StockFetcher Forums · Filter Exchange · 5% A WEEK FILTER (BASED ON TRO'S CROCK POT)<< 1 ... 6 7 8 9 10 ... 35 >>Post Follow-up
Kevin_in_GA
4,548 posts
msg #83506
Ignore Kevin_in_GA
11/25/2009 8:26:03 PM

Hi Kevin,

I've been reading this column "silently" and decided to come out of the closet. I'm amazed at the algorithms you've come
up with. I'm curious how you were able to backtest the filter, since you only want to go in at the opening price on Monday, or
theoretically on a later day in the week as long as your cost is less than opening price and a price 5 percent higher has not since triggered.

How can I set the backtesting to only trigger on Mondays at Monday's opening price?

Thanks
Serge
++++++++++++++++

First off, I did not develop this concept or the basic filter code. TRO did, and I think it is a very clever approach. Originally designed for 1-2% a week, all I did was modify the code and apply some weighting factors.

This is not readily backtestable, as it needs to be run over the weekend. However, you can use the offset function, as I did, and look back (use the data from each Friday to test the filter for the following week).

sdadesky
4 posts
msg #83534
Ignore sdadesky
11/27/2009 11:51:41 AM

Thanks Kevin,
I'll read the Crockpot thread. Kudos to both you and TRO then.
Serge

Kevin_in_GA
4,548 posts
msg #83537
Ignore Kevin_in_GA
11/27/2009 12:04:17 PM

No problem, Serge.

You may want to also look at the 5% short filter in this thread as well. Could be useful if the markets are actually tanking as we speak.

Kevin_in_GA
4,548 posts
msg #83595
Ignore Kevin_in_GA
11/29/2009 6:21:28 PM

Next week's plays, in descending order, are:


APWR
performance 19.50
reward 3.00


ZLC
performance 3.62
reward 2.92


PWER
performance 37.00
reward 2.85


RINO
performance 18.50
reward 2.85


IVAN
performance 18.50
reward 2.85


Kevin_in_GA
4,548 posts
msg #83692
Ignore Kevin_in_GA
12/1/2009 5:39:50 PM

PWER hit it's 5% target today.

Kevin_in_GA
4,548 posts
msg #83693
Ignore Kevin_in_GA
12/1/2009 5:47:30 PM

And on a less positive note, ZLC and APWR are both tanking, down 5% or so. Unlikely that they'll run up 10% between now and Friday.

Radiomuse
288 posts
msg #83788
Ignore Radiomuse
modified
12/3/2009 1:07:33 PM

Kevin, I really like this idea, just wondering about you came up with the 5% target and 10% risk thresholds? Also, the weighting scheme seems really heavy on the performance over the last 4 weeks, hardly affected at all by the 13 week. I was messing around a bit - using 45% weight on the last 4 weeks, and 55% on weeks 5 thru 13 (not all of the last 13 weeks). I'm having 2 "issues"...

- I can't figure out how the backtester works (especially how to get it to only select the top 5 stocks based on a particular variable)
- not a big deal, but I can't get the "sort by column 6 (or whatever) descending" to work

Fetcher[/*MEASURE THE DIFFERENCE FROM WEEKLY OPEN TO WEEKLY HIGH*/
set{whiop, weekly high - weekly open}
set{Long_Profit, whiop / weekly open }

/*DETERMINE THE FREQUENCY THAT 5% IS ACHIEVED OVER MULTIPLE TIME FRAMES*/
set{5_1wk, count(Long_Profit > .05,1)}
set{5_4wk, count(Long_Profit > .05,4)}
set{5_13wk, count(Long_Profit > .05,13)}
set{5_nextnine,5_13wk-5_4wk}

/*APPLY WEIGHTING FACTORS*/
set{var1,5_4wk*0.45} /*WEIGHT 4 WK AVERAGE AT 45%*/
set{var2,5_nextnine*0.55} /*WEIGHT 13 WK AVERAGE AT 55%*/

/*CREATE A WEIGHTED SUMMATION OF STOCK PERFORMANCE (MAX SCORE = 3)*/
set{reward,var1+var2}

/*DETERMINE THE RISK OF TRIPPING A 10% STOP LOSS*/
set{wloss, weekly open - weekly low}
set{max_loss, wloss / weekly open}

/*DETERMINE THE FREQUENCY THAT A 10% STOP LOSS IS TRIPPED OVER MULTIPLE TIME FRAMES*/
set{loss_1wk, count(max_loss > .10,1)}
set{loss_4wk, count(max_loss > .10,4)}
set{loss_13wk, count(max_loss > .10,13)}
set{loss_nextnine,loss_13wk-loss_4wk}

