.
Kevin
|
Kevin_in_GA
4,599 posts
msg #80669
- Ignore Kevin_in_GA |
10/5/2009 8:06:16 PM
Simple change of conditions for shorts:
|
heypa
283 posts
msg #80671
- Ignore heypa |
10/5/2009 8:19:46 PM
Takes real stones to play this since the % loss possible can be pretty high.
|
Kevin_in_GA
4,599 posts
msg #80676
- Ignore Kevin_in_GA |
10/5/2009 8:47:22 PM
Statistics argue otherwise, unless there is a big correction in the near future.
Target hit 13 weeks straight by each of the top 10 stocks on the list.
25 of the the last 26 weeks this was hit by each of the top 3 stocks.
I respectfully disagree with your assessment, but am glad you provided input. Thanks!
|
gabestogie
79 posts
msg #80680
- Ignore gabestogie |
10/5/2009 9:22:33 PM
Kevin:
Did you play around with the IBD watchlist on this scan yet?
Might be worth a shot.
Gabe
|
trendscanner
265 posts
msg #80681
- Ignore trendscanner |
10/5/2009 9:29:24 PM
I generally understand the sorting concept of this filter but I don't understand how you would use this for trading or what would generate the buy and sell signals for these stocks. Would you simply put 3 or 4 of the top filtered stocks on a watch list and wait for an intra-day or multiday pullback for a buy?
|
gabestogie
79 posts
msg #80684
- Ignore gabestogie |
10/5/2009 9:58:02 PM
Kevin:
Ran this scan using IBD watchlist and the top 5 from the IBD list were in the top 10 running it without it.
Gabe
|
Kevin_in_GA
4,599 posts
msg #80686
- Ignore Kevin_in_GA |
10/5/2009 10:13:14 PM
I generally understand the sorting concept of this filter but I don't understand how you would use this for trading or what would generate the buy and sell signals for these stocks. Would you simply put 3 or 4 of the top filtered stocks on a watch list and wait for an intra-day or multiday pullback for a buy?
+++++++++++++++
No. It's really quite simple. You buy at the open on Monday (say, each of the top 3 stocks from the screen, which is run before the open). You put in a 3% profit stop.
That's it. Every week you do this. Statistically, you would have made 3% a week every week for the past six months doing nothing but these two steps. Often you are in and out by the end of the day on Monday - that's it for the week. No more trading needed.
Set it and forget it.
|
Kevin_in_GA
4,599 posts
msg #80687
- Ignore Kevin_in_GA |
10/5/2009 10:19:20 PM
Kevin:
Ran this scan using IBD watchlist and the top 5 from the IBD list were in the top 10 running it without it.
Gabe
++++++++++
Understood, but remember that this filter looks only for those stocks which are statistically favored to give a 3% spread from the weekly open to the weekly high - IBD ranking has nothing to do with it here, only statistics.
Makes sense that stocks such as FUQI and others are there - they have done so well that just buying and holding would have given you a better return than this approach over the last six months.
|
kmcninvest
43 posts
msg #80691
- Ignore kmcninvest |
10/5/2009 11:38:59 PM
This is not an IBD filter as Kevin has posted in another thread. This is from TRO's filter. I modified it thinking that the most recent on the list would be more accurate. So I made a 13 week filter with colums for the 26 and 52 to break ties. This filter only missed once from the middle of July (on stocks I picked). That was when CAR dropped down but since has come back. I like what Kevin has done with the weighted score. This is a pure statistics filter but you still have to do research on the stocks that are on the list.
|