| Author |
Messages |
|
jfrance
Posts:5
 |
| 06-16-2006 9:36 PM |
 Alert
|
| I've always used reverse or back adjusted data...I'm swithcing to Pinnacle for my futures data. They have a ratio adjusted method available. I'm trying to decide which to go with...anyone have any experience, information, or opinion about the differences between these methods or which might be more suitable?
Thanks! |
|
|
|
|
murray
Posts:435
|
| 06-16-2006 9:51 PM |
Alert
|
| I still always use backadjusted |
|
|
|
|
gateway
Posts:12
|
| 07-04-2006 1:32 PM |
Alert
|
You might want to see two articles which deal with your question. The first is "Selecting the Best Futures Price Series For Computer Testing" by Jack Schwager, in Technical Analysis of Stocks & Commodities Magazine, Oct 1992, page 65.
The second is "Data Pros and Cons" by Thomas Stridsman, in Futures Magazine, June 1998, page 54. If you cannot find these articles, I can send them to you.
Mark Jurik was asked the same question, and his reply follows:
"The method that appears most sound (mathematically) requires manual labor. First, take the log of all prices for each contract. Next, make a continuous contract using reverse adjustment. Finally, take the inverse log of the continuous price series. The prices at the very beginning will be very small, relative to current prices, but who cares? For that time frame, all price movements are proportionally correct. Just keep in mind that all price related constants and thresholds at any point in time will need to be properly scaled (automatically). This forces one to think of all parameters in terms of market dynamics, rather than fixed values. A real good exercise.
Or you could leave the prices in log space. However, the time series early on will likely have negative values, including zero. Thus, all indicators need to be able to correctly handle this. (Ours do)."
Don't ask me to explain any of that; I haven't a clue what it means. Hope this helps. |
|
Precision Guesswork, Low Rates |
|
|
gateway
Posts:12
|
| 07-05-2006 1:48 PM |
Alert
|
You might want to see two articles which deal with your question. The first is "Selecting the Best Futures Price Series For Computer Testing" by Jack Schwager, in Technical Analysis of Stocks & Commodities Magazine, Oct 1992, page 65. You can purchase it from TASC for $4.95
http://store.traders.com/articles-volume-10-chapter-10.html
The second is "Data Pros and Cons" by Thomas Stridsman, in Futures Magazine, June 1998, page 54. If you cannot find this article, I can send it to you.
Mark Jurik, of Jurik Research, was asked the same question. His reply follows:
"The method that appears most sound (mathematically) requires manual labor. First, take the log of all prices for each contract. Next, make a continuous contract using reverse adjustment. Finally, take the inverse log of the continuous price series. The prices at the very beginning will be very small, relative to current prices, but who cares? For that time frame, all price movements are proportionally correct. Just keep in mind that all price related constants and thresholds at any point in time will need to be properly scaled (automatically). This forces one to think of all parameters in terms of market dynamics, rather than fixed values. A real good exercise.
Or you could leave the prices in log space. However, the time series early on will likely have negative values, including zero. Thus, all indicators need to be able to correctly handle this. (Ours do)."
Don't ask me to explain any of that; I haven't a clue what it means. Hope this helps. |
|
Precision Guesswork, Low Rates |
|
|
|