The CBOT product announcement of Market Profile in 1985 revolutionized market analysis. The
focus on intra-day market behavior offered more information than ever before possible. Couched
in the terms of pattern recognition, the CBOT Market Profile Manual suggested that analysis proceeded
from first recognizing the type of trading day. Day type was the base, the indicator of who was in
control of the
market; short or long term traders. Unfortunately, the number of day types varies with the
author and the same author might cite a different number of day types in different publications.
At CISCO we could not make pattern recognition work for us. In exploring alternatives we found
that it was very dicey to put too much faith in a single day's profile. In our book, Value Based
Power Trading, 1993, we noted that even a market in a solid trend had day to day variations that
on a single day basis made the trend unrecognizable. Broadening that concept led to the Three Day
Rule (2007). That rule, simply stated, says that market condition cannot be reliably found from a single day's data.
Three days of profiles together offers a much more reliable measure of market condition.
That missing link in profile analysis led to our work on value based analyses, starting with the
Overlay Demand Curve© (1988). (Dates such as (2007) refer to entries on the References page, which are
ordered by date)
Background on Market Profile
Market Profile (tm) appeared in 1985 as a product of the Chicago Board of Trade
that "integrates price, time and volume". It incorporates real time ticks to
generate a 'profile graphic' of market activity displayed in a 1/2 hour bar basis
(TPO's). This 'profile' often forms a bell shaped distribution when plotted as a
price-activity graphic.
The accompanying CBOT Market Profile Manual, 1985 (MPM 85), explains the uses of the profile
in terms of analyzing the bell-shaped graphic. End of day cleared data, the Liquidity Data
Bank (LDB), breaks down a day by cleared volume by type of member to show who
on the floor is active at what price levels. LDB voume data is the source of
profile value (profile value area).
The bell-shaped price-activity chart's similarity to a gaussian distribution
is used to draw inferences about the market such as the type of day (2 listed),
(day) condition of the market (trend or balance) and other profile dynamics. Value
as found from the cleared volume in the LDB, is the range of the first
standard deviation of the volume profile.
Several 'reference points' are
identified, more particularly in Market Profile Manual, 1991 (MPM 91). These reference points,
again day timeframe oriented, explain certain elements of a market's behavior
such as trade facilitation, initial balance, range extension, tails, who is most active
in the value area and
attempted direction. The book, Mind Over Markets, 1990, Dalton, et. al., (MOM),
addresses reference points in a somewhat more organized manner.
Reference points read from the bell shaped price-activity curve, including
day type, are prone to more than one interpretation. The goal
is to identify the source of market control: short timeframe traders like the
locals must trade to make a living and the long timeframe (other) traders
are looking for bargains and will not trade otherwise. The interpretations
are often judgement calls, depending upon ones recognition of the market patterns.
Market Profile Manual, 1985, states that it will take six months to a year
to gain the expertise required to recognize and interpret the various market forms.
In sum, the user is expected to become adept at pattern recognition.
Comment:
The basic premise of market profile, it's foundation, is that the bell shaped
price-volume curve from the LDB volume and the price-activity curve from TPO's
forms gaussian distributions. A gaussian distribution has a well defined form,
the bell shape. Two parameters (mean and standard deviation) define all
points on the curve. Experimentally, for a gaussian distribution, both the mean
and standard deviation can be found from the data. The identical formula can be applied to
non-gaussian data, but, of course, the interpretations are meaningless.
One problem with market profile is that there is little, if any discussion
about data validity; does the distribution at hand fit the bell shape model and
what to do if the bell shape is not present that day or the day shows both bell
parts and non-bell shaped areas. Tests of the quality of fit to a gaussian curve
are available in the statistics literature (e.g. chi-square) but not mentioned in
CBOT 1985, CBOT 1991 or MOM. A distribution may have a bell shape and still not
be gaussian. That would mean that at the very least, value obtained is questionable.
If the price-volume curve is not bell
shaped at all, it is obviously incorrect to apply gaussian statistical analyses. Since
bell shaped curves occur in balanced markets (and probably only in balances), it is
important to have tools to identifiy balance. None are offered in the Market Profile
Manuals or MOM.
Secondly, the profile methodology presented is essentially day timeframe. 'Market
condition' is used to describe the current day, not the multi-day market environment.
Simple observation of a string of profile days shows the wide variation of
shapes day-to-day. If a trader is to judge what a change in the distribution's form
means there must be guidelines. The difficulty, of course, is the almost infinite
variations in profile shape possible day-to-day or even intra-day.
The concept of finding value for the day from a gaussian distribution was
a great leap in market analysis. But not every day can be so analyzed.
