|
PureBytes Links
Trading Reference Links
|
Hi, folks:
I think that RobinHood's idea of normalizing the MACD is a very
fruitful thing to consider, if you are doing Explorations to "rank"
the relative values of MACD's (or their Histograms) across a large
group of stocks.
As a reminder, here are the relevant formulae:
MACD = FastEMA(C) - SlowEMA(C)
Histo = MACD - SignalEMA(MACD)
This presumes that the nature of your system is somewhat different
than the "classical" MACD signal logic. The classic approach is to
look for cases where the MACD (or its Histo) has crossed the zero
line (one way or the other).
If you use this approach, then there is no need for normalization,
since 0=0=0 regardless of the price range of the stock. Or, as the
Spanish would say, "all cats are gray at night".
However if you want to use the MACD (or its Histo) to comparatively
rank the "trend speed" (MACD) or "trend acceleration" (Histo) of the
stock's price action, then normalization of some sort IS required.
The "units" of the MACD are "change in dollar value". Most trading
decisions need to be made on dollar-value change VERSUS the dollars
at risk (where risk = stoploss related, equity related, or both).
Let's say, for example, we prequalify our trades by finding stocks
whose Histo's have just crossed up through the zero line. That does
not require normalization.
Now let's presume (for discussion) that amongst those recent
crossovers, we think that stocks which have a steeper MACD slope
offer better trading opportunities than ones with gentle slopes.
Keep in mind that a steep MACD slope indicates the FastEMA is
separating quickly from the SlowEMA.
In that case, we need to SORT the results of the Exploration based on
a column with a formula something like MACD(today)-MACD(yesterday).
Higher values represent faster increases in dollars per day.
However a change of 0.10 per share per day in the FastEMA-SlowEMA of
a $1.00 stock is much more significant to our pocketbook than that
same 0.10 change for a $100 stock ... we might typically own 100
shares of the $100 stock, but would have 10,000 shares of the $1
stock!
Thus the need for normalization. A comparison like that is better
done by first finding the change-in-dollars-PER-COMMITTED-DOLLAR,
then doing the Exploration Sort. That is, a comparison of 0.10/$1
verus 0.10/$100 would provide us a more useful metric.
One way to do this normalization is to first calculate the MACD, then
divide by the most recent price:
( Mov(C,FastMA,E) - Mov(C,SlowMA,E) ) / C
However to get a true normalization, we should use the
same "reference base" in the denominator as we use in the numerator:
( Mov(C,FastMA,E) - Mov(C,SlowMA,E) ) / Mov(C,SlowMA,E)
Either of these approaches will provide a metric for comparison
across multiple stocks that is useful for determining which has a
higher "trend speed". I personally prefer the second of the two.
A similar argument can be made for comparing the "trend acceleration"
differences between stocks, by normalizing their Histo values.
Jim Dean
------------------------ Yahoo! Groups Sponsor ---------------------~-->
Buy Ink Cartridges or Refill Kits for your HP, Epson, Canon or Lexmark
Printer at MyInks.com. Free s/h on orders $50 or more to the US & Canada.
http://www.c1tracking.com/l.asp?cid=5511
http://us.click.yahoo.com/mOAaAA/3exGAA/qnsNAA/zMEolB/TM
---------------------------------------------------------------------~->
To unsubscribe from this group, send an email to:
Metastockusers-unsubscribe@xxxxxxxxxxx
Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
|