/*APPLY WEIGHTING FACTORS*/
set{var11,loss_4wk*0.45} /*WEIGHT 4 WK AVERAGE AT 45%*/
set{var21,loss_nextnine*0.55} /*WEIGHT 13 WK AVERAGE AT 55%*/

/*CREATE A WEIGHTED SUMMATION OF RISK (MAX SCORE = 3)*/
set{risk,var11+var21}
set{performance,reward/risk}

add column performance
add column reward {reward}
add column risk {risk}
add column 5_4wk {4 wk reward}
add column 5_13wk {13 wk reward}
add column 5_nextnine {wk 5 to 13 reward}
add column var1
add column var2
add column loss_4wk {4 wk risk}
add column loss_13wk {13 wk risk}
add column loss_nextnine {wk 5 to 13 risk}
add column var11
add column var21

/*SCREEN FOR THE STRONGEST CANDIDATES*/
weekly open is above 5
average volume(90) above 500000
and reward is greater than 5.0
and risk is less than 2.5
sort by column 6 descending

and draw 5_1wk
and draw loss_1wk on plot 5_1wk
and do not draw reward

chart-display is weekly

/* Global Filter Text */


/* Global Filter Text */
]



Kevin_in_GA
4,548 posts
msg #83790
Ignore Kevin_in_GA
12/3/2009 2:04:05 PM

Sort by column 6 descending seems to be working when I run the filter.

I manually backtested this - choose the results for a specific Friday, and those stocks are used to determine performance for the following week. It basically mimics the idea of only running the scan on the weekend.

Then you need to see how each of the top stocks performed in that following week. If they hit the 5% target, if they hit a 10% stop loss, or just flatlined.

I picked a 10% stop loss somewhat arbitrarily, but based it on how the top stocks crashed and burned over the past 52 weeks. At 10%, you had a lower drawdown in total over the 1 yr period than 5% or no stop loss.

In practice, you need at least two 5% winners to offset one 10% stop out, which is hard in anything but a strongly up trending environment.

I did heavily bias the most recent 4 week performance, but it was because I felt that this had a momentum component to it that should be incorporated. High past volatility should not be weighted as highly as current preformance.

All of the filters I write and post are done with the full expectation that they will be modified and improved. please fell free to adapt to yourspecific needs, and share your results as you see fit.

Kevin

P.S. having gone through the forums quite a bit before starting to post anything, I'm glad to see that you are back at SF. I look forward to more interaction and discussion with you.


Radiomuse
288 posts
msg #83791
Ignore Radiomuse
12/3/2009 2:42:18 PM

Thanks Kev - I definitely hear what you're saying about placing more bias on the most recent 4 weeks performance. I'm going to experiment with some weights in Excel, in particular, I want stocks that have hit the 5% threshold in 3 of the last 4 weeks to pass if their record over the previous 9 weeks is really good. Might even use something different than the 4 and 13 week time periods (thinking 5 and 15). I'll let you know what I come up with.

Kevin_in_GA
4,548 posts
msg #83792
Ignore Kevin_in_GA
modified
12/3/2009 3:10:59 PM

I will say that, over the last 4-6 weeks, this approach has not been stellar. The current choppiness and sideways movement has robbed a lot of previous winners of their momentum (e.g., GNW, FUQI, CNO, MED). These guys rocked the house until late September, and have struggled with chop and profit taking since.

This week the top 3 will have only one winner, leading to a net loss if you tried to spread your risk at all.

There is also a 5% short filter on this thread as well, in case you have any interest in looking at that as well.

StockFetcher Forums · Filter Exchange · 5% A WEEK FILTER (BASED ON TRO'S CROCK POT)<< 1 ... 6 7 8 9 10 ... 35 >>Post Follow-up

*** Disclaimer *** StockFetcher.com does not endorse or suggest any of the securities which are returned in any of the searches or filters. They are provided purely for informational and research purposes. StockFetcher.com does not recommend particular securities. StockFetcher.com, Vestyl Software, L.L.C. and involved content providers shall not be liable for any errors or delays in the content, or for any actions taken based on the content.


Copyright 2016 - Vestyl Software L.L.C.Terms of Service | License | Questions or comments? Contact Us
EOD Data sources: DDFPlus & CSI Data Quotes delayed during active market hours. Delay times are at least 15 mins for NASDAQ, 20 mins for NYSE and Amex. Delayed intraday data provided by DDFPlus