CISCO Futures
Part 1. Auction Market Theory
Contents
Contents
Rewards
Trading Models
Standards
1-303-306-1521 1-800 800 7227 Fax 1-303-306-1572
Internet http//www.cisco-futures.com
Email dljones@cisco-futures.com
Auction Market Theory
Donald L. Jones
CISCO Futures 2002 ©
Part 2. Reward, Risk and Trading Models
Part 3. Standards: Potential and Opportunity
References
Appendix 1. Comparison of CAPM and AMT
Appendix 2. Technical Analysis Example
Appendix 3. Review of the Normal Distribution
Appendix 4. Fundamental and Technical Analysis
Appendix 5. Day Trading with Market Profile: Rules
Acknowledgements
The field of Auction Markets was developed quite recently in the history
of financial economics. J. Peter Steidlmayer laid the groundwork, changing
the focus from price to value. We have
borrowed freely from his works. Jim Dalton, one of the first to recognize
the value of Steidlmayer's ideas, contributed heavily with Mind Over Markets
(co-authored by Eric Jones and Rob Dalton). Eric Jones edited The Profile
Report, an important source of research on Market Profile in the early days.
Professor Tom Drinka, at Western Illinois University, was also responsible
for research and for bringing Auction Market techniques into the classroom.
My thanks go to Tom for reviewing this work as well as an earlier book,
Value Based Power Trading.
Closer to home, Christopher Young, in my organization, developed a computer
analysis of "day type", for types of Market Profiles. Day types are
'normal', 'trend', 'non-trend, etc. The idea was to catalog a time series
of day types
and chart the normal growth patterns of markets. When the data refused to
fit a pattern I realized that markets are not day-to-day serially
correlated. Inter-day volatility masks the trend even in relatively strong
trends. Lack of serial correlation implies that one could not blindly use
yesterday's data to trade today. A longer timeframe is needed to find the
'market condition', i.e. whether it is balancing, trending or in transition.
That was the breakthrough that led to Auction Market Theory.
Donald L. Jones
Aurora, Colorado March 12, 2002
Auction Market Theory
Auction Market Theory develops and clarifies the structure
of short timeframe non-equilibrium auction markets. The
primary variable is value, the region of price accepted by
the market. Based on market structure, the theory generates
trading strategy. A side result of the theory is the
capability of calculating reward to risk ratios for individual
markets. In turn, opportunity is categorized and standards
in the nature of indexes are set for the trading function.
Part 1. Auction Market Theory
Auction Markets
Very Long TimeFrame/Very Short TimeFrame Markets
Auction Market Theory; 3 Parts
Part 1. Development of AMT: The Theory
Part 2. Reward, Risk and Trading Models
Part 3. Standards: Potential and Opportunity
CAPM is Not a Guide for AMT
Auction Market Structure: Beginnings of AMT
Steidlmayer Original Decisions
Market Observables: Exchanges
A List of Market Observables
Information Derived from the Observables
Structure of a Trading Day, the Market Profile
Market Profile Recap
Longer Term Considerations, the Overlay Demand Curve
Market Profile, Multiple Days: Overlay Demand Curve
The Overlay Demand Curve and Market Condition
Auction Market Theory Development: The Elements
Elements of Auction Markets
General
Market Profile (MP)
Overlay Demand Curve (ODC)
The Structure is Complete
Volatility
An Application: Day Trading T-bonds on March 23, 2001
EOD Analysis, Step 1a, the Overlay
EOD Analysis, Step 1b, the Market Profile
Scenario for T-bonds on March 22, after the Close
Additional Evidence
Strategy for March 23
Most of the world's markets are auction based, e.g. futures, options, debt
and equities, derivatives, etc. The Auction market dynamic is
created by the participants. Daily price range is determined through
negotiation between the traders. As price is negotiated, some prices are
accepted by the marketplace and create heavy volume. Rejected prices, the
highs and lows, and sometimes opens and closes, are seldom traded and
generate little volume. By trading many times throughout a day, accepted
prices identify value.
Despite the ubiquity of auction markets, there is no general auction market
theory. Capital Market Theory (CAPM) (ref. 1 and Appendix 1) is a special
case for long term, equilibrium markets. CAPM is at base an
investment theory. It assumes a logical investor with a particular
level of risk tolerance. A portfolio approach provides diversification.
What is not provided is guidance for the trading upon entering, changing
and exiting the portfolio. The trading part of the investment
process is ignored.
Observation and experience shows that the trading part of portfolio management
can be critical to overall performance. For convenience, assume a one-hundred
percent rollover of a portfolio and a trading gain of two percent of the
average value on the net rollover. With an average portfolio gain of eight
percent, trading gain represents a twenty-five percent enhancement of return.
A two percent trading range is not unusual in the S&P 500 index, while the nasdaq
index range is much larger. CAPM makes the point that index
funds are superior because of the low rollover and hence low costs. But even
index funds have some rollover that might be traded rather than just rolled.
Every trade has a risk, expenses and offers an opportunity. There are good
and bad times to buy and sell. The trading function can conceivably
be a profit center instead of merely a cost.
Trading risk is not treated in the risk tolerance measure that initially
sets the portfolio. Identifying and managing the trading timeframe is the
substance of Auction Market Theory (AMT). Futures approach the ideal
in auction markets. Auction market analysis developed on a futures exchange.
Most explanations in the text will use futures because of convenience.
J. P. Steidlmayer (ref. 2), a member of the Chicago Board of Trade, first
described and interpreted the
day structure of auction markets. An extension of the day market concept was
added by Jones (ref. 4) to cover market condition, the environment in
which day trading takes place.
Sharpe, in the foreword to the reprinted reference 1, makes the point that
CAPM uses little empirical material (mercifully, he says, because of the
short half-life of empirical studies). Most of science starts empirically;
collecting and sorting data to derive basic principles.
AMT is based on empirical studies, in part because the underlying
mathematical description would be a non-linear, non-homogenous differential
equation with non-constant coefficients, and basically insoluble. Many
fields of science suffer the same defect, but airplanes fly even though
their wings are not infinitely long. Likewise, auction markets can be
understood phenomenologically through empirical evidence. An assumption
is that the probability of making better trading decisions improves with
market understanding. This is a general principle running through most
human endeavors. Successful decisions in most arenas come from data
coupled with a good understanding of the business at hand. Thus, auction
market theory is normative in effect, providing information for decision making.
Very Long Timeframe/Very Short Timeframe Auctions
An equity market can be idealized as a long timeframe, equilibrium market, as
in Capital Market Theory. Such an equilibrium market is an investment
medium. At the same time, equities trade every day in the short
timeframe non-equilibrium limit. Other markets such as commodity futures
spend virtually their entire lives in a non-equilibrium situation.
Non-equilibrium markets can thus be characterized as trading media.
Auction markets as a whole are a continuum, stretching from
the shortest, non-equilibrim limit to the long timeframe equilibrium
limit. However, the
non-equilibrium short timeframe limit theory developed here will be termed
Auction Market Theory (AMT) as contrasted with the long limit CAPM.
Auction Market Theory (AMT); 3 Parts, a Brief Review
Part 1. Development of AMT: The Theory
The primary aim of Auction Market Theory is to identify the characteristics
of short timeframe non-eqilibrium auction markets and to present a coherent
description of their behavior. Achieving this goal will place market analysis
on a quasi-quantitative basis, removing the mystique and misinformation
surrounding market activity. An allied aim is to develop a procedure to
utilize current market knowledge as a guide to future (trading) action.
The approach is empirical. Observables are identified, cataloged and
interpreted to provide a structurally sound market description that applies to
real world situations. AMT analyses will identify the salient elements
of a market; including value, market condition, risk, volatility and other
items in a market's structure. These market elements will, in turn, be used
to generate a strategy for trading.
AMT (theory) develops the principles governing short
timeframe market phenomena. Often, prediction of future events qualifies
a theory. Financial markets are replete with market predictors and market
predictions. It is an axiom of Auction Market Theory that future prices
are not predictable. Current market condition is found
from AMT principles. Market condition is the state of the market, e.g.
local equilibrium or balance, local trend or transitions between the two.
The state or condition is time dependent in that one timeframe may be in
one state (say a market is balanced in a twelve day view, while trending
by a four day measure).
AMT analysis shows that changes to the current condition can be the
basis for market decisions. A strategy developed subject to the
event of change from one market condition to another is
non-predictive (normative). While predictions of future price are not made,
logical conclusions from market behavior are. Incidentally,
CAPM operates much the same way, picking portfolio candidates on the basis
of past behavior.
Part 2. Calculation of Reward and Risk
The aim of trading is the reward, the gain from a trade.
Reward (plus or minus) results from the trading process. Auction Market
Theory does not
specify trading procedures; rather trading strategies. It will be
shown that maximum potential reward in a day trade depends only on the entry
point. In certain situations (e.g. balanced markets) entry points are
readily identifiable. Reward maxima for a trading event can then be found.
Reward alone, while significant, is better understood in terms of it's
relation to the risk involved. The rarely known reward-to-risk ratio is
important to appropriate money and risk management. A risk estimate can be
found from a
simplified standard trading model that uses entry prices and risk from an
AMT analysis process. The maximum time frame of a trade is from trade entry
to market close. The maximum possible reward is the largest price excursion
from entry to close. This is defined as the Potential
of the trade. Potential for a trade in such a case, is defined by one
variable, the entry point. Potential
in day trading plays somewhat the same role as the index in CAPM, i.e. as a
marker for success. In AMT, Potential measures opportunity. These points
will be developed in Part 2 below.
Part 3. Standards: Potential and Opportunity
Standards are desirable but rare as a basis for evaluating models and
markets. If an investor's stock portfolio value declines while the
market index is declining more,
the investment is described in positive terms (it beat the market). In
non-equilibrium trading there are no indexes. A loss is absolute, a gain is
absolute. But, losses can carry different connotations. A loss in a market
offering little opportunity is quite different
from a loss in one with much opportunity. Potential is the arbiter, it sets the
standard. If one trades a market that consistently offers low Potential,
one should change markets. If one does poorly in markets that consistently
offer substantial opportunity, one must examine one's trading methodology.
Part 3, below, will use auction market principles 'best trading' risk and
reward to risk ratios to catalog markets.
CAPM is Not a Guide for AMT
Although the focus of AMT is on short timeframe trading markets, the
long-term (CAPM) is important psychologically. The public tends to think
of short timeframe
trading in terms of the more familiar longer term investing.
The two are quite different. Speculation, betting on risks and opportunities,
in the day timeframe has a totally different risk and reward structure from
investing. The long term upward trend of equities rewards a buy and hold
strategy. The same strategy in speculation is fatal. A trader/investor
who is not aware of the difference is doomed. A comparison of CAPM and
AMT is in Appendix 1.
Auction Market Structure: Beginnings of AMT
Short timeframe auction markets, although not an an equilibrium system,
do spend time in quasi-equilibrium, balanced trading.
As a non-equilibrium system, AMT is
difficult to describe in a closed mathematical formulation. Much
of the
analysis of near term markets is necessarily empirical. A benefit is that
AMT, in practice, is not obscured by mathematical formalism. The
average trader can readily understand and follow the principles. That is
not true of CAPM with it's efficient frontier.
AMT is applicable to trading strategy in a wide variety of instruments;
securities, options, interest
rates, futures contracts and other double ended markets .
Regardless of the market, the timeframe for a given trade is specific
to that trade. It ranges from seconds to minutes to many days, determined
by the market itself, in it's relationship to value and value change. A
certain percentage of
losses are inevitable in a trading environment. Leverage may exist,
as in futures or
margin trading in equities or more generally in hedge funds.
Consequently, unlike a long term investor, a trader should risk much less
than 100% of an account's capital on each trade. Performance is measured by
appreciation alone, but
it can be gauged against an ideal, the Potential.
Beginnings of Auction Market Analysis: Steidlmayer Original Decisions
J.P. Steidlmayer started as a floor trader at the Chicago Board of Trade in
1963. Over time he developed the concepts from which evolved the Market
Profile volume charts and an auction market description as well (ref. 2). The
seven market concepts below are a brief overview of his work.
1) All the publications he searched attempted to predict the
market. They failed.
He surmised that prediction of the future should not be a goal.
2) Graham and Dodd's book Security Analysis showed that value could be
found in the present tense. He projected this result to his trading.
