Kevin_in_GA
4,599 posts
msg #114789
- Ignore Kevin_in_GA
modified |
8/7/2013 4:02:54 PM
Jimmyjazz: Hard to do for two reasons:
1. There is no way to set the day of the week in a filter. I have asked SF for this (a function such as dayofweek() which would return a number from 1 to 5 for each trading day) but no response on this to date.
2. What if the weekly high was on Monday? Now it is Tuesday but you are trying to see if the weekly high will be hit ... this would screw up any backtesting.
A variant of this would be to look at each day of the week to see if - for that day - a similar set of stats could be generated. But you would still need to have a dayofweek() function or equivalent, which currently is not an option.
|
jimmyjazz
102 posts
msg #114791
- Ignore jimmyjazz |
8/7/2013 6:53:14 PM
I haven't dug through the code logic -- are you using weekly data (i.e. weekly open/close/high/low) or are you using daily data parsed in M-F chunks? If the former, then I agree the filter would be tough to write.
If the latter, then I guess it comes down to the ability to scan historical data in Tue-Mon batches.
Anyway, great filter, and much appreciated. It has made me a nice consistent return over the past couple of months.
|
Eman93
4,750 posts
msg #114793
- Ignore Eman93 |
8/8/2013 12:29:13 AM
I just saw this and is very interesting... nice work as always Kevin.
You can only run it like 15 min after the close on Friday and until Monday open, to get true weekly data, any other time you are getting the last 5 trading days of info.
|
jimmyjazz
102 posts
msg #115476
- Ignore jimmyjazz |
9/23/2013 8:21:24 AM
Can someone explain the results on this filter from this weekend, where DRYS showed a performance of "0.0" having had a 4-week reward:risk ratio of 4:0 and a 13-week ratio of 11:0? "Reward" was 2.85. Am I not understanding "performance"?
|
Kevin_in_GA
4,599 posts
msg #115477
- Ignore Kevin_in_GA |
9/23/2013 8:50:02 AM
Must be a glitch in how SF handles a division by zero.
|
jimmyjazz
102 posts
msg #115480
- Ignore jimmyjazz |
9/23/2013 9:03:25 AM
What "should" DRYS performance be?
|
jimmyjazz
102 posts
msg #116270
- Ignore jimmyjazz |
10/26/2013 10:26:49 AM
Something interesting happened after trading this week: NEWL triggered both the long and short versions of this filter.
Long: Performance = 1.06 (Reward = 2.92, Risk = 2.77)
Short: Performance = 2.06 (Reward = 2.85, Risk = 1.38)
|
pharmacist.umang
3 posts
msg #124322
- Ignore pharmacist.umang |
7/9/2015 11:15:51 PM
Does this filter still works Kevin? Please share your latest experience with this.Cheers
|
Kevin_in_GA
4,599 posts
msg #124324
- Ignore Kevin_in_GA |
7/10/2015 8:38:49 AM
I have not been using this filter for quite a while so I can't comment on its recent value. This was more of an expansion of prior work done by TRO on this idea.
|
pharmacist.umang
3 posts
msg #124344
- Ignore pharmacist.umang |
7/10/2015 8:55:13 PM
Where is 'LIKE" button Kevin...? where is your filter for LIKE? Thank you.
|