[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Rollover workarounds for TS?



PureBytes Links

Trading Reference Links

At 10:22 PM -0500 11/22/99, Jim wrote:

>     One contract ends...a new one
>starts...PLEASE Quit trying to tie them
>together. It makes NO sense! Really.! Just
>think about it.

At 12:22 PM -0500 11/25/99, Randy wrote:
>
>There someone has finally said it. I have traded the SP and bonds for
>years without any thought of a continuous contract. Jim is correct.


At 11:30 AM -0600 11/25/99, Stewart Taylor wrote:

>Well I have traded for years using continuous contracts (but Im not a
>systems trader). In the context of my experiance and trading style Jim is
>incorrect.
>
>In the context of his experiance and trading style, Jim is 100% correct (as
>are you).
>
>There are lots of ways to skin the cat, lets just not assume that we have
>the only way.


"Continuous contracts" are simply a convenience to allow backtesting 
over a longer period without having to test and optimize on each 
individual contract. There is nothing wrong with using them if you 
construct the continuous contract correctly. But "correctly" depends 
upon what characteristic of the prices your trading system uses.

If your system looks at the local relationship between 
opens/highs/lows/closes of the bars (as most systems do) then the 
pattern of trades will not be affected by adding or subtracting a 
constant from all prices.

You can easy test your system to make sure this is true. Simply 
replace "Open" by "Open + 50", "Close" by "Close + 50", "High" by 
"High + 50" and "Low" by "Low + 50", using some constant (50 in this 
example) for all prices.

If the trades are unchanged then back adjusting older contracts by 
adding/subtracting a constant from the prices will not affect the 
trades it makes. Thus, if you construct the continuous contracts by 
adding/subtracting a constant from old prices, you will get the same 
trades using continuous contracts as you would have gotten by testing 
using the individual contracts.

If your system holds a position through a rollover period, of course, 
this will be different than you would have done with the real 
contracts. But remember, this is just a simulation using old data 
which will never again repeat. So the differences in back testing 
during the rollover period are probably less of a factor than the 
differences between current to past market behavior.

Bob Fulks