olathegolf
119 posts
msg #90593
- Ignore olathegolf |
3/31/2010 12:54:18 AM
Has anyone experienced problems indicating that data used by SF (for penny stocks) does not contain sufficient significant digits? I was playing around with simple filters for stocks less less than 0.1 and noticed some peculiar "Change %" reported by SF. Upon further inspection, it appears that the data may be rounding off to the nearest 100th instead of nearest 10,000th. It's also possible that the data stream received by SF does not contain this granularity. That makes a huge difference when you calculate the percentage change from the previous days close.
Run the following:
[price < 0.1
close is 50% below close 1 day ago
volume > 50000]
For March 30, 2010, I get 33 hits.
I looked on http://www.google.com/finance and checked each of the stocks and determined the following:
1. 7 stock "Change%" reported by SF are correct.
2. 26 stock "Change%" reported by SF are incorrect.
I've run many backtests with penny stocks and I believe a large percentage of these are in error.
I hope I am missing something here. I have not gotten a satisfactory response from SF to explain this. Please review and advise if you can shed some light on this.
Thanks.
|
olathegolf
119 posts
msg #90631
- Ignore olathegolf
modified |
4/1/2010 3:29:52 AM
Apparently SF has fixed this issue. I have not formally heard from them but the 33 stocks retrieved previously has been reduced to 11. The 11 appear to be correct. Thanks.
Update: I received the following response from SF, 4/1/10:
"Thank you for the feedback. As indicated previously, the data precision issues you highlighted, with respect to our end-of-day updates, have been resolved. Additionally, as evidenced by your results, StockFetcher does receive and employ floating point precision up to 4 significant digits for pricing data."
|
moby
314 posts
msg #90636
- Ignore moby |
4/1/2010 8:42:08 AM
|