3) He posited a basic unit of market activity, now known as the
TPO (price and time location of trading activity). TPO's are measured
in half-hour segments of the trading day. Steidlmayer selected the half-hour
time frame because that seemed to be the time it took for new information to
get represented into price change.
4) He found that grouping a day's set of basic units forms a bell
shaped distribution of price - volume over a day, which he named a
Market Profile (tm). The bell shape
called to mind the 'normal distribution' so widely applied elsewhere to
bring "order out of chaos".
5) Price-over-time measures value.
He saw that relative value at a price is equal to the sum of the basic
units (TPO's) at that price.
6) Traders/investors seek a fair price. Markets auction 'too high' and
'too low' in the search for fair prices. Markets accept fair prices
with enhanced volume and reject unfair prices by way of low volume. The
price range is reached by negotiation among the traders.
7) His experience showed that there are two sorts of market,
'(balanced) day trading' where value is static and the
market moves around a static fair price; and 'longer term' where value
is moving (sometimes called 'trend days').
Market Observables: Exchanges
Organized financial auction markets have been present in the United States
since 1792.
As markets expanded, rules were implemented and the exchange structure,
with memberships (seats), took form. Exchanges formed the arena for the
growth of the well-defined auction markets that exist today.
Steidlmayer's pioneering work occurred within the exchange structure.
In developing Auction Market Theory, a somewhat different path is
taken from the Steidlmayer work. Here the market observables, self evident
facts clear to all, forms the starting point. Nevertheless, reference 2 forms
the base for much of the development that follows.
Significant Market Observables
1) Possibly the first rules in 1792 set the meeting place, time and trading
hours. These
rules are, in the case of electronic markets, now being stretched to include
non-floor, non-exchange hours. Futures exchanges set margins based on
volatility (risk).
2) Auction prices are arrived at by negotiation.
3) Some prices are accepted by the market and show considerable trading
activity during the market day.
4) Other prices, often the highs and lows, are quickly rejected,
trading little.
5) A market may be accumulating (balancing) or distributing
(moving).
That is, markets spend time in balance and sometimes trend.
6) Some participants seek quick gains, e.g. day traders, while others trade
a longer term strategies (position/swing traders).
7) Some days many traders are active in a market, other days few
trade. Some days a market will have high volume, some days not.
8) Some markets usually have wide daily trading ranges, others are
noted for consistently small ranges.
9) The heaviest trading often occurs in the opening and closing periods.
10) Markets display little day to day serial correlation.
11) Markets cycle from balance to trend and back to balance.
12) An auction timeframe can be long or short (e.g. the market has been in
balance for 15 days, or the market broke out of balance and has been trending
for 3 hours).
13) Members on the floor have different functions (scalpers, commercials,
specialists, trading for the public, etc.).
14) Electronic platform markets clear, or complete trades faster than the
floor.
Information Derived from the Observables
Market observables deal primarily with prices and trading (volume). Auction
Market Theory owes it's utility to value. In this section, the first
steps are taken to convert the one-dimensional price to the two-dimensional
value description of auction markets. An illustration of price analysis
versus value analysis is in Appendix 2.
First, the observables are fleshed out.
1) Trading hours.
Non-electronic markets open and close at set times.
Trading is tied to the market timeframe. Orders accumulate between the close and the
next day open, creating an order backlog. At the close, traders who do not
wish to hold overnight must exit their positions. In this simple way, market
hours exert a level of control over trading structure. One of the ramifications
of trading hours is a specialized behavior pattern in the first hour of trading.
This timeframe is called the initial balance because floor members seek
to define the trading range for the day from the order backlog. How well they
accomplish a stable balance affects the subsequent trading for the day.
At the market close, those who must trade tend to create an extended closing
range. At the very least, the closing range is deleterious to the public
trader since their 'market on close' orders most often get the worst side of
the range.
Electronic markets tend to show much the same structure as the floor because
arbitrage exists and they often take their cues from the floor. As electronic
trading matures, those markets will likely lose a large part of their dependence
on the exchange floor.
Exchanges, or clearing authorities for some electronic exchanges,
interact with the public principally in setting margins. Margin is
'earnest money' guaranteeing the broker that losses are covered.
A trader's interest in margins in part is the
amount of money that must be deposited in order to trade.
Far more important
is how the exchanges set margins. With a lot of experience
backing them up, the exchange margin is set to mirror the risk.
Trading models inevitably have a risk function of some sort.
However, anytime exchange margin is changed, there is a change in
the risk being taken.
Since exchange margins are not
necessarily what a broker charges, (brokers often charge more),
keeping track of exchange margins takes some effort.
2) Prices are Set by Negotiation
Each trade has a buyer and a seller. In figure IDO 1, price auctioned
as high as 6054 until no one would bid higher. Likewise, price auctioned down
to 6036, where no one would take less. There were many buyers and sellers
at the middle prices. Each completed trade was negotiated between
a buyer and a seller.
CONTRACT: DEC 01 S FRANC (CME-IMM) TRADING DATE: 10 26 01
TRADING BEGINS 0720 (CST) CLOSES 1400 CHICAGO TIME
PRICE VOLUME Volume Plot x = 10
6054 10 x
6052 20 xx
6051 28 xxx
6050 84 xxxxxxxx
6049 136 xxxxxxxxxxxxxx
6048 182 xxxxxxxxxxxxxxxxxx
6047 464 xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
6046 210 xxxxxxxxxxxxxxxxxxxxx
6045 536 xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
6044 186 xxxxxxxxxxxxxxxxxxx
6043 254 xxxxxxxxxxxxxxxxxxxxxxxxx
6042 166 xxxxxxxxxxxxxxxxx
6041 122 xxxxxxxxxxxx
6040 74 xxxxxxx
6039 60 xxxxxx
6038 24 xx
6037 6 x
6036 12 x
Figure IDO 1. Swiss franc volume by price. Oct. 26, 2001. Minimal trading
occurs at the top and bottom prices. The three top and three bottom prices have
only 3 percent of the day's volume. The middle six (6043 - 6048) have
70 percent of the trading. The disproportionate volumes at 6045 and 6047 are
at least partly artifacts of the way orders are placed (e.g. five is a popular
trading point, diminishing the six next to it). Volume data is from the
Chicago Mercantile Exchange Liquidity Data, with volume in 'sides' (two
sides equals one round turn).
With no idea of when the trading at a particular price took place
one would be hard pressed to tell from figure IDO 1 just when a price
may have traded. That is not much of a problem in a
congesting market like this one, since
there are many opportunities at all the prices in the value area (central
70 percent of trading, 6043 - 6048).
Recasting the price - volume plot into a Market Profile and a half-hour
bar chart adds a substantial level of information as in figures IDO 4
and IDO 5.
The half-hour bars
are identified by letters; y, z, A, B,..., where each letter signifies
a time span. The y's are for the period 07:00 to 07:30, z is for 07:30
to 08:00, A is for 08:00 to 08:30 and so on. The letters are called
TPO's or time-price-opportunities. Collapsing the bars to the price
axis creates the Market Profile. TPO counts are commonly used in place
of actual volume since they embody both price and time.
3) Accepted Prices
Referring to figure IDO 1 again, the price range 6039 to 6050 is heavily
traded. A trader entering the day wishing to sell or buy at 6044 would
have many opportunities to do so. A wide range of prices were approved
for trading by the participants. These constitute the accepted prices for
this day. The graphic is of an accumulating market, one which is cohesive,
compact and in which there is good agreement on the location of acceptable
price (value). An accumulating market has a single price - volume
distribution roughly in the shape of a bell curve.
4) Rejected Prices
Prices not accepted by the market generate very light volume. Such prices
rarely trade and the trader who wants to do business there has little
opportunity to do so. The trader who wished to buy at 6035 had no opportunity
at all. In figure IDO 1 the upper three prices and the lower three prices
are one-third of the day's trading range but only 3 percent of the volume.
5) Accumulation and Distribution
The accumulating market of figure IDO 1 has given way to a distributing,
or moving market the next day. Figure IDO 2 appears to be in two parts;
the range from 6106 to 6122 and the range 6122 to 6143. The close on
10/26 at 6048 is far below today's close of 6134. That 86 point difference
is $1075 per contract. Without outside evidence we would guess that
price continued upward throughout the day, but one cannot be sure without
further investigation. The Market Profile in figure IDO 6 will solve
the problem.
CONTRACT: DEC 01 S FRANC (CME-IMM) TRADING DATE: 10 29 01
TRADING BEGINS 0720 (CST) CLOSE 1400 CHICAGO TIME
PRICE VOLUME Volume Plot x = 20
6143 86 xxxx
6142 50 xxx
6141 228 xxxxxxxxxxx
6140 194 xxxxxxxxx
6139 308 xxxxxxxxxxxxxxxx
6138 842 xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
6137 548 xxxxxxxxxxxxxxxxxxxxxxxx
6136 1022 xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
6135 384 xxxxxxxxxxxxxxxxxx
6134 334 xxxxxxxxxxxxxxxxx
6133 496 xxxxxxxxxxxxxxxxxxxxxxxxx
6132 684 xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
6131 468 xxxxxxxxxxxxxxxxxxxxxxx
6130 836 xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
6129 794 xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
6128 520 xxxxxxxxxxxxxxxxxxxxxxxxx
6127 240 xxxxxxxxxxxx
6126 252 xxxxxxxxxxxxx
6125 122 xxxxxx
6124 214 xxxxxxxxxxx
6123 52 xxx
6122 28 x
6121 366 xxxxxxxxxxxxxxxxxx
6120 124 xxxxxx
6119 16 x
6118 112 xxxxxx
6117 322 xxxxxxxxxxxxxxxx
6116 126 xxxxxx
6115 402 xxxxxxxxxxxxxxxxxxxx
6114 326 xxxxxxxxxxxxxxxx
6113 1286 xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx.....xxxxxx
6112 576 xxxxxxxxxxxxxxxxxxxxxxxxxxxxx
6111 260 xxxxxxxxxxxxx
6110 72 xxxx
6109 78 xxxx
6108 56 xxx
6107 16 x
6106 4 x
Figure IDO 2. Swiss franc volume by price. October 29 is the next trading day
after October 26 of figure IDO 1. The trading range is twice as large and
the orderliness of IDO 1. has disappeared. Volume today is 12,844 compared
to the much lower 2,574 of yesterday. This market has moved over $1,000
in one day (close to close).
6) Day Traders and Swing/Position Traders
Day traders, by definition, are out of the market by the close. They have
no long term effect on the market, since they are holding positions only
a fractional part of the day and not at all overnight. Markets do move,
and sometimes rather violently. Exchange members, as a group, tend to be flat
or hedged at the close. Longer term demand, the sort that moves markets,
comes from those who hold positions past the close. These are the position
traders in the futures, options and debt markets; and the institutions
and public in equities. Day traders are the opportunists who jump on a
move and hold a short time. Position traders have the
patience to hold longer term, creating demand.
7) Day-by-Day Variation in Market Activity
Perception of opportunity drives the day and short term position trader.
The trading day in figure IDO 1 was quiet and few traders acted on what
little opportunity existed. The overnight, off-exchange market, which is
substantial in currencies, had a 50 point ($625) move. This alerted the
exchange timeframe traders. In the first
hour of the next trading day price jumped some more and there were over 4000
contracts traded, 160 percent of the previous day total. Opportunity appeared
and was eagerly taken.
8) Heavier Trading at Open and Close
It was pointed out in item 1) that markets with floor trading develop a
backlog prior to open. The necessity of day oriented trading exiting
prior to close develops a backlog there, also. Since both open and
close have associated ranges, demand is artificially changed
at these particular times. With the rise of electronic markets, artificial
demand from exchange openings and closings will disappear.
9) Trading Ranges
Short timeframe traders gain by quickly recognizing an opportunity, seizing
it and just as quickly exiting. Return is a function of the speed and range
of the day's internal movements. So is risk. A quantitative measure of
risk in the day timeframe evaluates average swing size. Some swings are more
risky to trade than others. At the other extreme, some markets are so un-
risky that their opportunity is unacceptably small to the day trader.
One of the capabilities of Auction Market Theory is to explore and quantize
this aspect of auction markets.
