> A perfect system would be long for all postive return days and short for all negative return days.
I withdraw that statement. It is not true. An up day that was up by less than the average would result in negative detrended value, but would still be desirable to have long rather than short, from a gain standpoint. Mike --- In [email protected], "Mike" <[EMAIL PROTECTED]> wrote: > > I cannot comment on the suitability of either approach other than to > say that Aronson uses a fixed adjustment accross all bars in the > period under evaluation. The value of which was calculated from only > those bars being evaluated. > > Specifically, the average market return of the period being evaluated > is deducted from each bar, whereas using a moving average (or linar > regression?) would be subtracting a rolling average. > > The strategy returns are then calculated by adding the detrended > market return for the day, when long, or subtracting it, when short. > > The strategy returns are then compared to the market returns. A > perfect system would be long for all postive return days and short > for all negative return days. At least, that is how I interpretted > the book. > > Mike > > --- In [email protected], "wavemechanic" <timesarrow@> > wrote: > > > > How do you intend to use the detrended price? Usually it is used > in cycle work. Is that what you are after? > > > > I also wonder if there is much difference between Aronson's results > and the commonly used Close - (ma() or ref(ma()) or linear > regression) approach. > > > > Bill > > ----- Original Message ----- > > From: Louis Préfontaine > > To: [email protected] > > Sent: Sunday, March 02, 2008 10:42 AM > > Subject: Re: [AmiBroker] Re: Detruding... log > > > > > > Hi, > > > > The rules doesn't matter. The goal of the detrending is to > verify if a rule has a positive or negative bias. One has to > substract the average daily log of a stock from the daily log > difference between two days and then apply the rules to the results. > It sounds complicated, but I think it is worth the trouble! > > > > Louis > > > > > > 2008/3/2, Mr. Valley <valleymj@>: > > > > I have not read that book. > > What are the rules? > > > > -----Original Message----- > > From: [email protected] > [mailto:[EMAIL PROTECTED] Behalf Of Louis Préfontaine > > Sent: Sunday, March 02, 2008 8:02 AM > > To: [email protected] > > > > Subject: Re: [AmiBroker] Re: Detruding... log > > > > > > Hi, > > > > Do you think the two codes can do what Aronson recommands? I > mean: would I be able to backtest the detrended code to see how my > rules perform in it? If yes, how would you do it? > > > > Thanks! > > > > Louis > > > > > > > > 2008/3/1, Mr. Valley <valleymj@>: > > > > Like This? > > > > Detrend = Close * > > > > log (Ref(O,-2) / Ref(O,-1)) - log(MA(C/Ref(C,-1),BarIndex ()- > 1)); > > > > Plot > > > > (Detrend,"Detrended Returns",6,1); > > Plot > > > > (0.00,"",2,1); > > -----Original Message----- > > From: [email protected] > [mailto:[EMAIL PROTECTED] Behalf Of louisprefontaine > > Sent: Saturday, March 01, 2008 4:19 PM > > To: [email protected] > > Subject: [amibroker] Re: Detrending... log > > > > > > Anybody can help? > > > > Thanks, > > > > Louis > > > > --- In [email protected], "Louis Préfontaine" > <rockprog80@> > > wrote: > > > > > > I am trying to build a formula to "detrend" the market. > > > > > > What I want to set is something like this > > > > > > Close of day 0 * ( log (open day2/open day 1) - average > log > > return of > > > every day of the data available. > > > > > > Anybody can do that? > > > > > > Thanks, > > > > > > Louis > > > > > > > > > > > > > > > > > > > > > > > > > > > > > ------------------------------------------------------------------ -- > ---------- > > > > > > No virus found in this incoming message. > > Checked by AVG Free Edition. > > Version: 7.5.516 / Virus Database: 269.21.2/1304 - Release Date: > 2/29/2008 8:18 AM > > >
