|
PureBytes Links
Trading Reference Links
|
As the developer of software that has emphasized MFI for over a decade,
thought I'd put in my two cents.
Dtrader stated:
>You are correct in stating that the concept was "invented long ago by
someone else" ... assuming we believe that such things are
actually"invented" as opposed to just being observed and noted.<
>the first public observations and notes that I am aware of were written by
Richard Wyckoff.
He noted that heavier volume with little price movement characterized
turning points in the market.<
My observations:
The underlying concepts were also understood by the inventor of Equi-Volume
charts (Arms?), which have been around a long time. These are bar charts
that vary in width proportionately to volume. A "squat" appears on these
charts as a short fat bar (a "go" is a tall fat bar, etc.).
In the mid to late 80's Bill Williams publicized the concepts in his
Profitunity Indicator work and used the terms "squat", "go", "fade" and
"fake". There were a lot of letters to the editor of TASC between Bill and
Charlie Wright as to who was the inventor. Charlie MAY be the unknown
trading guy at Omega referred to earlier in this thread.
My software displays these indicators plus one of my own -- a "LOOK". This
is a bar with increased volume but no significant change in the MFI. It is
quite common on the S&P which often zigs on volume surges without changes in
what I call the "character" of the market (meaning the MFI).
All published literature on MFI that I have seen misses the true value of
these indicators. Specifically, a squat is defined as an increase in volume
and a decrease in the MFI. (the other signals also deal with simple
increases or decreases in the volume and MFI). I learned a decade ago that
this creats some kind of signal on almost EVERY bar. My software looks for
"significant" changes in volume and "significant" changes in MFI. This
results in far fewer, but more reliable signals.
A few observations (don't have time to do a treatise now, but if there is
enough interest, I could write something after the holidays):
- In addition to significant changes in volume/MFI, you should have
high volume in and of itself for the particular instrument. I call it
"who's voting?". If the volume is light, presumably the locals are
controlling the "character of the market" -- if high volume, presumably the
deep pockets around the world are in control. So... a low volume squat has
less meaning in terms of true resistance in the market than a high volume
squat, even if the % change in volume and MFI were identical.
- Fades and Fakes are not usefull in my experience -- they just
complete the symetry (volume can go up or down (2X), MFI can go up or down
(2X) = FOUR situations). Fades and Fakes are the result of decreases in
volume. Less volume to me means less reliable information.
- Squats can be very helpful when they are at the current extreme
of the market -- they often are very precise in timing for entry. However,
you don't always get a squat when it is time to exit, so I don't find them
as helpful on exit.
- Go's typically can't be day-traded. By the time the Go appears,
the market has run a tremendous distance (by definition) and unless you have
a break out stop, you usually have difficulty getting in in time. IF you
are already in, you have some reason to hang around for at least a little while.
- A test of a high volume squat is well worth looking for. They
usually hold, but if not it frequently is a good breakout signal. I call
this a "mu" setup from the Greek letter.
- Humans are better at seeing "squatiness" in the market than are
computers. My favorite example (true life) was a downdraft in the Swiss on
a five minute bar chart. One bar had a range of about 25 tics and a volume
of 45 tics. Then next bar had a range of about 8 tics and a volume of 43.
The market shot out of this formation to make new highs. The MFI had
dropped significantly, but the volume also dropped 2 tics. HOWEVER, for the
entire preceeding hour of trading the highest volume was in the low 20s.
iow the volume had increased significantly, from TWO bars back. The human
eye would see this, but it's a lot of programming to have a mechanical
indicator catch it.
- Related to the preceeding, people who try trading with squats
often go looking for them (a mistake!). If the 5 min bar doesn't have a
squat, maybe the 3 min does. I've always wanted to do a calculus approach
to "squattiness" in the market, but haven't had time. Conceptually, you
could do the indicator based on a tic chart with variation in the number of
tics (or amount of time) you look back and what you consider to be
significant changes. You no longer would get out foxed by which minute the
charting software uses to start the next X-min bar, etc.
I have never adequately back tested this, but know some people on this list
that may have. Back in the early days of TS, someone did try testing based
on 5 minute bars in the Swiss and had disappointing results which were
contrary to what I "saw" going on. After a lot of long distance debugging
we saw that trades were being entered on the close (typically a high volume,
low MFI bar -- frequently a squat) and stopped out on the open in a
pre-Globex market! The loses were far more than intended when designing a
day-trading system. Unfortunately, the testing stopped. I'd be glad to
work with anyone on the list who is proficient at back testing to suggest
some conditions and strategies.
Finally, I use some of Bill Williams material in my manual (with his
permission) and will not comment further on who invented this stuff.
Larry Ehrhart, WINdoTRADEr
www.windotrader.com
(773) 871-4687 voice
(773) 871-4612 fax
|