10) Serial Correlation and Forecasting
There is little day-to-day serial correlation
in auction markets (see e.g. Speculation, Hedging ..., Labys & Granger, 1970;
or reference 4, p19). This makes today's trading a poor predictor for tomorrow.
Consequently, it would be a questionable aim of a market theory to propose
forecasting market behavior. AMT trading is based on understanding the
current market situation and having a strategy for change; not predicting
"tomorrow's" price.
11) Members Functions
So far market knowledge is equated to an understanding of
value based data displays. A market is also comprised of people,
the public and the members and/or professional traders. Four classes of
futures members inhabit the floor. Non-members must interact with them.
It is to the public's
advantage to understand member's motivation. Class 1 are the Locals or
scalpers, the other side of virtually every transaction. They work for
themselves, provide liquidity and are most comfortable with balanced
markets. Class 2 are the commercials who's job is to trade and hedge
for their
companies. These are the businessmen of the floor. Their company
will be a large commercial firm, e.g. Morgan Stanley. Since commercials
know both the cash and futures markets, they are the best informed
traders on the floor. They too work best in balanced markets.
In addition to their "business" they may speculate when prices
are out of line (capping is discussed later). Commercials
typically do five to fifteen percent of the volume. Class 3
are members clearing for other, off-floor, members. This class accounts
for around five to ten percent of the volume. Lastly, Class 4 clears for
us, the public. We, the public, are typically twenty to thirty percent
of the day's trading volume. Chicago Board of Trade and Chicago
Mercantile Exchange release the Liquidity Data Bank reports with
volume-price-member type statistics.
CBOT VOLUME REPORT
TRADING DATE: 03 22 01
CONTRACT: JUN 01 T-BOND (CBOT) DAY
TRADING BEGINS 0720 (CST);CLOSES 1400;TPO SYMBOLS ARE Z$ABCDEFGHIJKL
FIRST PERIOD IS 10 MINS;SUBSEQUENT PERIODS ARE ALL 30 MINS
PRICE VOLUME %VOL %CTI1 %CTI2 %CTI3 %CTI4 BRACKETS(*)
10708 2036 0.6 45.6 14.7 4.5 35.2 F
10707 5694 1.8 59.0 8.7 12.2 20.1 F
10706 5934 1.9 60.5 3.8 6.8 28.9 FIK
10705 8342 2.6 57.6 2.9 5.9 33.6 FIKL
10704 13868 4.3 56.4 3.6 11.5 28.5 EFIKL
10703 14320 4.5 54.0 5.8 5.5 34.7 EFIJKL
10702 12186 3.8 61.5 12.3 6.2 20.0 EFGHIJKL
10701 20582 6.4 56.9 9.7 7.9 25.5 EFGHIJKL
10700 15382 4.8 57.2 8.5 6.7 27.6 DEFGHIJKL
10631 23526 7.4 50.5 6.5 6.7 36.3 CDEFGHJKL
10630 32526 10.2 56.7 7.5 6.0 29.8 CDEFGHJL
10629 19146 6.0 57.2 4.3 9.6 28.9 CDEGHJLM
10628 24108 7.5 56.3 6.6 7.9 29.1 BCDEGHLM
10627 14762 4.6 54.5 5.7 10.9 28.9 BCDEGHLM
10626 13938 4.4 55.1 9.2 5.5 30.3 BCDEGH
10625 12528 3.9 59.8 3.9 13.3 23.0 BCEGH
10624 8466 2.6 61.7 2.8 7.4 28.0 BCE
10623 19036 5.9 61.1 5.1 5.7 28.2 BCE
10622 5384 1.7 57.5 4.5 4.4 33.6 BE
10621 2104 0.7 57.7 6.7 5.9 29.7 BE
10620 582 0.2 78.7 0.0 0.9 20.3 BE
10619 1210 0.4 60.6 0.0 2.4 36.9 ZAB
10618 6980 2.2 53.8 1.5 3.5 41.2 Z$AB
10617 8616 2.7 59.9 7.3 8.1 24.8 Z$AB
10616 8616 2.7 55.9 2.1 7.8 34.2 Z$A
10615 5056 1.6 54.0 5.7 9.0 31.2 $A
10614 8106 2.5 61.5 3.5 9.9 25.1 $A
10613 5006 1.6 63.2 2.2 7.2 27.4 $A
10612 1900 0.6 58.6 3.9 7.6 29.8 $
10611 4 0.0 50.0 0.0 0.0 50.0 $
%CTI1 %CTI2 %CTI3 %CTI4
VOLUME FOR JUN 01 T-BOND (CBOT) DAY 319944 57.0 6.1 7.6 29.3
VOLUME FOR ALL T-BOND (CBOT) DAY 320350 57.0 6.1 7.6 29.3
70% VOLUME SUMMARY
PRICE VOLUME %VOL %CTI1 %CTI2 %CTI3 %CTI4 BRACKETS
10704 225338 70.4 56.3 6.8 7.9 29.0 BCDEFGHIJKLM
10624
*The MARKET PROFILE is a registered trademark of the Board of
Trade of the City of Chicago 1984. ALL RIGHTS RESERVED.
Figure IDO 3. Liquidity Data Bank for T-bonds, March 22, 2001.
Column headings: Price, Volume (in half contracts), %Volume for
each price, %CTI1 is volume percentage for the local members, %CTI2 is
volume percentage for the commercial members, %CTI3 is volume percentage
for the off-floor members and %CTI4 is members acting for the public.
On the far right, BRACKETS refers to the Market Profile.
Below the volume table, totals show the average percentages of volume
for each of the four member classes. 70% Volume Summary is the value
area.
T-bonds are quoted in 32nds. The price 10708 stands for 107 and 8 32nds.
The next price tick above 10631 is 10700. A move from 10600 to 10700
is $1000 for the one unit jump. A move from 10621 to 10622 is one price
tick, worth $31.25.
Liquidity Data Bank reports are a more comprehensive version of
Market Profiles. The value area is defined by trading volume
as opposed to using the TPO's as in IDO 5.
12) Electronic Trading Platforms
Electronic platforms are developed primarily to cut trading costs.
A side effect is that trades clear faster and more accurately. Member - public
interactions are minimized. In some cases simultaneous trading occurs
on the floor and on the electronic market. At this time,
many traders are placing orders electronically through their computers but
the trades are still routed to a trading desk and pass through the standard
floor procedures. While how a trade is accomplished is of little moment
theoretically; practically there can be a big difference in costs.
As electronic trading grows, the influence of floor members will
diminish. The current effects of exchange trading hours will no doubt
diminish as well. These changes will little affect the fundamental
characteristics of auction markets. Most likely, a trader familiar with
auction market basics will have minimal trouble incorporating such changes. For
instance, if opens and closes disappear entirely, a new timeframe to replace
the exchange day will become apparent, probably defined by
trading volume. Currently, the 24 hour electronic markets show consistent
times of higher volume tied into the market being traded.
13) Markets Cycle
Markets continually move from balance to testing the balance,
to trend, to testing for end of trend and back to balance.
The time spent in any one phase may be long or short. There is no valid
method of predicting when the phase may
change. Further, any phase may be arbitrarily short. The change from
balance to trend can occur in minutes or may take days of testing the
balance. Figure IDO 1 was a balance day, the next day that balance has
cycled into a trend phase. The progression is known: a market in balance
today will surely break out in the future.
Structure of a Trading Day, the Market Profile (tm)
Figure IDO 1 illustrates the price - volume structure of one quiet day
in a balanced market. Missing is information on the time of day
any particular volume occurred. By slicing a day into sub timeframes and
posting the segments in order, a picture of time behavior emerges.
Figure IDO 4 shows the trading of figure IDO 1 in half-hour bars. Contract
volume is now "time" volume.
MARKET PROFILE* REPORT FOR 10 26 01
AND SEGMENTED AUCTION
COMMODITY -- S FRANC (CME-IMM) DEC 01
Price Brackets Segmented Auction
6054 F
6052 F J
6051 C F J
6050 C F G J K L
6049 C D F G H J |K |L
6048 A C D F G |H | |J |K |L
6047 z |A |C |D E |F |G |H | |J |K >L
6046 |y |z |A | |C |D |E |F |G |H | | | |L
6045 >y |z |A | |C |D |E |F |G >H >I > > |
6044 |y >z |A | |C |D |E |F |G |H |I | | |
6043 y |z >A >B >C >D >E >F >G | |I | | |
6042 z |A |B |C |D |E |F |G | | | | |
6041 z A |B |C |D |E |F | | | |
6040 z A |B |C |D |E | |
6039 A B
6038 A B
6037 A B
6036 B
Figure IDO 4. Swiss franc price by time. Oct. 26, 2001. Time is
denoted by vertical bars: y is 0700 - 0730, z is 0730 -0800, A is
0800 - 0830 and so on until L, 1330 to 1400. These letter groups
are just 1/2 hour bars. The display is called a "Segmented
Auction" since the day is segmented into half-hour time periods.
The time source of the quasi-bell shape of figure IDO 1 is
now apparent. The lower part was built in early day trading,
the upper, later. End of day trading was near the highs.
Combining figures IDO 1 and IDO 4 displays both volume
and time. Many markets do not report volume along with the price
ticks. In fact, some markets never report volume at price, just
end of day totals for all prices. A surogate for volume (demand)
is price over time. Each of the letters in figure IDO 4
denotes trading at that price within that time frame. The auction market
assumption is that frequency of trading at a price measures
relative demand. e.g. at 6050, trading first occurred in the
C timeframe, then recurred in F, G, J, K, L timeframes a
total of six periods throughout the day. At 6054 there was trading
only in F time period. The assumption is that there is
more demand at 6050 with six events than at 6054 with one
event, i.e. 6050 won the popularity contest. Figure IDO 1 shows that
indeed there was much more volume at 6054 than at 6050.
Events (the A, B, C's) are called TPOs, short for Time-Price-Opportunity
(or That-Price-Occurred).
One who wanted to trade at 6050 had at least six chances, six different
half hour periods where trading occurred at 6050, while
someone wanting to trade at 6054 only had the F time frame to do
it in. Opportunity at 6050 is approximately one-sixth the
opportunity at 6054. A more precise measurement of relative opportunity
could be obtained by using 15 minute bars, or even shorter.
Practically, TPOs greatly simplify defining market structure
within the trading day. Collapsing the TPOs from the half
hour bar into the price - volume distribution of figure IDO 5
completes the development of the Market Profile.
MARKET PROFILE* REPORT FOR 10 26 01
AND SEGMENTED AUCTION
COMMODITY -- S FRANC (CME-IMM) DEC 01
Price Market Profile Segmented Auction
6054 F F
6052 FJ F J
6051 CFJ C F J
6050 CFGJKL C F G J K L
6049 CDFGHJKL C D F G H J |K |L
6048 ACDFGHJKL A C D F G |H | |J |K |L
6047 zACDEFGHJKL z |A |C |D E |F |G |H | |J |K >L
6046 yzACDEFGHL |y |z |A | |C |D |E |F |G |H | | | |L
6045 yzACDEFGHI >y |z |A | |C |D |E |F |G >H >I > > |
6044 yzACDEFGHI |y >z |A | |C |D |E |F |G |H |I | | |
6043 yzABCDEFGI y |z >A >B >C >D >E >F >G | |I | | |
6042 zABCDEFG z |A |B |C |D |E |F |G | | | | |
6041 zABCDEF z A |B |C |D |E |F | | | |
6040 zABCDE z A |B |C |D |E | |
6039 AB A B
6038 AB A B
6037 AB A B
6036 B B
Figure IDO 5. Swiss franc Market Profile. Oct. 26, 2001.
The price - time distribution
is quasi-bell shaped. TPO volume peaks in the middle prices
(6050 to 6040) and then tails off toward the upper and lower limits. There
is very little support for trading at the highs and lows of the day.
The highs and lows are rejected. Prices in the middle are
accepted.
A Market Profile tends to even out the volume
trading spikes such as those at 6047 and 6045 in figure IDO 1. The
same is seen to be true in figure IDO 6, the profile of the data in
figure IDO 2.
Price-over-time, in line with general usage, is designated value.
Value maps out a wide area in the mid region of a balanced market. It is
inappropriate to say for figure IDO 5 that value for the day is 6045.
Rather, one would state that the market finds value in the 6050 - 6040
area. It is appropriate to say "the market rejects prices
in the neighborhood of 6054 and 6036".
