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

Re: Trendiness or Congestion measurement - r-squared + Linear regression Slope



PureBytes Links

Trading Reference Links

Al,

(1) What do you mean by a "13 period slope"? A moving average? Or something 
else?

(2) I trade primarily Cdn stocks. What exchange is DD? I'd like to download 
it and see what you're talking about.

(3) Could you give an example of this trading technique with going long?

Thanks.

Bob.


At 09:56 AM 05/02/01 -0600, you wrote:
>Yes, I think of Slope as the direction.  The value indicates the degree of
>advance/decline.  R-squared is a measurement or "presence of trend", as I
>use it.  I just completed a trade on DD that could paint a picture to aid
>in interpretation.  If you will pull up the chart DD, open a 23 period
>r-squared indicator
>(I use the oscillator style of display), then overlay a 13 period slope
>(line display), observe the period around 12/29.  You will see how the
>slope began the decline while "r" was still at a high value.  When "r"
>began to weaken, with Slope still in a downward movement, the first close
>below the lowest low of that period (1-3) made a low risk entry for a short
>sale.  When Slope turned up for two consecutive days (1-24) and "r" was
>above 80, this was, for me,  time to cover the short.
>
>I am also using a 23 period slope to observe the 13/23 crossing, but I have
>not seen any significant benefit.  Actually, I think this confuses the
>picture.
>
>As to your comment that Slope moves all over the place, that would suggest
>to me that  there is very high volatility in that stock,or a tradable trend
>is not in place.  Please send me a stock symbol so that I may plot what you
>are seeing.  I am watching TXN, but the Slope signals have been late.  But
>here again, there is no trend, as I see it.  Rather, it is in a trading
>range.  Again, this is simply way that I use the two indicators.
>
>Al Taglavore
>
>----------
> > From: Sean Taylor <sean.taylor@xxxxxxxx>
> > To: metastock@xxxxxxxxxxxxx
> > Subject: Re: Trendiness or Congestion measurement - r-squared + Linear
>regression Slope
> > Date: Sunday, February 04, 2001 11:24 PM
> >
> > Al,
> > I've been trying to combine r-squared with Linear Regression Slope.  Not
> > always easy determining how to interpret them.
> > I know r-squared values below a certain amount mean there is no
>statistical
> > significance. But with larger numbers, the Slope value can still move all
> > over the place.  Is it more the direction of the slope rather than the
>value
> > itself?
> > It's probably just trial and error.
> > Thanks,
> > Sean
> >
> > ----- Original Message -----
> > From: "Al Taglavore" <altag@xxxxxxxxxx>
> > To: <metastock@xxxxxxxxxxxxx>
> > Sent: Friday, February 02, 2001 11:02 AM
> > Subject: Re: Trendiness or Congestion measurement
> >
> >
> > > Consider using 'r-squared' with regression slope; "r" indicating the
> > > strength of the trend, sans noise, and the slope to indicate the
>direction
> > > of the trend.
> > >
> > > Al Taglavore
> > >
> > > ----------
> > > > From: Sean Taylor <sean.taylor@xxxxxxxx>
> > > > To: Metastock List <metastock@xxxxxxxxxxxxx>
> > > > Subject: Trendiness or Congestion measurement
> > > > Date: Friday, February 02, 2001 9:09 AM
> > > >
> > > > I've recently re-discovered the quality of the User manual.  The
> > > > interpretations of various indicators offer useful advise.
> > > > Take the Stochastic. I'm looking to use Overbought/Oversold
>indicators.
> > > > The manual recommends that these work best in non-trending markets
>and
> > > that
> > > > 'r-squared', CMO and VHF might be used to determine this.
> > > > I tried r-squared  (RSquared(C, 14) <0.27) which improved the system
> > > results
> > > > significantly.  But it does seem to traverse extremes very readily
>and
> > > give
> > > > false signals.
> > > >
> > > > Are there some favorites out there for indicating congestion or
> > > trendiness
> > > > and their strength ?
> > > >
> > > > Thanks,
> > > > Sean
> > >


==============+==============
Robert L. Webb
Webb.Bob@xxxxxxxx
===============+=============