|
PureBytes Links
Trading Reference Links
|
Ok Steve,
Your criteria for the dinner may not be able to be satisfied because in the
way Metastock compresses daily to weekly data. Frankly, I'm too lazy to
check, but I'll forego the dinner if you'll indulge me with a nice bottle of
wine after reading my comments below.
Hint: www.virtualvin.com ;-)
The method I provided for calculating the synthetic bars is VERY accurate if
you want a nice rolling 5 PERIOD data, i.e. every calendar Friday on the
daily bar chart showing weekly synthetic output will match the Friday weekly
close shown on a weekly compressed chart most, but not all, of the time.
Understand that the synthetics are not true bars but just the values. Values
can be plotted if needed, but I generally just call them in other formulae.
When I find a need to plot them I usually plot the TYPICAL PRICE instead.
Weekly compression used in Metastock is based on the calendar week, M-F.
Means anytime you have a holiday during the calendar week the compressed
weekly bar shown in Metastock DOES NOT reflect a full 5 days of trading.
Where my synthetics will show true O, H, L, and C over the last 5 trading
days, compressed weekly data in Metastock may reflect only 3 to 4 days
trading during holiday periods.
This "error" naturally throws off any indicators based on the data and will
account for occasional differences in indicator values between with my
synthetic weekly and Metastock's compressed weekly (especially if there are
large price movements around the period in question). FYI, its my
observation that these differences are not cumulative and values do
synchronize after a few weeks.
So which data is more accurate to use for trading? If your trading system
depends on a calendar week, like BUY S&P Monday on open, SELL Monday on
close, (you'd be surprised how well something this simple works!) then
Metastock's compressed weekly data is more accurate. But, with most other
systems the synthetics are very accurate and work better.
As for Paul and his formulae, they were a noble attempt but made you work
more difficult than needed. He gave you flexibility with his use of the
INPUT function, but this may only cause you headaches if you need to
reference this indicator in later formulae. Moreover, because he plots %K
and %D in the same indicator you get a stochastic that's impossible to read
since you can't differentiate between the lines, i.e.. same color same
style. BTW, he also erred in using a simple moving average to smooth %D
instead of an exponential.
Better to plot %K and %D using separate indicators, IMO. Take care when
referencing the %K formula in %D if you used the INPUT function as it will
only calculate %K with default parameters. Solution is to use another %K
formula in the %D indicator but just don't plot it.
As for making monthly synthetics, I've found using 21 bars of data works
well.
Hope this closes the competition,
Rick
Tokyo, Japan
-----Original Message-----
From: Steven Buss <sbuss@xxxxxxxxxxx>
To: Metastock-list <metastock-list@xxxxxxxxxxxxx>
Date: Tuesday, January 20, 1998 4:04 AM
Subject: The Value of Multi-period Indicators Illustrated
>Last week, I promised to provide a template that illustrated what I think
is
>valuable about Alexander Elder's multi-period perspective and why I've been
>making such a fuss about it. The attached graphic fulfills my promise.
>
>I've attached a small .gif that contains a chart of the S&P500 for most of
>1997 through Friday 1/16/98.
>
>Three indicator windows are included below the price chart.
>
>- The first indicator window contains three indicators calculated from
>daily S&P500 data.
>- The second indicator window contains the same three indicators
calculated
>from actual weekly (not synthetic daily) data.
>- The third indicator window contains the same three indicators calculated
>from actual monthly (not synthetic weekly or daily) data.
>
>The three indicators are:
>
>1 Default parm stochastic (5,3,3) ==>> Heavy red line with light blue ma
>and 20/80 lines light red.
>
>2 MACD based on an exponential ma (9,18,4) ==>> Heavy green with zero line
>light green.
>
>3 A Smoothed Rate of Change (5,3) as described in Elder's book "Trading
for
>a Living" ==>> Dark Blue with zero line light dark blue.
>
>I like using these three indicators because they mostly move together and
>can be used as confirmations of one another. In the text below, I'm going
>to reference the stochastic indicator only but in fact I mean below all
>three indicators taken together.
>
>Examine the price chart and the indicator windows together. Assess the
>quality of a daily stochastic breakout or breakdown above 20 or below 80
>with relationship to price in view of what the weekly stochastic is doing.
>
>Question 1: Isn't what the weekly stochastic is doing a better confirming
>indicator for the eventual quality of a daily stochastic breakout or
>breakdown than daily price itself? I think the answer is yes.
>
>Question 2: Would you like to create a Metastock exploration that provides
>you with the list of issues for which there has been a breakout (breakdown)
>of the daily stochastic and the weekly stochastic is above 20 (below 80)
and
>sloping up (down)? The reason for my initiating the "San Francisco Dinner
>Challenge" is that I think this would be a valuable thing to be able to do
>and it currently cannot be done without going through the kinds of
>shenanigans
>that Paul and Rick have gone through.
>
>You may remember that I said I thought that Elder's contribution was unique
>and valuable.
>
>Please note also that, at this writing, as far as I know, only Alexander
>Elder has a
>clear description of what this graphic illustrates. Rick has mentioned
that
>others
>discuss it. Do you have specific references for us Rick?
>
>More on Paul and Rick's work on the synthetic weekly stochastic from the
>daily data later...
>
>Steven Buss
>Walnut Creek, CA
>sbuss@xxxxxxxxxxx
>
>
|