|
PureBytes Links
Trading Reference Links
|
Well, I'll certainly take a shot at this. First of all, if you want to
look at the same price series with varying resolution P&F charts, the
place to go is Murphy's "Technical Analysis of the Futures Markets".
Please note that there is a new edition of this book called "Technical
Analysis of the Financial Markets" (I think) which may not still have
this display.
Let's start with a simplified version of the algorithm. Let's assume
that we are using a single price for each time period rather than the
high and low. That way we can chart anything that has a varying price
over time. And let's fix the reversal size at three boxes, so the only
tunable parameter we have to worry about is the box size in ticks. A
tick is the smallest possible price increment; for a stock that trades
in eighths, it's an eighth, for T-bond futures it's 1/32, for corn
futures it's 1/4 cent, and so on.
Let's start with the smallest possible box size, one tick, and with a
very fast sampling rate, say once a minute. Let's assume we are
currently plotting X's (moving up). The algorithm says: "If the next
price is up at least one box, plot X's up to the next price. If the next
price is down at least three boxes, move right a column and plot O's
down to the next price, starting one box below the top of the column of
X's we just finished. If neither of these is true, don't plot anything."
I think it's pretty obvious that you would get a lot of columns this
way, and that this would give you a very detailed picture of the
movement of the price series. The only finer resolution you could get
would be if you sampled every trade and reversed every time the
direction changed. But let's just stick with our one-minute samples and
one-tick boxes for now. But let's plot a whole week's worth of trading
this way. What you will get is a very wide chart with a lot of columns.
Now imagine you've got a big wide wall to put this chart on. Stand close
to the chart so you can only see ten or fifteen columns. Then slowly
walk backwards until you can see the whole week. From a distance, the
columns tend to merge together, and you can see where the trend lines
(bullish/bearish support/resistance) would be.
Now let's triple the box size but leave everything else the same. You
don't need as wide a wall to plot the week of trading, you don't get as
many reversals, and you don't have to step as far back to see the whole
chart. But I contend that the major features -- the trend lines, the
support and resistance levels -- are going to be more or less the same
as they were on the one-tick box chart.
But as we keep increasing the box size, giving a coarser and coarser
resolution on the chart, eventually we're going to be down to a single
column! That column will represent a whole week's worth of trading. If
you keep up the chart, eventually you'll have a chart of one year or
five years of activity with multiple columns. You'll have trend lines
and support and resistance levels on a different time scale.
So what's the "right" box size? It depends on what time scale you want
to trade! If you're a futures day trader, you're probably going to use a
one-tick box, plot every trade that comes across the wire and enter
short when you see a resistance level form or long when you see a
support level form. You'd exit on a stop based on how much you're
willing to lose, when you saw the opposite level form (resistance if
you're long, support if your're short), or at the close, whichever came
first. If you're a long term stock trader, you'd probably use a daily or
weekly chart and have a box size of 1/3 what you would consider a
significant reversal, or use the default box sizes for stocks that have
been handed down by the ancients (Cohen, Burke, Tom Dorsey). You'd
probably only trade long because of margin constraints, so you'd only
trade stocks that showed bullish patterns on their charts.
Still, you shouldn't need multiple charts for a single tradeable. The
box size should be chosen appropriate to your trading time scale and
your trading decisions made on the basis of the patterns on that single
chart.
--
M. Edward Borasky znmeb@xxxxxxxxxxxx http://www.teleport.com/~znmeb
If God had meant carrots to be eaten cooked, He would have given rabbits
fire.
-----Original Message-----
From: gary bodnar <gbodnar@xxxxxxxxxxxxxxxx>
To: RealTraders Discussion Group <realtraders@xxxxxxxxxxxxxx>
Date: Sunday, February 07, 1999 11:08
Subject: RE: GEN: Why P&F Box Sizes?
>Hello all
>
>Maybe one of you P&F expert can explain why two different box sizes
would
>show something different, for example, box size at .10 cent would be
>bullish while box size @.15 would be bearish.. In other word,
>the .10 cent box is showing the X signals while the .15 cent box is
showing
>the O signals.. In all I assume one should keep track in different box
>sizes in order to have a bigger picture of the price movement??
>
>Respectfully yours
>Gary
>
|