Balance days consistently display the quasi-bell shaped curve. That
shows a general agreement on value. Note that the distribution is not
ideally normal--this is experimental data (e.g. there is not enough
data to apply econometric analyses).
Trend days often show small bell shaped price - volume distributions
(ref. 4, pg 78 -84), albeit in a manner modified to accept the change
in value.
MARKET PROFILE* REPORT FOR 10 29 01
AND SEGMENTED AUCTION
COMMODITY -- S FRANC (CME-IMM) DEC 01
Price Brackets Segmented Auction
6143 K K
6142 KL |K L
6141 KL |K |L
6140 HKL |H | | |K |L
6139 EHKL |E | | |H | | |K |L
6138 BEHKL B | |E | | |H | | |K |L
6137 BDEHKL B |D |E | | |H | | |K |L
6136 BDEHKL B |D |E | | |H | | |K |L
6135 BDEHJKL B |D |E | | |H | |J |K |L
6134 BDEHIJKL B |D |E | | |H |I |J |K |L
6133 BDEHIJKL B |D |E | | |H |I |J |K |L
6132 BDEHIJK B |D |E | | |H |I |J |K |
6131 BCDEHIK B C |D |E | | |H |I | |K |
6130 BCDEFHIK B C |D |E |F | |H |I | |K |
6129 BCDEFGHK B C |D |E |F |G >H > > >K >
6128 BCDEFG B C |D |E |F |G | | | | |
6127 BCDEFG B C >D >E >F >G | | | | |
6126 BCFG B C | | |F |G | | | | |
6125 BCFG B |C | | |F |G | | | | |
6124 B B | | | | | | | | |
6123 B B | | | | | | | |
6122 B B | | | | | | | |
6121 B |B | | | | | | | |
6120 B |B | | | | | | |
6119 zB z |B | | | | | | |
6118 zA z A | | | | | |
6117 yzA y |z |A > > | | | |
6116 yzA |y |z |A | | | |
6115 yzA |y |z |A | | |
6114 yzA >y |z |A | |
6113 yzA |y >z >A | |
6112 yzA |y |z |A | |
6111 yzA y |z |A | |
6110 yzA y |z |A | |
6109 zA z A | |
6108 zA z A | |
6107 zA z A | |
6106 z z | |
Figure IDO 6. Market profile for SF on October 29, 2001. A trend day.
The volume profile, figure IDO 2, shows the same general structure, but
market profile shows timing within the movement. Overnight trading
in the inter-bank market moved price upward as noted (from about 6050 to
the 6114 region). For the first three periods the market accepted 6114
as the new balance. But this was merely a pause, not end-of-trend.
The next jump in B period (8:30 - 9:00) found a new balance around 6133.
Structures like the two distributions in IDO 6 illustrate the
inefficiency of markets. An overnight price rise of about 50 points
obviously came
on an increase in demand. The exchange market opened around 6114 and
traded there for about an hour. However, the latent demand in the market
became slowly recognized, driving price to the 6135 region, which became
the balance for the day. Can this scenario be proved? No, it is not
possible know the opinions and thought processes of all traders. However,
this "slowly catching on" behavior is quite prevalent in value shifts. Another
example of inefficiency we will discuss is the "short covering rally" in which
the demand is latent on the opening, price is driven up strongly, the shorts
are often covered in about an hour, and then price slowly recovers
over several hours as
the true demand becomes apparent to the traders. A profile of this action
represents a capital P. This behavior has been named the "P" distribution,
as in Peter (Steidlmayer). The P distributions show a time lag of about two
hours, the time it takes the market to digest the real demand.
Market Profile Recap
Within a trading day, the natural interplay of trading forces generates a
price - volume
curve similar to the well known bell shape of the normal distribution. The
volume per price around the middle prices far outweighs the volume per price
farther away. The middle
seventy percent of the volume is designated Value by analogy with the
+/- one standard deviation of the normal curve (see figure 2). Prices at
either end of the 70 percent value region define the Value Area.
Prices outside the Value Area are increasingly less significant the
further they are from the value region, with the day's highs and lows
having the least volume and being least significant (ref. 5).
Longer Term Considerations, the Overlay Demand Curve (tm)
Any auction market may be, at a particular
time, (1) in balance (temporary equilibrium), (2) in a trend or (3) in
transition between equilibrium and trend.
The trend phase breakout from a balance can often be identified in a short
time, depending on the speed with which it develops. If there is a quick
breakout, the change to trend is fast, e.g. overnight (the Swiss franc on
10/29, figures IDO 1 and IDO 2). An equilibrium or balanced market
condition becomes recognizable only only after it is well underway,
e.g. several days after the end of a trend (see ref. 4, pg 77 for run - pause
behavior in trends).
Market Profile, Multiple Days: Overlay Demand Curve
Market Profile pictures the 'day' structure of a market.
It's time frame is the length of the trading day and the information it contains
reflects the participant's feelings for that time frame. Reference
points for the profile (value area, point of control, etc.) are 'local' and
have an equally short term validity (of the order of a day). Carry over into
the next day depends on the rate of change of the participants change in
attitude. Each day is relatively independent of the former day; the coupling
from day to day is normally very weak (ref. 4, pg 19). But a trader seeks
information to act on and previous market activity inevitably has some
effect. That carry-over effect is measured by consolidating several days of
Market Profile, into what is called the 'Overlay Demand Curve'.
FIVE DAYS OF MARKET PROFILES
MARKET PROFILE* REPORT FOR 03 16 01 - 03 22 01
COMMODITY -- T-BOND (CBOT) DAY JUN 01
Day ID ==> 5 6 7 8 9
Price 03 16 01 03 19 01 03 20 01 03 21 01 03 22 01
10708 F
10707 F
10706 FIK
10705 FIKL
10704 EFIKL
10703 EFIJKL
10702 EFGHIJKL
10701 EFGHIJKL
10700 DEFGHIJKL
10631 CDEFGHJKL
10630 CDEFGHJL
10629 A CDEGHJL
10628 AB BCDEGHL
10627 AB BCDEGHL
10626 ABCD y BCDEGH
10625 ABCD y BCEGH
10624 ABCD yz BCE
10623 ABCDE yz BCE
10622 ABCDE yz BE
10621 ABCDE yz BE
10620 zABCDE yz BE
10619 zABCDE yz yAB
10618 zABCDEL zB yzAB
10617 zABCEFL zABCGHJ yzAB
10616 zBCEFL ABCGHJ yzA
10615 zBCFL ABCGHIJK z
10614 zBCFGIL y ABCFGHIJK z
10613 yzBFGHIL yz L ABCFGHIJK z
10612 yzFGHIJL yzA L ABCEFGGHIJK z
10611 yzFGHIJL yzABG L BCEFHIKL z
10610 yzFGHIJL yzABCG L BCDEFHKL
10609 yzFGHJKL yzABCDG L BCDEFKL
10608 yzFGHJKL yzABCDEFGHI KL CDEFL
10607 yzGHJKL ABCDEFGHIJ KL CDEFL
10606 yGJKL BCDEFHIJ KL CFL
10605 JK CEHIJK KL
10604 JK yBJKL
10603 KL yzABCDEJK
10602 KL yzABCDEJK
10601 KL yzABCDEJK
10600 L zACDEJK
10531 L zEFJK
10530 L zEFGIJK
10529 EFGHIJ
10527 FGHI
10526 I
Figure IDO 7. Five sequential days of Market Profiles. Mar. 16 - 22, 2001.
To illustrate: The five days of figure IDO 7 appear to be trending down for
three days and then trending up. There is, of course, noise and volatility
included in the data. Further, each day may have it's own rumors, power
plays, reports and the like. We are left with an unknown variation from day
to day, even if the (real) value does not change. Each of the days appears
to have a relatively well developed bell shaped curve of value. However the
center of value, the maximum TPOs (POC) varies from day to day:
POC (Point of Control)
3/16 10611
3/19 10608
3/20 10602
3/21 10612
3/22 10631
The scatter in POC averages out to 10611. The first four days average is 10608.
The latest day, 3/22, seems to be out of line, a little stronger than the average
deviation. We leave the question of day 5 for the moment and return to
figure IDO 7. Our eye may well have fooled us. At least in the first four days
our perception of trend was based on an average deviation of only about
3 points in the POC. The volatility (half-hour ranges) is easily 8 points.
The Overlay Demand Curve and Market Condition
Experience shows that successful trading with the Market Profile alone
is difficult to attain. Missing is the context
for analysis, the longer term structure of the market.
One trades differently in a trend than in a balance. But trend, balance or
transition phases are often not one-day phenomena. It usually takes several
days for a recognizable balance to develop. The same is true for the transition
from a trend to a balance. Time is a factor, due to noise, volatility and
changing
market situations. Extending to a longer timeframe can be accomplished by
collecting and assembling a series of day timeframe Market Profiles.
Since there is little day-to-day serial correlation between profiles, merely
stepping from one profile to the next is too confusing, as noted in the discussion
following figure IDO 7. Rather, the set of
profiles must be integrated, collected in such a way that their salient
characteristics are preserved with much of the noise and some of the volatility
excised.
Summing several days of profiles cancels out the noise (rumors, news, etc.)
that is a part of each day. The resulting Overlay Demand Curve(tm)
turns out
to contain a deeper level of information. This is similar to a tornado that,
while comprised of many individual wind cells, is totally unlike any of them.
Overlays (tm), in the same way are more than the sum of their
Market Profiles.
The Overlay converts data from the day timeframe to the longer term
required for understanding the over-all market behavior. An Overlay answers
the question what is the market doing with a time specific response (e.g.
the market has been in balance for the last four days, while trending up over
the last 17 days). The response "balance last four days, within an overall
17 day trend" quantifies the market's condition.
An Overlay provides, among other things, market condition,
distribution limits for balanced markets, trading entry prices and provisional
stop loss points. Market condition is the starting point and foundation
for all subsequent trading decisions.
Overlays can be set at any length. A convenient grouping for futures is found
to be four periods: five, 10, 15 and 20 days.
The Overlay of the five days of Market Profile of figure IDO 7 is in
figure IDO 8. We define a distribution (bell) to be limited by three TPO's as
an estimate of the distribution cut-off at +/- two standard deviations,
the approximate 95 percent confidence level of a normal distribution.
Thus, the upper limit is 10706 and the lower limit is 10528. There is
a single distribution, so the market is in balance, on a 5 day basis.
Insofar as the normal distribution analogy holds, a price outside the upper
or lower limit is likely
to be the start of a different distribution. Put another way, a breakout
from an Overlay limit is an alert for a change in value. Or, a breakout
from balance is an alert of a potential change of market condition from
non-directional to directional (trending).
TPO VOLUME OVERLAY AND PRICE ROTATION PROFILE
JUN 01 T-BOND (CBOT) DAY
03 16 01 TO 03 22 01
PRICE DYS L/F ROT PROFILE * TPOS TPO VOL OVERLAY *
10708 1 9 9 1 X
10707 1 9 9 1 X
10706 1 9 9 3 XXX <== Upper Dist. Limit
10705 1 9 9 4 XXXX
10704 1 9 9 5 XXXXX
10703 1 9 9 6 XXXXXX
10702 1 9 9 7 XXXXXXX
10701 1 9 9 8 XXXXXXXX
10700 1 9 9 9 XXXXXXXXX
10631 1 9 9 10 XXXXXXXXXX
10630 1 9 9 11 XXXXXXXXXXX
10629 2 59 59 10 XXXXXXXXXX
10628 2 59 59 8 XXXXXXXX <== Close
10627 2 59 59 7 XXXXXXX
10626 3 59 589 9 XXXXXXXXX
10625 3 59 589 10 XXXXXXXXXX
10624 3 59 589 10 XXXXXXXXXX
10623 3 59 589 9 XXXXXXXXX
10622 3 59 589 9 XXXXXXXXX
10621 3 59 589 9 XXXXXXXXX
10620 3 59 589 11 XXXXXXXXXXX
10619 3 59 589 11 XXXXXXXXXXX
10618 3 59 589 12 XXXXXXXXXXXX
10617 3 59 589 15 XXXXXXXXXXXXXXX
10616 3 59 589 15 XXXXXXXXXXXXXXX
10615 3 59 589 16 XXXXXXXXXXXXXXXX
10614 4 59 5689 19 XXXXXXXXXXXXXXXXXXX
10613 5 59 56789 22 XXXXXXXXXXXXXXXXXXXXXX
10612 5 59 56789 22 XXXXXXXXXXXXXXXXXXXXXX
10611 5 59 56789 24 XXXXXXXXXXXXXXXXXXXXXXXX
10610 4 5 5678 25 XXXXXXXXXXXXXXXXXXXXXXXXX
10609 4 5 5678 22 XXXXXXXXXXXXXXXXXXXXXX
10608 4 5 5678 24 XXXXXXXXXXXXXXXXXXXXXXXX
10607 4 5 5678 21 XXXXXXXXXXXXXXXXXXXXX
10606 4 5 5678 20 XXXXXXXXXXXXXXXXXXXX
10605 3 5 567 10 XXXXXXXXXX
10604 2 67 7 XXXXXXX
10603 2 67 11 XXXXXXXXXXX
10602 2 67 12 XXXXXXXXXXXX
10601 2 67 11 XXXXXXXXXXX
10600 2 67 10 XXXXXXXXXX
10531 2 67 7 XXXXXXX
10530 2 67 7 XXXXXXX
10529 1 7 5 XXXXX
10528 1 7 5 XXXXX <== Lower Dist. Limit
10527 1 7 1 X
Figure IDO 8. Five Day Overlay Demand Curve of June 2001 T-bonds 3/16 - 3/22.