It is unfortunate that no discussion is made of these situations. For example, what should the
trader do if yesterday was gaussian, but today is not; but the market is still not
trending?
Lastly, the patterns that trading decisions are based on are often poorly defined. Identifying
'Day Type' illustrates the recognition problem. Day Type is a starting point in the Market
Profile scheme for market analysis. A day's analysis begins by recognizing the profile form (day type)
which tells who is controlling the action (short time frame or long (other) time frame).
In the words of CBOT Market Profile, 1991, pg 4
...the market has only a finite number of behavior patterns and that finite
number is universal from market to market.
and from Mind Over Markets, pg 19:
...Market Profile as a whole tends to fall into readable patterns in the day
timeframe, determined by the degree of involvement of the other timeframe
participant. These patterns, when properly identified, can increase the
day trader's success, as well as provide information regarding what the market
is trying to do in the longer term.
CBOT Market
Profile, 1985, listed two day types. CBOT Market Profile, 1991 has four.
Steidlmayer on Markets, 1989, posts six. MOM, 1990, has nine. Steidlmayer on
Markets, 2003, gives five. Markets in Profile, 2007, does not have an index
item for Day Type!
The primary writers in the area do not agree with each other or even themselves
on the number of basic day types that exist. But day type presumably shows the areas of
dominance of the trader types (short and long timeframe) and so is the starting
point for profile analysis. How can the student be expected to
learn holistic profile market analysis if the guidelines are so variable and so
poorly defined?
A real danger is that in spite of the revolution in market analysis effected
by market profile methodology, it leads traders to expect too much from each
piece of data. All traders want to take a reference point, say value area or
initial balance, and predict the coming market activity; or to be able to compare
one day's value with the next to explain and project the market's behavior.
If we could interpret todays market's, as demonstrated by Steidlmayer in CBOT
Market Profile 1991, we would hold the keys to untold wealth. Steidlmayer was in
a unique position of personally knowing the players on the floor as well as being
an astute observer of market behavior. This is a far cry from the situation of
the normal trader. Reality says "you are in a complex
environment and it is highly unlikely that knowledge of any one or two market elements
will insure your success". It is experimentally apparent that one day's data
is inadequate to measure Market Condition. Further, since market shapes run
the gamut from well developed bells to completely directional straight lines, there should be
some measure of where the crossover occurs between distributions amenable to value
analysis and those that are not; if one is to use pattern recognition.
Background for Value Based Analyses
At CISCO we initailly tried the holism path and found it impossible to read
a market objectively. Starting with a profile, we first had trouble determining
the day type consistently. In some cases day type was clear, others, not at all.
Without a day type upon which to base our analysis, we were stymied.
Early on we wrote a computer program that would input a day's price - TPO data and
output day type. This was a surprisingly difficult program because there are
no sharp demarcations between day types. Our expectation was that once we
knew the day type unequivocally we would find statistical agreement between
day type and market behavior the next day. We found no aid in having day
type, so far as trading the next day was concerned. (1986, CISCO internal report)
So, by 1987 our path diverged from holism to objectivity. First we found that
value area, by the CBOT definition, could not be calculated for any but
CBOT markets. we researched the Tick-TPO, or Meta-Profile value method and that
is now the standard for the industry.
Next we realized that one day's profile could be understood only within
the context of the market as a whole, i.e. the (multi-day) Market Condition. That is,
Market Condition to us means the collective behavior of several days (minimum of three).
That led us to develop the
Overlay Demand Curve to determine whether the market is in balance or not.
The critical point here is that the reference points
used in the holistic interpretation, such as value area, are NOT valid
unless the market is stable, i.e. in balance. (1988)
In 1992 we examined the Commercial trader and their identification of value.
This study relied on the LDB data, which displayed the CTI2 (commercial) volume.
That data will not be reported after January 1, 2008. (1992)
Then we questioned sampling periods. Is a 30 minute time frame too long,
too short or just right for the intra-day bars that specify the TPOs?
Sample periods are affected by the run-pause
nature of markets. A very short time sample, say a few minutes, could be
in the noise. A very long time, say a trading day, wipes out the intra-day detail. We
found the best time to be about 25 minutes, close to the 30 minute timeframe
originally chosen by Steidlmayer in CBOT Market Profile, 1985. So the
standard 30 minute time slices are fine. On our Meta-Profile intra-day
display (CMaPS) we offer both 30 and 15 minute profiles--the shorter one for closer
examination of run-to-pause or pause-to-run transitions. (1993)
We examined the Liquidity Data Bank intra-day reports for day trading.
