|
PureBytes Links
Trading Reference Links
|
Conrad,
Yes, your suggestions would work for me. I presume you mean your comments to
apply to back- or forward- adjusted contracts.
>From: cb <cpbow@xxxxxxxxxxxxx>
>Reply-To: cpbow@xxxxxxxxxxxxx
>To: Jim Duc <firestar100@xxxxxxxxxxx>
>CC: omega-list@xxxxxxxxxx
>Subject: Re: Continuous vs Perpetual - which one is the best!
>Date: Sun, 13 Jun 1999 22:53:45 -0400
>
>J
> >
> > I think CSI agrees that negative numbers can be a serious problem
>depending
> > upon what traders are doing. Therefore, because of “the strong chance
>that
> > an inflation-sensitive market could produce negative price quantities
>into
> > the past” which “could discredit the accuracy” of the results, they
>offer a
> > subset of adjusted contracts they call Proportional Back Adjusted (aka
>ratio
> > adjusted). These are back- or -forward adjusted series that adjust “by
> > increasing or decreasing successively further distant contracts by a
> > percentage to raise or lower the entire history by the same proportion”
> > instead of by the better known actual price difference between one
>contract
> > and the prior older one. CSI says that “Proportionally adjusted series
> > prepared through ratio multiplications cannot go negative, so there is
>never
> > a need to elevate a series out of negative territory.” I have no
>experience
> > with this method of adjusting. Would someone who has care to comment?
>How
> > well does it work? How do your results differ (for better or worse) from
> > other methods?
> >
>
>I don;t think that the solution for negative numbers is proportionally
>adjusted contracts. This is a *different* type of result bec. it
>changes the bar heights and distances betw. bars, unlike addition
>adjusted contracts. As someone else suggested if your 20 yr back
>adjusted contract goes negative, try 2 10 year contracts.
>
>If you have to must have all the data in one file then, if negative
>results are obtained after creating the cont. ctct., the program should
>find the most negative number and add that value to the whole contract.
>Thus the properties of "gap adjustment by addition/subtraction" are
>maintained; (you just have had one extra addn/subtr). the only problem
>with this is that current prices on the latest contract will not be
>"real". For testing purposes that will not be a problem. You would
>probably have to actually trade from the current contract. I don't
>think this is a problem, but those who want to trade from the cont. ctct
>as well would not like it.
>
>Conrad Bowers
|