The label L/F gives the range of the earliest day (5) and the most recent
day (9). The Rotation Profile (ROT PROFILE) is the range for each of the
five days presented in Market Profile form. It allows the relative dates
of trading to be resolved. In this case, the 9's show the latest day's
trading to be near the top of the 5 day distribution.
Earlier it was noted that one effect of disequilibrium is that Auction
Market Theory has an empirical base. That point
is demonstrated in the following discussion of the Overlay of figure IDO 8.
It will
be seen that the tenets of AMT leads one directly to the analysis. Unraveling
the
details inside a particular market situation draws on both market knowledge
and experience. There are
many possible Overlay timeframes, for all different markets. The shortest one
would use would have at least the most recent market condition in its
entirety.
This one covers five days of the U.S. T-bonds. The market may be in one of
four conditions (balance,
transition to trend, trend and transition to balance). Figure IDO 8 shows a
balance (10706 - 10528).
Further, trader activity and interest in the market may be of various kinds.
Yet, the Overlay and Market Profile structures are universal, bringing
the
market forces together in a single display. The analyst's task is to read
the market from AMT principles and set the strategy for the next day. The
basics will all be there, but the
reasons for the behavior will often come from experience and market
understanding.
Observables are the raw data of Auction Market Theory. They range from
prices and their behavior to constructs such as Market Profile and Overlay
Demand Curves. The constructs bring value into the picture. With value
in hand, theory development can proceed.
Auction Market Theory Development. The Elements.
Now that the observables and their consequences have been considered, it is
time to collect and sort out the descriptors of auction markets. General
information brings auction markets into focus; what they are, how they are
constituted. This information is the framework for understanding the details
of the theory.
Elements of Auction Markets
General
Since financial auction markets are of diverse sorts, some of these general
elements are prone to change. The viewpoint here is of exchange based
trading. Movement to off-exchange trading will certainly alter the
effects of exchanges and members. However, nothing in the forseeable
future will change the fundamental way auction markets behave, the way
prices are negotiated and the way demand affects value.
1) Future price levels are not predictable.
2) Markets display little day to day serial correlation.
3) Auction market: marketplace in which price is arrived at by negotiation.
4) Double-sided auction markets see activity by both buyers and sellers.
5) Auction markets are not generally equilibrium systems.
6) An auction market's structure is continuously evolving, being revalued.
7) An auction market is in one of two conditions: balancing or not balancing.
8) Demand fluctuates over the day timeframe.
9) Demand change drives price change.
10) Markets generate price data: ticks, open, high, low, close
11) Price ranges are developed by negotiation.
12) Some prices are accepted, some are rejected.
13) Market volume varies day to day.
14) Market range varies day to day.
15) Markets accumulate (balance) and distribute (trend).
16) Markets cycle (phases are: balance, test, trend, test, balance, etc.)
17) A market phase can exist for a short time (minutes) or long (days).
18) Market hours control aspects of trading: openings, closes, initial balance.
19) Heavier trading is seen on exchange opens and closes.
20) Auction markets are traded for short term appreciation.
21) Traders seek value.
22) Accepted prices define value.
23) Value is price over time. P x T = V x constant
24) Value based (price-time) data: Market Profile, Overlay Demand Curve.
25) Short timeframe trader holding period is from minutes to hours.
26) Longer timeframe trader holding period is days.
27) Longer timeframe traders move markets by accumulating positions.
28) Public day traders and public long timeframe traders seek trends.
29) Long term trending markets are controlled by long timeframe traders.
30) Four types of exchange members (floor, commercials, off-floor, public).
31) Most members seek balanced markets.
32) Most members are short-time frame traders.
33) Electronic platforms clear faster
Market Profile (MP)
Market profiles graphically display generally understood
market characteristics (price - volume). A market day displays price change
from demand, rumors, aborted moves and trading strategies. Market Profile
displays the structure of the day.
1) Price - volume structure: a day's trading is quasi-bell shaped.
2) Accepted and rejected prices are displayed. Usually they are highs and lows.
3) The TPO is the minimum time for information to be decoded by the market.
4) TPOs developed because of market inefficiency in mirroring demand.
5) The Value Area is easily identified by it's bell shaped bulge.
6) The most heavily traded price is near the middle of the distribution (POC).
7) Volatility is measured by the average range of the half-hourly bars.
8) Market Profile displays market congestion.
9) Balancing markets are congesting.
10) Pausing trends are congesting.
11) Market Profile displays Initial Balance (first hour).
12) Demand change is often delineated in 'double (running)' Market Profile.
The Overlay Demand Curve identifies the phase, the market condition.
The Overlay is a multi-day construct.
Market profile is a day structure. It's interpretation
requires a larger framework, the phase of the market. A balanced profile
in a balanced market phase would be interpreted differently than the same
profile within a trending market.
Overlay Demand Curve (ODC)
1) ODC collects day markets into longer term structures.
2) Demand Curve structure of ODC shows market condition, its phase.
3) ODC development shows continuation or change in phase.
4) Breakout from balance is an alert for trend.
5) ODC shows time evolution of a market with varying length Overlays.
6) Balanced markets define the distribution limits and risk.
7) ODC is Quasi-bell shaped, locates value region and shows price rotation.
8) In trending markets, ODC shows trend components including pauses.
9) The ODC distribution delineates expected and unexpected market behavior.
10) Expected/unexpected behavior leads to "if this, then that" analysis.
11) ODC may contain internal trends.
12) ODC balance may be skewed, indicating non-balance influences.
The Structure is Complete
Auction markets can be analyzed and understood with the elements posted
above. These elements are the base of Auction Market Theory. Since the theory
is empirical it is not falsifiable, i.e. it's validity rests upon it's
utility.
But markets change and new definitions
may be required or old ones altered. Certainly there will be change in
the exchange structure if electronic trading dominates. Or, the TPO, which
is defined on a 30 minute basis, may move to another timeframe as new methods
of information dissimination develop. Such change will not alter the
fundamental ways auction markets work.
The elements of the theory are both descriptive and numeric. Facts of
auction market behavior are useful for qualitative decisions, e.g. increasing
volume alerts the trader to other possible changes. Numeric elements are
the stuff of hard trading decisions. Numerics are, e.g. the Market Profile
value area prices, the rejected highs and lows; the Overlay Demand Curve
limits and risks and the price of the pause in a trend. The numerics can lead
to trading rules, e.g. buy if market price exceeds the upper limit.
Volatility
Volatility is a natural part of all auction markets. In CAPM it is the risk,
arrived at statistically. AMT is slightly different. Volatility is related to
the trading range; small in quiescent periods, larger in more active
markets. It changes from day to day. Fluctuation grows with volume in the
day timeframe. Daily trading
range does give a gross estimate of market fluctuation.
A better working estimate of volatility describes activity within the day.
Market Profiles are based on half-hour periods. Half-hour timeframes
break down the day into manageable parts. More importantly, a half-hour
appears to be the minimum average time for changes in demand to be reflected
in the value (see Steidlmayer Observations, 3).
We define the (AMT) volatility as the average range of the half-hour time
periods of a Market Profile. In figures IDO 9 and IDO 10, these are:
y z A B C D E F G H I J K L Average
March 21 8 8 6 10 12 4 6 9 6 8 5 6 7 6 7.4
March 22 4 8 7 12 9 7 17 11 10 9 7 7 8 11 9.1
The average of the half-hour bars approximates the risk of a trade stop-out
from either the long or short side. It is the 'noise' risk. If one sets
a risk (stop-loss) smaller than the noise, then the probability is high
that simple market fluctuation will cause trade exit. The volatility, then,
becomes the minimum risk one should take on a trade.
Practically, volatility has another important use. It is a sensitive measure
of market congestion. Balanced markets (congestion) have low volatility.
Trending markets have larger volatilities. March 21 is clearly congesting,
as observed in figure IDO 9. March 22 (figure IDO 10) is a combination
trend (periods y through F) and congestion (periods G through L).
The 90 day average volatility for T-bonds (as of March 13, 2002) is 8.3.
Minimum is 3.9 and maximum is 15.5. Assuming about the same range in 2001,
both March 21 and 22 are near the average. Large volatility increases rarely
precede the start of a trend, although sometimes the general market tenor,
as measured by volatilty, rises prior to directional movemant. Volatility
is more of a coincident indicator, which helps to uncover trend end. In the
table below, volatility offers a tip-off to market intentions.
UU MAR 02
DATE OPEN HIGH LOW CLOSE BAL VTY ULIM LLIM
1/28/ 2 113250 113880 112610 113550 YES 303 113900 111800
1/29/ 2 113600 113825 109750 110050 NO 546
1/30/ 2 110050 111575 108075 111550 NO 775
1/31/ 2 111550 113000 111300 113050 NO 405
2/ 1/ 2 112875 113225 111850 112350 YES 350 113600 108700
2/ 4/ 2 112325 112400 109100 109525 YES 471 113000 108700
2/ 5/ 2 109550 110150 108225 108900 NO 614
2/ 6/ 2 108925 109450 107700 108375 NO 600
2/ 7/ 2 108625 109500 107625 107700 NO 578
2/ 8/ 2 107625 109675 107550 109650 YES 483 110300 107750
2/11/ 2 109700 111275 109425 111025 NO 308
2/12/ 2 111050 111325 110250 110750 NO 337
2/13/ 2 110725 112150 110525 111875 NO 383
2/14/ 2 111900 112550 111175 111675 NO 367
2/15/ 2 111650 111800 110300 110475 YES 387 112400 109850
Table IDO-T1. S&P emini March 2002. Market demand interpretation
aided by the volatility. BAL is 5 day balance as discussed in the
Overlay Demand Curve section. ULIM and LLIM are the Overlay balance
limits. VTY is the half-hour bar average volatility for the day.
Start with the Balance as of close Jan 28.
Jan 29, breakout on down side alerts for start of trend.
Close of Jan 29: Price lower, volatility at 546 is up 80 percent.
Interpretation: volatility implies demand is still present.
Close of Jan 30: Trend bottomed out at 108075. Closed higher.
Interpretation: short timeframe trend is over. Higher volatility
is not directional and is disregarded.
Start with the Balance as of close Feb 4.
Feb 5, breakout on down side, volatility up 30 percent, price lower.
Interpretation: short timeframe trend is probably still in place.
Feb 6, price moves down slightly, volatility is only 27 percent above entry.
Interpretation: demand or trader interest is not growing.
Feb 7, price continues down, volatility is down to 22 precent above entry.
Interpretation: demand continues to decay.
Feb 8, local bottom reached at 107550, close is higher, market in balance.
Interpretation: trend is over.
Start with the Balance as of close Feb 8.
Feb 11, breakout on the upside, close at breakout price, volatility lower.
Interpretation: breakout not supported by demand increase.
Volatility is another valid way to check markets for congestion. As
reference point for market condition, volatility adds to the visual measures
discussed in figures IDO 6, 9, 10 and 12.
