mharrisnj
4 posts
msg #158563
- Ignore mharrisnj |
2/9/2022 4:56:09 PM
I found a response to a user question about how to back-test a filter beyond the 250 day limit when using the "date offset" keyword.
==================================================
Q: Is there a limit to 'date offset'?
A: Yes. The "date offset" feature is limited to approximately 1 year of historical data. For testing prior periods, we recommend using the "Backtesting" feature.
==================================================
You can also see it here: https://www.stockfetcher.com/help/Is-there-a-limit-to-date-offset/100391
Can anyone shed some light on how to access this feature? As I mentioned, the "date offset" keyword can be used in a filter, but it only goes back about 1 year. For some filters, a single year can not give a good measure of the expected performance over time..
|
nibor100
1,031 posts
msg #158575
- Ignore nibor100 |
2/11/2022 1:27:22 PM
If you look at the Announcements Forum you'll find SF discontinued their Backtesting feature in 2015.
However, you can run any filter in SF up to 999 days in the past using the drop down box inside the date box.
Note, you will not be able to go back that far on a normal SF chart because they are limited to 3 years of history.
Hope this helps,
Ed S.
|
mharrisnj
4 posts
msg #158637
- Ignore mharrisnj |
2/21/2022 9:51:06 AM
Hi @nibor100. Thank you for the insight and for indulging a noob.
Having never seen or used the back-testing feature, I am not really sure what has been disabled.
For example, both of these yield the same results
===========================
/*temp*/
offset 0
close 100 days ago below 2000
add column close 100 days ago
sort column 5 descending
===========================
and
===========================
/*temp*/
offset 100
close below 2000
add column close
sort column 5 descending
===========================
However, the offset keyword is only good for looking 250 days in the past whereas the nnn days ago can look much farther. Neither of these methods is very useful without a way to automate decrementing the look-back number and gathering results. Was the back-test feature of old able to provide tables of results spanning a given time period?
|
nibor100
1,031 posts
msg #158661
- Ignore nibor100 |
2/23/2022 5:27:34 PM
@mharrisnj,
While it doesn't really help you no matter what SF's backtesting feature used to entail, the answer to your question is I don't know as I subscribed to SF long after they had discontinued their specialized backtesting feature.
If you search on backtest(ing) you'll probably find old posts about that feature and some of the issues members were having with it back then.
SF is not the only stock service that found backtesting is a bit harder to implement properly than one might think.
Ed S.
|
marine2
963 posts
msg #158662
- Ignore marine2 |
2/24/2022 12:39:06 PM
With SF’s former back testing feature once you learned its nuances it did provide you enough info on how well your filter performed over time. SF did not remove the back testing tool because of its complexity / problems they removed it because it was costing them too much to maintain. We the subscribers pleaded with them to keep it for its higher subscription service and charge accordingly but to no avail. They cut the tool out and let you go to another subscriber service for that nifty tool. Anytime you have filter creation your service should also provide a tool to test it over a durated period of time. SF in this regard failed us.
|
mharrisnj
4 posts
msg #158673
- Ignore mharrisnj
modified |
2/25/2022 1:24:39 PM
Agreed it is an important tool. I don't really see how anyone could trade a strategy without knowing how it performs over time...seems like the most basic requirement.
I trade on tradestation and their tools only let you backtest a strategy once you pick the stock or other instrument. It does not help you pick a stock. I guess I have to widen my search for the right tool. Thanks for your insights.
|
nibor100
1,031 posts
msg #158676
- Ignore nibor100 |
2/25/2022 3:25:29 PM
Edgerater and Amibroker and StrataSearch(discontinued) are the most robust backtesting tools that I've used though none of them, nor any other backtest sw I've tried covers as many stocks as the 15,000+ stocks that SF can filter thru.
Depending on what you are after there are some filter adjustments in SF that can be done for limited backtest results.
Ed S.
|
mharrisnj
4 posts
msg #158715
- Ignore mharrisnj |
3/3/2022 1:32:37 PM
Thank you Ed. I am looking into Edgerater and Amibroker.
|