Our finding: The clearing delay of about an hour is too slow for most day traders. (1993)
We also examined known trends for day-to-day serial correlation. That is,
in a long trend is the trend apparent day to day? We found no serial
correlation from one day to the next, some in day to second day and
and good correlation basis day to third day. These findings confirm our
later measures of balance periods requiring three days to define a
balance. This study is on pages 19 - 23 of our book, Value Based Power
Trading, (1993)
Back in 1973 when we were beginning to manage money, we ran into the
Efficient Market Hypothesis. This theory, championed by the Chicago School
of Economics, inferred that all technical trading models were based on
invalid principles. If
true, our trading model was doomed. We reasoned that if we could find
any predictablilty in market data, that would invalidate the Hypothesis.
We showed that in a mature futures contract (front month at least 60 days
old) a new contract high would be followed by a yet higher high within
the next ten days with an 82 percent probablilty. Mission accomplished,
Efficient Market Hypothesis disproved. Work by others many years later
confirmed our finding. (1973)
A recurring theme in CISCO research is the validity of the data.
In 1999 - 2000 we developed the Advice Engine, a daily listing of all
markets in balance, with periods of 5, 10, 15 and 20 days. To merit
inclusion, there has to be both a 5 and a 10 day balance.
This list of balanced markets (end of day) is for trade guidance the next day.
The report is in two parts: Part 1 lists all markets in balance at end of
day. A sub Part 1, the Select Table, are those futures in Part 1 that a
CISCO proprietary algorithm chooses because they are deemed to have
have better potential. Part 2 lists the breakouts
of the postings of the previous day. CISCO maintains the daily Advice Engine Report
in a research database. (2002)
A diversification
tool, the Advice Engine Table also offers a quantitative methodology for testing our
ideas on the importance of balance dynamics. Results from backtesting
of models are highly suspect. We instead devised a test method that does not rely
on a model. First we choose a test future (generally but not necessarily) from the Select
Table. Then we search the history database over the timeframe chosen for breakout trades.
The resulting table of trades shows the date of the breakout, the maximum extent of
the price run and the risk that is listed in the original Advice Engine Table. We call
the maximum price run after breakout the 'Potential'. Potential is not profit, it is
the amount a perfect trader could made on the trade. The results for a SP test for
2005 - 2006 are: Out of the approximately 500 trading days, there were 268 days of balances.
Of these 268 balances, 131 broke out, with an an average
Risk of $643 and average Potential of $1,229. Understand, this is not a trading
model and to get 100% of the potential you would always have to pick the top (or bottom).
A good trader might get 40% or 50% of the Potential.
This 'Potential' approach is for "proof of principle". It shows that, for SP, trading
breakouts offers a positive return, large enough to interest a trader. A 'Figure of Merit';
Potential divided by Risk, in this case is 1.91. A subscriber can search the database for
markets of their choice, just as we did with SP above. (2001)
In response to a query, in 2002 we compared the value from the Market Profile (cleared
volume) and the Meta-Profile in changing markets, when there was a breakout. We
found that there were some differences--sometimes Market Profile would reflect
the change better, sometimes it was Meta-Profile. The article is in our References section. (2002)
Also in 2002, we showed how to develop a trading model with Auction Market Theory.
Everyone wants a model, preferably one that wins regularly with little drawdown.
As we all know, such models are in short supply. But, there is a proper approach to
model development that many of those selling training do not follow.
This article takes the trader through model development in detail. We make the point that
models can be developed that are valid within Auction Market Theory, but the fine
points of trading must come from you, the trader. You probably will not find that ideal model,
but you probably can develop a model to fit your needs. (2002)
Volatility is ever present. At best it is noise that affects one's stops. At worst
it destroys a trade. This article, written in 2003, explores volatility as it affects
a trade. A market may fit the market profile standard (bell shape, gaussian distribution)
with wide ranging prices, or narrow. Patterns may be similar for the two cases, but
high volatility may cause stops to be hit that would not be the case in low volatility. (2003)
In our effort to encourage our customers to diversify, we revisited the Trading opportunity
concept. Again using the Advice Engine database, we show how to use the Figure of Merit
in selecting trading opportunity. (2004)
Accepted wisdom in the auction market trade is the belief that the public are dummies.
Sometimes we are. Most especially when we are taking advice from the insiders. A way
to not be the dummy is to understand markets. Value analyses can prepare you to
look at markets objectively. This does not guarantee that you will win, but it does
insure that you are harder to fleece. Article on site. (2004)
The article on the Market's Second Chance revolves around the fact that markets
run and pause, run and pause continuously and on numerous timeframes. This antedates
the article on run-pause in 2007 (above). The point is that the run-pause nature
of markets tells you to not give up if you missed a trade, it may come around again.