Volatility calculations are tied to the timeframe. If a different timeframe
is selected (say 15 minute bars) the volatility will be unique to that
timeframe. However, the only valid volatility is the one associated with
the appropriate timeframe, the timeframe that best reflects the time delays
inherent in the market. That timeframe is thirty minutes in the data in
this report.
An Application: Day Trading T-bonds on March 23
Auction Market Strategy for March 23, Market Condition from Overlays:
At the end of a trading day we are faced with the decision of how to
trade tomorrow. A swing/position trader will first attend to the
trades that are still on. A day trader will presumably have no current
trades. For this example we assume no positions left over at the
close of March 22.
Our general approach is to collect the information available on value and
market condition. These data will include the latest day's behavior
and at least the market of the day prior. Then we factor in what we know
from the theory of markets. Lastly, we set our strategy for the next day.
Both day and swing traders start their analyses at the same place--with the
market condition.
Market Condition at the close of March 22 from figure IDO 8 is:
MC1) Market in 5 day balance, with limits 10706 and 10528, close 10628
MC2) Balance is skewed toward the top
MC3) Latest day trading (L/F = 9) concentrated at upper prices
From the previous 5 day Overlay of March 21 in figure IDO 13:
MC4) Market in 5 day balance, limits 10626 and 10527, close 10611
MC5) Balance is symmetrical
MC6) Latest day trading (L/F = 9) mostly above the midpoint
Conclusions from Market Condition (MC) behavior
MC7) On March 22 the balance broke out on the upside but did not hold
MC8) Price ran up to 10708 (14 ticks = $437), then pulled back to close
at 10628, a sign of weakness
MC9) At end of day, market is back in balance (is this a failed breakout?)
Market Condition Preliminary trading decisions (TD) for March 23
TD1) Swing trader will go long above 10706 or short below 10528
TD2) Risk will be around $325, the one standard deviation level.
Recall that the market condition provides the framework within which value
based trading decisions are made.
Auction Market Value Analysis (MV) for March 23
At the end of trading on March 21 the value area is 10617 to 10609.
Market Profile for the day is unremarkably congesting (figure IDO 9).
MV1) Value Area 3/21: 10617 to 10609, 8 points ($250)
LENGTH OF FIRST PERIOD = 10 MINS
MARKET PROFILE* REPORT FOR 03 21 01
AND SEGMENTED AUCTION
COMMODITY -- T-BOND (CBOT) DAY JUN 01
Price Brackets Segmented Auction
10626 y y
10625 y |y |
10624 yz |y |z |
10623 yz >y |z |
10622 yz |y >z | |
10621 yz |y |z | |
10620 yz y |z | | |
10619 yz y |z > | | |
10618 zB z | |B | | |
10617 zABCGHJ z |A >B >C > > | |G |H | |J | |
10616 ABCGHJ A |B |C | | | |G |H | |J | |
10615 ABCGHIJK A |B |C | | | |G |H |I |J |K |
10614 ABCFGHIJK A |B |C | | |F |G |H |I |J |K |
10613 ABCFGHIJK A |B |C | | |F |G |H |I |J |K |
10612 ABCEFGHIJK A |B |C | |E >F >G >H >I >J >K >
10611 BCEFHIKL B |C | |E |F | |H |I | |K |L
10610 BCDEFHKL B |C |D |E |F | |H | | |K |L
10609 BCDEFKL B |C |D |E |F | | | | |K |L
10608 CDEFL C |D |E |F | | | L
10607 CDEFL C D |E |F | L
10606 CFL C F L
TPO Analysis
CENTER 10612
VALUE AREA FROM TPOS
UPPER 10617
LOWER 10609
Figure IDO 9. Market Profile for T-bonds, March 21, 2001.
After the seven point drop in the first two periods, the market
is in congestion the rest of the day.
The latest trading day, March 22, has value area of 10705 to 10625.
It shows congestion, trend and then large congestion.
MV2) Initial trading is slightly above and inside previous value
MV3) Trend: breakout from the congestion at 10620 with a run to 10628
MV4) Congestion for the rest of the day, a sign of trend termination
MV5) Close of 10628 is well down into the congestion region
Conclusions from Market Value behavior
MV6) Value is higher on the day, but got there early (B period)
MV7) Market showed congestion early, during first hour or so
MV8) Market spent last 5 hours in congestion
MV9) Except for the quick run in B period this is a congesting market
MV10) Value at 10705 - 10625 provide support/resistance for tomorrow
MV11) Price nearing 10705 (upper limit = 10706) is a warning of impending
breakout
MV12) Price below 10625 is a sign of weakness
Trading Strategy (TS) for March 23, Basis both Condition and Value:
Note that all the information used is market developed. Also remember
that market condition can change overnight as happened in the Swiss franc
example. The trader reads the market and determines a strategy based
on current conditions. Any substantial change will be obvious, requiring
an upgraded analysis.
TS1) The market is in balance. Price above 10706 is an upside breakout
Price below 10528 is a downside breakout
TS2) Risk on breakout for the swing trader is around $330
TS3) Risk on breakout for the day trader is around $160
TS4) Early congestion followed by massive later congestion on 3/22
is indicative of a market confused about underlying demand
TS5) A breakout tomorrow is unlikely because of the congestion picture
in the last few market hours of 3/22.
TS6) This is a low priority market for the breakout swing trader
TS7) If tomorrow open is still in the upper area of the Overlay, day
traders are looking to short any turndown. If prices reach
near the bottom of the Overlay, we will seek to buy bottoms.
TS8) If the upper limit (10706) is exceeded, day traders change to looking
to buy into upturns.
TS9) Upper Limit (10706) and upper value area (10705) are nearly
coincident. Price there is strongly bullish.
TS10) Day traders turn bearish below 10625, seeking to short downturns.
Trading strategies TS1 through TS10 come from a direct reading
of the auction market variables. Another seasoned trader may use the same
data in a somewhat different way, differentiated by experience. The starting
point is the same: trading on
3/22 began with an upside thrust, a breakout, and then traded down while
congesting. The previous day, 3/21, ended in a much more symmetrical
balance and that day's Market Profile was likewise quite normal for
trading in a balance.
So 3/22 is a colossally failed breakout. Why? How soon in the day's
development could a market savvy trader catch on? Congestion tells the
tale. We are looking for that transition from trend to balance. We can
recognize congestion graphically as in figure IDO 10. But if we know
more about markets, we have a chance to do some intelligent guessing.
LENGTH OF FIRST PERIOD = 10 MINS
MARKET PROFILE* REPORT FOR 03 22 01
AND SEGMENTED AUCTION
COMMODITY -- T-BOND (CBOT) DAY JUN 01
Price Brackets Segmented Auction
10708 F F
10707 F F
10706 FIK F I K
10705 FIKL F |I | |K |L
10704 EFIKL E F | | |I | |K |L
10703 EFIJKL E |F | | |I |J |K |L
10702 EFGHIJKL E |F |G |H |I |J |K |L
10701 EFGHIJKL E |F |G |H |I |J |K |L
10700 DEFGHIJKL D |E |F |G |H |I |J |K |L
10631 CDEFGHJKL C D |E |F |G |H | |J >K >L
10630 CDEFGHJL C D |E |F |G |H | >J | |L
10629 CDEGHJL C D |E | |G |H | |J | |L
10628 BCDEGHL B C D |E | |G |H | | | |L
10627 BCDEGHL B C D |E | |G |H | | | |L
10626 BCDEGH B C |D >E > >G >H > | | |
10625 BCEGH B C | |E | |G |H | | | |
10624 BCE B |C | |E | | | | | |
10623 BCE B |C | |E | | | | |
10622 BE B | | |E | | | |
10621 BE B | | |E | | |
10620 BE |B | | |E | |
10619 yAB |y | |A |B | | | |
10618 yzAB >y |z |A >B > > | |
10617 yzAB |y |z |A |B | |
10616 yzA y >z >A | | |
10615 zA |z |A | | |
10614 zA z A | | |
10613 zA z A | | |
10612 z z | |
10611 z z | |
TPO Analysis
CENTER 10631
VALUE AREA FROM TPOS
UPPER 10705
LOWER 10625
Figure IDO 10. Market Profile for T-bonds, March 22, 2001.
After moving out of the y-z-A congestion the market struggled to
a top in F period. From C period through the rest of the day
the market is congesting.
Short Covering Rally
A common phenomena in markets is the 'short covering rally'. Conceptually,
imagine that many of the local members on the floor end the day short,
rather than their more usual flat. After a sleepless night, they come to work
eager to exit. As professionals, they know better than to exit all at once.
Each one is looking for an exit that hurts the least. Some trade immediately
and some wait.
The net is that the market sees demand
over the period in which the members are buying in their shorts. This period
is typically an hour or two. During the time the members are net
buying, public interest is aroused. The public carries the price on up until
they realize demand has evaporated. But this takes time. The market
is not efficient. The TPO shape of a short covering rally is that of a capital
P. Price runs up, stopping past the point where the excess demand is gone.
Then there is a period of backing and filling, forming the loop of the
P. Look at figure SC 10 again. Do you see the P?
Now we understand the overloading toward the upper prices in the Overlay
for March 22 (figure IDO 8). The upside breakout was likely driven by a short
covering rally. It was merely an accident that the rally occurred near the
breakout of the Overlay. Now we have evidence for the failure of the
trend. No wonder the Market Profile for March 22 did not fit in with the
prior four days.
Additional Market Analysis from Short Covering Data
TS11) The odds are that the Overlay tomorrow will pull back, i.e. 10708 is
a local high.
TS11) Unless new upside demand enters the market, the odds are that the
Overlay tomorrow will pull back, i.e. 10708 is a local high.
TS12) Understanding the probable cause of the rise on March 22 does
not substantially change our strategy for March 23. Corroboration
adds confidence in the original analysis.
Buy/Sell Confirmation of the Original Premise for Short Covering
We cannot look into the minds of the floor traders. But sometimes
we can see what they have done. The Chicago Board of Trade releases an
end-of-day Buy/Sell report. These data list the
four classes of member's volume at each price and also how much of the
activity is buying and how much is selling.
The Buy/Sell Report for
March 21 is in figure IDO 11. For the Locals, CTI1, it lists the buying,
selling and net for each price, and totals at the bottom. Floor traders
indeed ended the day selling more than they bought by over 1000 contracts
(2108 sides = 1054 equivalent contracts). Yes, on the 22nd, Locals probably
came to work with latent demand and an itch to get out.
Updated on March 21, 2001 at 20:56 for US 01M Traded on March 21, 2001
Net Buy and Sell/Bracket Information
___________________________________________________________________________
Price Volume CTI1b CTI1s CTI1n CTI2n CTI3n CTI4n Half-hour Brackets
Z$ABCDEFGHIJKLM
_____________________________________________________________________________
10626 2010 53 644 -591 -35 -206 832 Z
10625 1796 516 264 252 20 98 -370 Z
10624 864 259 294 -35 5 -48 78 Z$
10623 5834 1663 1575 88 26 278 -392 Z$
10622 3914 1086 1143 -57 280 57 -280 Z$
10621 4696 1776 1215 561 -70 -97 -394 Z$
10620 6726 1974 2307 -333 66 -20 287 Z$
10619 5198 1690 1439 251 -207 -41 -3 Z$
10618 4188 1503 1333 170 4 -45 -129 $B
10617 7388 2113 2736 -623 -263 322 564 $ABCGHJ
10616 12732 3572 4117 -545 357 -166 354 ABCGHIJ
10615 24336 6729 7848 -1119 458 -155 816 ABCGHIJK
10614 22922 7033 7287 -254 345 -596 505 ABCFGHIJK
10613 23874 6659 6593 66 -404 -95 433 ABCFGHIJK
10612 13172 3902 3748 154 200 -426 72 ABCEFGHIJK
10611 15886 4586 4862 -276 -14 -62 352 BCEFHIKLM
10610 16566 4226 5195 -969 16 -232 1185 BCDEFHKLM
10609 12748 3718 3643 75 -491 174 242 BCDEFKL
10608 16040 4379 5010 -631 163 -211 679 CDEFKL
10607 12728 4177 2897 1280 -339 355 -1296 CDEFL
10606 1246 519 91 428 0 0 -428 CVL
___________________________________________________________________________
Grand 214864 62133 64241 -2108 117 -1116 3107
Total
Figure IDO 11. Buy/Sell statistics for T-bonds (day), March 21, 2001.
