siliconhippy
37 posts
msg #33493
- Ignore siliconhippy |
10/4/2004 7:10:11 PM
I ran the following filters, assuming that they would give me the same results, but they don't. Why?
What I did was take a simple Muddy filter, offset it by 1 day, and then mimicked the turn back by re-writing the filter as if I were looking at the filter 1 day ago.
siliconhippy
|
cegis
235 posts
msg #33495
- Ignore cegis |
10/4/2004 7:58:38 PM
siliconhippy,
You shifted the first and last clauses of the filter, but you didn't shift the average volume or close price parts (both need "1 day ago" added). So, the two filters are not equivalent...
HTH,
C
|
siliconhippy
37 posts
msg #33496
- Ignore siliconhippy |
10/4/2004 9:18:53 PM
cegis,
The "equivalent" filter for 1 day ago will then be:
Well, it still is not the same: produces exactly same results as above. And it shouldn't make a difference either:the avgVol and close price range should be universal over the filter range anyway.
Some other problem....
siliconhippy
|
defghca
150 posts
msg #33498
- Ignore defghca |
10/5/2004 7:08:21 AM
indeed, there seems to be a difference in the results from using "1 day ago" and "date offset is 1"
returns 3122 matches
returns 3191 matches
maybe Tomb should look into it or we should post in the bugs section
|
cegis
235 posts
msg #33499
- Ignore cegis |
10/5/2004 9:01:15 AM
siliconhippy,
> And it shouldn't make a difference either:the avgVol and close price range should be universal over the filter range anyway.
Huh?? Average Volume(90) will NOT be the same as average volume(90) 1 day ago. Average volume(90) will sum the volume over the past 90 days, then divide by 90; average volume(90) 1 day ago will sum the volume from 91 days ago through 1 day ago, and divide by 90. The only time these two values will be the same is if the volume 91 days ago is the exact same as volume yesterday.
Close price range will similarly be different than that 1 day ago (think a stock going from 4.99 to 5.01, 5.01 to 4.99, or similar "boundry" issues).
As for the current discrepancy, it's probably in the way you shifted "decreasing over the last 3 days". I'm not sure if "decreasing over" includes no change (your shift only checks for "below"). In other words, is a stock whose close over the past 3 days were 5.67, 5.66, 5.66 considered "decreasing over the past 3 days"? Also, I'm not sure if the phrase "HAS BEEN below" is the same as "IS below". You may be able to just use "close 1 day ago has been decreasing over the last 3 days" to do the shift.
One last thing, there's a Global Setting that will exclude stocks with zero volume today from your results. This can have the effect of different number of results between using "1 day ago" and "offset 1 day", since a stock that may have matched "close 1 day ago is below 6" might have had zero volume. This stock would then be eliminated from the "close is below 6 date offset is 1", because this phrase is executed as "close 1 days ago is below 6 and volume 1 day ago is greater than 0".
Or, maybe there's a bug in SF...
HTH,
C
|
siliconhippy
37 posts
msg #33500
- Ignore siliconhippy |
10/5/2004 10:20:44 AM
defghca and cegis,
Good comments, guys. We should let TomB sort it out.
cegis: what I meant was that small changes should not matter, and I got very different ticker results, with only 1-2 out of 6-8 each in common.
It would now help to have an updated manual with some remarks about such similar (but not same) filter phrases.
cheers,
siliconhippy
|