Roughly, a market offers about three run-pause phases per day. (2004)
Market Profile/Meta Profile History 1985 - 2004 is a review. (2004)
Auction Market Value Theory is, we believe, the first comprehensive theory of
auction markets. Economists for long have held to the Sharpe-Markowitz gaussian
distribution theory for equities (Capital Market Theory, or CAPM). Their idea
that the market was an equilibrium structure has proved wrong. Steidlmayer and
Koy, Markets and Market Logic, brought forth many elegant insights about markets
and how they function. But there is no objective theory in their work. Possibly
the most comprehensive study of problems with CAPM is Why Stock Markets Crash,
by Sornette. That study gave us the courage to develop an axiom based theory. (2005)
One of the first findings to come from Auction Market Value Theory was The
Market Unit, a detailed study of the market cycle, balance, trend, balance, trend,
etc. There it was shown that a balance - trend - balance cycle could four days,
seventeen days or any other period. There is no regularity to the periods.
a first conclusion is that the standard moving average smoothing is just
wrong because there is no internal cycle to smooth. Newer version above:
Financial Markets Auction Market Value Theory. (2005)
Market Waves are not Cycles is an application of the Market Unit finding,
addresses technical analysis--to the detriment of technical analysis. (2005)
Meta-Profile for Day Trading is another follow-on of Auction Theory. (2005)
Day Trading with Market Value is a continuation study, comparing Meta and
Market Profiles. (2005)
Markets Do Not Turn on a Dime examines the time the market takes to change
it's mind. Usually this is several TPO periods. (2005)
An example of combining Visual Graphic for the set up and Meta-Profile
for the intra-day information, this is an example of a strategy analysis
using the Visual Graphic (5/16) and then the current day profile. This is the
first of a two day set. (2005)
Follow-up for Visual Graphic of 5/17 (trading on 5/18). (2005)
Day Trading the London Subway Blast is a study in herd instinct trading.
There was a great amount of potential for the value trader. (2005)
Pattern Day Trading. Use of the bell shaped distribution. Much of this
is superceded by the later Dialogue on Two Systems (2007). (2005)
Pivot Points and Candlesticks critiques. On the basis of profile/value
theory neither of these trading methodologies have much utility. (2005)
Three Major Auction Market Discoveries explore the basics of Financial
Markets Auction Market Value Theory. (2005)
Financial Markets Auction Market Value Theory. (2005)
Five Billion Dollar Loss on Natural Gas Trading is a study in point
of 'winging it' instead of following value principles. (2006)
Drawdown Studies with Advice Engine. An update. Drawdown measure is based
on the Advice Engine record. Balances with their upper and lower limits and the risk,
are listed one day and checked the next day to see if breakout occurred. If they
broke out, the Potential of the run (how far it
went after breakout) is compared to the original risk. this is a measure of
validity of the Advice Engine method. While not a trading model, the average
size of the Potential, compared to the risk, is an objective measure of the
validity of the selections in the table. This is a 'Proof of Principle'
measure: i.e. do I want to spend time developing a model from Advice Engine
data? How often does a trader have an opportunity to set out to develop a model
from a methodology that has undergone Proof of Principle testing? (2006)
Advice Engine Search Background Info
Short Term Congestion. Run-pause congestion analysis is introduced
for the very short term, 15 to 45 minutes. Each trading day offers an average
of three run-pause cycles. Trading decision making is helped by knowing what the
market is doing intra-day. (2007)
The Run Pause Profile (tm)
Value Analytics offers a major change to profile analysis. Heretofore
profile analysis was principally day oriented. Now profile reference
points are treated as flow variables; how they change over the last
three days is more important than their latest day value. How can this be?
It is simply that short-timeframe, intra-day situations can quickly develop
and as quickly fade away; not affecting the overall market condition at all.
A blip; here today, gone tomorrow. (2007)
Value Analytics: A New Departure in Profile Analysis
Value Analytics Background
The Three Day Balance Rule requires a three consecutive day set of data to define
balance. (2007)
Profile Value Area Three Day Rule
Dialogue on the Two Systems compares holism in the basic profile
analyses with the objectivism of critical value analysis as in the Three Day Rule. (2007)
The common thread running through all our studies is validity of the data. If trading is
to be a science, the data used in the analyses must be valid. The original Market Profile
analyses, and still today, treats any calculated value area as a valid measure of value;
with further analyses proceeding from that assumption. But our work on Market Condition has
shown that a one day snapshot profile may not be representative of a market and often is not.
Consequently, relying on simple reference points (point of control, value area) can give a
totally misleading read of a market. If your data is not valid, how can you possibly succeed
by following it?