CTI1, floor traders buy (b), sell (s) and net (n) volumes at each price
culminates in a net sell of 2108 sides (side = 1/2 contract). The other
three classes of traders (CTI2 = Commercials, CTI3 = Off Floor Members and
CTI4 = Members Trading for the Public) show the net only. Market Profile
symbols are Z = 07:20 to 07:30, $ = 07:30 to 08:00, A = 08:00 to 08:30.
B = 08:30 to 09:00 and so on.
Additional Market Analysis from Buy/Sell Data
TS13) At the end of March 21 the Locals were net short 1054 contracts.
Analysis for March 22 would suggest a potential net demand
from the floor traders.
Commercial Capping
Paragraph 11) mentioned commercial capping; the process where the commercial
members (CTI2) sell heavily at the top (or buy heavily at a bottom) to
push price back to balance. March 22 T-bonds moved up on demand that
was exhausted at the top. Did the commercials aid the price drop?
In figure IDO 3 the CTI2 average volume for the day is 6.1 percent
of the total. Going down the %CTI2 column we see the first two values
of 14.7 and 8.7. Both are substantially larger than the 6.1. The path
of price in F period (10:30 to 11:00) is down from 10708 to 10630.
Indeed, it appears the commercials capped and drove price well back to
the middle.
Additional Market Analysis from Commercial Capping Data
TS14) Commercial selling at the top indicates the public does not have
enough buying power to keep the upward trend in place. Again,
commercial data confirms analyses TS4, TS5, TS6 and TS11.
Volatility Information
In the volatility discussion above, the breakout day has a volatility of 9.1.
Since the long term average is in the 8's, 9.1 is just above the average,
and hence does not imply trend. In this instance, volatility merely confirms
the other conclusions. Another market situation may well find volatility
much more important in helping unravel the market's message.
Conclusion
Auction Market Theory shows the structure and patterns of auction
markets. It provides the tools to convert price to value and value change
(Market Profile) and to market condition and risk (Overlay Demand Curve).
The theory deconstructs a
market from it's current condition. To look inside, so to speak.
In addition to value, condition and risk, one can know which
prices are accepted, which rejected. It is often possible know what
the members are doing. Yes, markets can be understood. One has
the salient facts and these facts lead to conclusions. We call these
conclusions 'strategy'. Understanding ones markets imbues us with
a confidence unfamiliar to most traders. When you know, and know that
you know, confidence replaces fear.
A big advantage of understanding the market and setting up a strategy
based on that understanding is that if a strategy turns out to be
wrong it is very quickly apparent. The swing trader will know when
a breakout fails. A swing trader will also have a strong clue when
a trend falls into congestion--in hours, not days. Day traders, too,
will usually know when market conditions change, say from balance to
trend, and so can react accordingly.
The generality of the theory makes it a starting point for much new
market research. One area, just being explored, is the measurement
of reward to risk ratios. An early finding is that the Dow Jones Index
has a reward to risk ratio about twice that of the SP Index. Another
area is categorizing markets by trading opportunity. Now that the
initial development is in place, and with a theory to lean on, there is a
vast arena of practical market applications waiting to be discovered.
Unfinished Business
But wait, a lot of analysis went into developing a strategy for trading
on March 23. How did it work out? The trading strategy TS1 - TS14
indicated a small liklihood of any further upward activity. No new demand
entered. The market of March 23 confirmed the analysis. It is a classically
'dead' market. The events of March 22 took the wind out of the trader's
sails. The day market on March 23 opened at 10610 and stayed within 10
ticks of that price all day long (figure IDO 12).
LENGTH OF FIRST PERIOD = 10 MINS
MARKET PROFILE* REPORT FOR 03 23 01
AND SEGMENTED AUCTION
COMMODITY -- T-BOND (CBOT) DAY JUN 01
Price Brackets Segmented Auction
10615 H H
10614 BCH B C H
10613 BCFGH B C F G |H
10612 BCFGH B C F |G |H | |
10611 yzABCDFGHI |y |z A |B |C |D | |F |G |H |I | | |
10610 yzABCDEFGI >y |z |A |B |C |D |E |F >G > >I | | |
10609 yzABCDEFI |y >z >A >B >C |D |E |F | | |I | | |
10608 yzABCDEFIJK y |z |A |B |C >D >E >F | | |I >J >K >
10607 zBCDEFIJK z |B |C |D |E |F | | |I |J |K |
10606 zBCDEJKL z B |C |D |E | | | | |J |K |L
10605 BCDEJKL B C |D |E | J |K |L
10604 BCDEJL B C D E J L
10603 BDEL B D E L
10602 DEL D E L
10601 D D
TPO Analysis
CENTER 10608
VALUE AREA FROM TPOS
UPPER 10611
LOWER 10605
IDO 12. Market Profile for T-bonds, March 23, 2001.
The market congested all day.
Recalling some of our analyses:
TS4) Early congestion followed by massive later congestion on 3/22
is indicative of a market confused about underlying demand
TS5) A breakout tomorrow is unlikely because of the congestion picture
in the last few hours of 3/22.
TS6) This is a low priority market for the breakout swing trader
TS11) Unless new upside demand enters the market, the odds are that the
Overlay tomorrow will pull back, i.e. 10708 is a local high.
TS12) Understanding the probable cause of the rise on March 22 does
not substantially change our strategy for March 23. Corroboration
adds confidence in the original analysis.
The Market Profile of March 23, in figure IDO 12, fits neatly into the
Overlay of March 21. The breakout on March 22 is shown to be a
transient, not due to any pemanent change in demand or value. Thus, the trader
can totally discard the action of March 22. Trading action of March 22
did not alter the value picture of the market. Trading analysis for Monday,
March 26 can be based on figure IDO 13, the Overlay of March 21!
TPO VOLUME OVERLAY AND PRICE ROTATION PROFILE
JUN 01 T-BOND (CBOT) DAY
03 15 01 TO 03 21 01
PRICE DYS L/F ROT PROFILE * TPOS TPO VOL OVERLAY *
10629 1 6 1 X
10628 1 6 1 X
10627 1 6 1 X
10626 2 9 69 4 XXXX
10625 2 9 69 5 XXXXX
10624 2 9 69 6 XXXXXX
10623 2 9 69 6 XXXXXX
10622 2 9 69 6 XXXXXX
10621 2 9 69 6 XXXXXX
10620 2 9 69 9 XXXXXXXXX
10619 2 9 69 9 XXXXXXXXX
10618 2 9 69 10 XXXXXXXXXX
10617 3 59 569 13 XXXXXXXXXXXXX
10616 3 59 569 13 XXXXXXXXXXXXX
10615 3 59 569 15 XXXXXXXXXXXXXXX
10614 4 59 5679 18 XXXXXXXXXXXXXXXXXX
10613 5 59 56789 23 XXXXXXXXXXXXXXXXXXXXXXX
10612 5 59 56789 25 XXXXXXXXXXXXXXXXXXXXXXXXX
10611 5 59 56789 27 XXXXXXXXXXXXXXXXXXXXXXXXXXX
10610 5 59 56789 30 XXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
10609 5 59 56789 25 XXXXXXXXXXXXXXXXXXXXXXXXX
10608 5 59 56789 27 XXXXXXXXXXXXXXXXXXXXXXXXXXX
10607 5 59 56789 24 XXXXXXXXXXXXXXXXXXXXXXXX
10606 5 59 56789 23 XXXXXXXXXXXXXXXXXXXXXXX
10605 4 5 5678 15 XXXXXXXXXXXXXXX
10604 3 5 578 12 XXXXXXXXXXXX
10603 3 5 578 18 XXXXXXXXXXXXXXXXXX
10602 3 5 578 19 XXXXXXXXXXXXXXXXXXX
10601 3 5 578 19 XXXXXXXXXXXXXXXXXXX
10600 3 5 578 18 XXXXXXXXXXXXXXXXXX
10531 3 5 578 13 XXXXXXXXXXXXX
10530 3 5 578 13 XXXXXXXXXXXXX
10529 2 5 58 11 XXXXXXXXXXX
10528 2 5 58 11 XXXXXXXXXXX
10527 2 5 58 4 XXXX
10526 1 5 5 2 XX
10525 1 5 5 2 XX
10524 1 5 5 1 X
Figure IDO 13. Five Day Overlay Demand Curve for June 2001 T-bonds
3/15 - 3/21.
Part 2. Reward and Trading Models
Rewards
Initial Conditions
Market Condition
Balance
Transition to Trend
Trend
Transition to Balance
Trading Models
Model Evaluation
One Necessary Parameter
Finding Trade Entry
A Standard Model for Auction Markets
The Basic Model
Conclusion
Gains and losses are the natural result of trading. One's reward is
is the cumulation, the sum of the trading results less the fixed costs
of the endeavor. While the fixed costs are amenable to some control, by
far the larger part of one's reward lies with the trading gains. This
apparent fact leads most traders to concentrate on developing trading
models, the second part of this section. But just jumping to model development
is putting the cart before the horse.
Missed in most trading
research is the need to quantify what one might expect as a reward.
Putting 'Reward' into perspective means putting one's expectations into
some quantifiable form.
Any individual trade can be a winner or a loser. Experience of successful
traders places the expected win ratio at 30 to 50 percent. If, in practice,
one wins 50 percent of the time there is a 50 percent chance of one loser in
a row, a 25 percent chance of two losers in a row and one-tenth of one percent
chance of ten losers in a row (one in a thousand). Any month with 125 trades
(about six per day)
will likely have a string of seven losses somewhere in the period. If one
trades regularly, a long string of losses at some point is a virtual certainty.
When a string of losses hits, the ordinary trader begins to have doubts. Not
only is the overall reward in danger but one's methodology comes into question.
The entire enterprise starts to crumble.
What is the solution to the natural behavior of the laws of chance in the
trading enterprise? Reward preservation has several facets. These are:
1. Trading should follow a well defined and researched
model as
discussed in the Trading Models section below.
2. Any market traded should be understood from the probablilty
of return standpoint. Some markets offer
substantial opportunity
and some offer little. Opportunity is quantized
as Potential,
the maximum possible gain offered after trade
entry. Risk is
another factor. One desires to maximize the
Potential/risk ratio.
Potential and risk/reward are treated in Part 3,
Standards.
3. Diversification. All markets traverse periods of high
activity
and low activity. A once high opportunity market
can become one of low
opportunity. Or a high opportunity market may
become overheated, with
the accompanying risk generating a low reward to
risk ratio. Clearly,
one should be aware of the opportunity/reward/risk
factors of all
the markets in one's trading universe.
Diversification studies of
spreading one's trading risk among several
markets all show the
benefit of such practice.
Trading is a disequilibrium market activity. Price appreciation
opportunities
are found only where value is changing. The most potentially profitable
situations
are the ones where value change is quasi-monatonic, a trend moving in a fairly
straight line. Capturing
appreciation requires recognition of the inception of a trend and a
realization
of when the trend is over. Normally, the recognition and monitoring process
is incorporated into
a set of trading rules, a model. Specification of a
completely defined model
removes all flexibility from the trader. On the other hand, specifying
no model at all makes it difficult to proceed with a research study (and
orderly
trading!). Successful traders seem to follow an 'expert-systems' process
where there are some
rules, with the additional overrides of market knowledge as needed.
We are guided here by the 'expert' concept; defining some parameters and rules,
while
leaving substantial room for decision to the trader. The aim is to have as a
base,
a bare bones model that is simple, and yet fundamentally sound. A "plain
vanilla"
model, while easy to understand, still provides the unequivocal entry point
needed to measure a trade's Potential.
Trading is highly personalized. Some traders have a feel
for the market.
Others seek an organized model to guide their trading. These
latter are in the vast majority. They can be identified as "non technical"
traders since they will not, or can not develop their own models.
Help is offered by the industry. A recent issue of Stocks and Commodities
Magazine (July 2001) listed over 400 companies that sell trading products and
data. Very few of the 400 (possibly only one) apply Auction Market Analysis.
If the trader is to follow Auction Market techniques in trading, a
model must be found that is driven by value change, not price change.
Ideally, such a
model can be utilized by both the non-technical and more advanced traders.
Anyone who understands the rudiments of Auction Market Analysis could trade
the basic model, while the more advanced traders could build on the basic
model in developing their own personal methodology.
Initial Conditions
Profiting from the appreciation potential offered by changing markets requires
a model that is active if value is changing and that gets out or
stays out when
value is constant and stable. The most reliable and lowest risk alert
point is found when a market in balance breaks out. This is the only instance
in which a change in the quasi-normal distribution can be pinpointed.
Markets in which value is changing are often
'trending'. But 'trend' is a market condition that requires a time frame
and 'size' in it's definition. That is, a market may be trending
over the last three days, while on a longer timeframe (say 20 days) that same
market is in balance. Or a market can be in a trend that is so small that
the volatility hides it. The ideal trend trading model needs to be able
to identify market condition, to recognize a trend start and to determine
when the trend is over.
A valid model will most likely be time frame dependent. i.e. one sort of
model for
day trading, somewhat different rules for swing (multi-day) trading and
yet another for the long timeframe (CAPM). The primary initial condition
is that the market be in balance. The market selected should be
one that has a recent history of offering opportunity.
Market Condition
The trending phase of the market is just one of four conditions (balance, test,
trend, test, balance, etc.). The
form, the procession is fixed, although a particular phase, such as transition
to trend, may be very, very short to the point of non-existence. The time
spent in each condition varies according
to the market situation (a balance may be 3 days, 7 days or three months).
Balance
Balance is distinguished as a single price - volume distribution
of the Overlay Demand Curve (Fig. 4) with the latest close inside the
balance. Balance is a quasi-equilibrium state.
End prices (limits) are known, i.e. alerts for end of balance or breakout
from the single distribution.
Value is known. Although a trend is often hard to discover,
balance is easily recognizable. Risk is defined as a portion of the
distribution analogous to one standard deviation (practically, a common
approximation is one-eighth of the range of the balance).
Transition to Trend
A transition from balance moves the market into a dis-equilibrium
market condition, one with multiple distributions. A successful
transition results in the
start of a trend, i.e. a change in value. A failed transition,
i.e. a false breakout, results in trade nullification.
Trend
Trending is just price change reflecting changing demand.
A trend is a quasi-orderly dis-equilibrium market condition.
The path of a trend is generally not smooth, exhibiting runs and pauses.
The trend's final pause marks the beginning of the transition
back to balance; consolidation back into a single distribution.
Transition to Balance
This transition is from a dis-equilibrium market condition back
into (short term) quasi-equilibrium. Although the transition
itself may be fast (it usually is not), several days must pass
before the balance is apparent. As a balanced condition emerges, the
first day or two of the new balance are hard to recognize because of
volatility. After a few days, the Overlay process integrates out much
of the volatility and the single distribution becomes apparent.
Trends are easier to recognize when viewed within the context of the four
conditions, i.e. trend closely follows balance. Balance is easiest to define,
since it is the phase where value
is not changing. As balance evolves into transition, a measurable event, called
breakout, occurs.
Breakout is merely where price moves past the balance limit. This phase
can more properly be called 'alert for trend', since it is a test of demand.
If the breakout fails and
no trend develops, market condition normally reverts back to balance.
Theoretically, one might view the situation as the market breaking out and
rapidly passing through transition to trend, trend and transition to balance.
Or the failed breakout can be seen as simply that, an erroneous breakout
that was not supported by a change in value. Either way, the net result
is the same, a failed trade.
Trading Models
An assumption is that a valid trading model will be in tune with the
market. That is, parameters developed will come from current market
conditions (i.e. value), not from some predetermined measure (e.g. a ten
day moving average).
There are a multitude of trading models in the literature and offered by
vendors. Traders want them. Most are based on 'technical analysis'.
Some commercial models may be productive, but the evidence for success is
scanty. A fundamental problem with 'technical models' is that they are
based on fixed formulas and use price as the primary input variable.
Formula models suffer from the absence of 'expert' input. Many traders have
data bases. If a formula is known, there is no limit of traders that can
apply it. Traders will gravitate to the model and the
zero sum characteristics of trading will work to minimize the gains.
Adding costs makes the process negative-sum, and hence most will lose.
Statistics on futures and equities day trading show that up to 90 percent
of day traders lose.
Avoiding the pitfalls of formula trading is no mean task. In a zero-sum
environment the returns are 50 percent to winners and 50 percent to losers.
Costs ensure that less than half of the wager will be paid to winners.
This statistic
is inevitable in short term trading. To be profitable one has to somehow
be in the winning group--either winning a high percentage of trades or by
making one's winning trades significantly larger than ones losers, or both.
Appreciation accrues only when value is changing. A competent model will
be highly attuned to changing value, not necessarily changing price (see
figure IDO 7 and figure IDO 8 and discussion). It is apparent that in
the competition for trading return the edge goes to the trader who
understands markets.
Model Evaluation
The ideal for any model evaluation is to compare it to a standard.
Long term growth of the economy is reflected in market indexes. Over the
longer term of ten years or so, it is possible for up to 100 percent
of investors to win. Market indexes grow. CAPM portfolio returns can be
measured against an index. No index exists
for AMT. In the same time frame, one trader may do well while another may
lose. There
is no underlying index because there is no agreement on how to catalog
returns in the trading environment. In the period of a day,
a floor trader might make twenty trades, averaging 2 ticks gain per trade.
A (public) day trader might make three trades, averaging 10 ticks per trade.
Another day trader might make only one trade for 15 ticks. Who did better?
It is hard to tell. The three are not really comparable since they took
different risks.
Yet, the trading arena
can be good (offer large opportunity for gain) or bad (low opportunity).
How does one know good opportunity from bad? The key is in the Potential,
as developed more fully in Part 3 on Standards.
In a balance, as IDO 8, one knows the range (10706 to 10528) and the Octant
risk of 5 ticks. Risk derives from volatility. Further, risk and potential
reward are proportional (in general). A five point risk translates
to $156. Trading for a potential $156 reward is reasonable. If the market
volatility were low and risk
is 2 ticks, the potential reward is $62. Considering costs of some $30 and
the possibility of loss as well, the average trader would decline the
trade. If the trader used a particular model over a period of time, there
would be a track record. With experience with the model, one could know
which market conditions are good. But the problem remains; how
would one rank this particular model against other models? Standards are
needed.
One Necessary Parameter
An absolute evaluation of any day trading model can be made by defining
a single prior condition.
That condition is the entry price. Assume that
all models enter at the same price and a trade is not allowed to extend past
the exchange close. Then during the trade there is one best point to exit.
The only problem is locating the entry.
Finding Trade Entry
Specifying trade entry can be accomplished many ways. We will explore
and choose a tested Auction Market methodology using the balanced
Overlays. As before, we specify a breakout from the balance to
be an alert for trend start. In this case that breakout price
becomes the entry. No trading rules are defined
for the initial risk, the management of the trade throughout the day or
for the exit. A trader will make all the management decisions.
Obviously, different traders will make different decisions within the rules.
The standard for this trade is simply the point of highest open trade
equity during the period from entry to close. This standard is called the
Potential. To paraphrase, once trade entry is attained the Potential
becomes fixed, but appreciation from trading is not.
A Standard Model for Auction Markets
Requirements of a 'standard' model are 1) information on the balanced markets
within a set of all markets in the defined arena (futures, securities, etc.),
and the posted breakout price (s) from the balance. Possibly many
models meet these requirements. We will discuss one that has been
published, one that has been used by many traders as a starting point in
learning Auction Market Analysis. This is the Basic Model in the
Trader Control Package (TCP) of CISCO Futures (ref. 6). Requirement
1) is met by a list of balanced markets for the day: two such sources are
the Bracket Screen and the Advice Engine Report, both CISCO
products. Both lists are essentially just compilations of Overlay parameters.
The Basic Model
The Basic model is included in the CISCO training course. A training tool,
Basic Model comes in several varieties;
for day trading, swing trading and for responsive trading of balanced
markets (non-breakouts). Rules for the most elementary form of the breakout day
trading Basic Model are:
(1) limit trading to balanced markets only
(2) go long on a breakout of 1 tick above the Upper Limit
(3) go short on a breakout of 1 tick below the Lower Limit
(4) use a trailing stop of one Octant
(5) if not stopped out during the day, exit on the exchange close.
Only item (4) violates the freedom of action post entry discussed in
Model Evaluation. A stop-loss risk limit in the courses is a pedagogical
tool for teaching risk control. For the purposes of this discussion, item
(4) is not necessary. Practically, item (4) is useful as a starting
point in many research studies, since the risk is tied to the value range.
Conclusion
Part 2 presents a simple model with rules that are easy to implement.
With the unambiguous entry provided, Potential can be found for a wide
varitey of markets.
The next section on 'Standards of Comparison',
will turn to the development of trading efficiency measurement using the
Potential found with the Basic Model.
Part 3. Standards of Comparison
Contents
Standards
Trade Identification. Advice Engine Report
Experimental Standard
Potential
Models
Locating 'Best Trading Risk'
Figure IDO 14. Advice Engine Report June 13, 2001
Figure IDO 15. Select Table June 1 to June 1 15, 2001
Reward to Risk Ratio
Figure IDO 16. SP Reward/Risk Study November 6, 2000 to June 26, 2001
Non-equilibrium day-trading for appreciation is much less organized than long
term investing where returns can be compared to average market movements (indexes).
Day trading seeks out short term opportunities. There is no
over-all order in the short term, no index for comparison. An absolute test,
the test of ruin, does exist. But what of the trader who is successful at least
part of the time? How is that success rated? Could that trader have done
better? If so, how much better? There are two aspects to the rating
question for traders. First, how well does the trader locate opportunity?
Second, once opportunity is found, how well is it handled? Both questions can
be answered within the context of Auction Market Theory (AMT). Question one
has a solution in the Advice Engine Report discussed in this section, that is,
Advice Engine locates trading opportunities.
The answer to question two hinges on the Potential. Potential permits
comparisons: how well did Basic Model do compared to the Potential? How well
did a trader's model compare? Now the researcher can have an absolute to
measure models and performance against.
Trade Identification. Advice Engine Report
Short term breakout trading success comes from finding trends. Locating
a trend in progress and trading it is a high risk process. Trends are of
uncertain length and vary in intensity. Their price - volume distribution
is changing and thus is hard to catalog. Risk is uncertain, but the increased
volatility accompanying a trend insures high risk. Lower risk trading
opportunities are better found in the pre-breakout market. For futures, one
source of trading candidates is found in the Advice Engine Report.
Essentially the same approach can be used in equities and other auction markets.
A balanced market contains the distribution value based market
parameters that identifies breakout prices prior to breakout.
Risk is lower because 1) market condition is in equilibrium so
the distribution is stable and 2) volatility is low.
A breakout is a high probability, low risk
precursor of the start of a trend. When a trend does not develop
from a breakout, the loss is limited to the low risk of the
balanced quasi-equilibrium distribution, not the high risk of a trending,
thoroughly non-equilibrium
distribution. When a trend does develop, the trade offers the advantage
of early entry. The trade is then well positioned to take
advantage of the appreciation offered by the market.
Experimental Standard
Measuring the effectiveness of a trading procedure or model is more complicated.
The market situation in which the model operates is not in equilibrium. This
makes finding a closed mathematical solution of the optimum unlikely. The
trading function
is different for each trader and is fundamentally unknowable in general.
Yet, an experimental standard can be developed.
The primary condition is only
that a trade be initiated at the breakout point defined by AMT. Other
subsidiary conditions relating to time can be applied for special cases, but
for the basic standard for day trading, we specify that all trades not
otherwise closed out will be terminated at the close of exchange trading for
the day. Conditions, then, are two: enter on the AMT breakout, exit
no later than the close for the day. No information is needed on the
explicit model followed by the trader.
Potential
A trade can be either profitable or unprofitable. The best possible
result comes when the trade is not stopped out but exits on a close
that is the highest post entry price (if a long trade) or the lowest
post entry price (if a short trade). Normally, the best open-trade price
occurs not at the close, but at some intermediate time after initiating
the trade. The best open-trade price is the Potential of that trade.
Potential is always positive or zero, while the actual trade may win or lose.
Potential measures a trade's maximum possible return. Potential
plays a role akin to the index in CAPM. It is a quantity against which
the