-
web site: www.sta-uk.org The Journal of the STA
1
Members might be interested to know that the committee
iscurrently working on an initiative to develop a strategic
alliancewith Robert Walters Associates, the recruitment
consultants. Itis hoped that, through Robert Walters, the STA will
be able toestablish a dedicated technical analysis recruitment
service forthe benefit of members of the Society. At the time of
going topress some of the details are still being finalised but we
hopeto be able to make a full announcement shortly.
The Societys website has now been up and running for twoyears
and we thought it would be a good idea to look at howwe might
expand the site to make it more inter-active. Ifanyone has any
ideas or suggestions as to what they wouldlike to see included,
please contact Gerry Celaya.
Anyone wishing to take the Societys Diploma should notethat the
next examination will be held on October 4th andthe cost for taking
it will be 235. For the first time this yearthere will be a choice
of two levels for our introductorycourse on technical analysis. One
course will last for fourweeks and the other for six weeks. The
cost for these courseswill be 225 and 295 respectively. Both
courses will beginon 31st October. In recent years we have had a
growingnumber of enquiries about whether the Society would
beprepared to run a distance learning course on technicalanalysis
and we are now in the throes of preparing one. Weare grateful to
all the people who have agreed to contributeto this project. More
details will be released later in the year.
As some of you will have seen in the IFTA Update, the IFTAboard
is spearheading a drive to build a Body of Knowledge
reference work on technical analysis. The workload is
beingdivided up between the member societies. Adam Sorab is
co-ordinating the northern European input on the subject ofprice.
He would be very pleased to hear from members of theSTA who would
like to be involved in this project.
On behalf of all the members who attended the summerparty, we
would like to thank Reuters for hosting such anexcellent occasion.
Glyn Bradney started the evening offwith a talk explaining how he
tries to identify energy levelswithin the markets. His ideas
prompted a large number ofquestions from the audience and those of
you who werenot able to attend will be pleased to know that Glyn
haspromised to do a comprehensive write-up of the subject inthe
Journal.
Finally Luise Kliem is stepping down from the board and wewould
like to take this opportunity of thanking her for all thework that
she has done. In particular, Luise has been anextremely active
member of the last minute speakers club,often agreeing to step in
and talk at the monthly meetingswith only a day or sos notice when
the scheduled speakerhad had to cry off for some reason.
COPY DEADLINE FOR THE NEXT ISSUE31 AUGUST 2000
PUBLICATION OF THE NEXT ISSUEOCTOBER 2000
July 2000 ISSUE No. 38
MARKET TECHNICIAN
FOR YOUR DIARY
IN THIS ISSUE
Wednesday 13th September Introduction to Ichimoku Cloud
FormationsNicole Elliott and Yuichiro Harada, Industrial Bank of
Japan.
October 5th 7th IFTA Conference
Wednesday 11th October STA AGM
N.B. The monthly meetings will take place at the Institute of
Marine Engineers, 80 Coleman Street, London EC2 at 6.00 p.m.
Exam Results 2
Book review 3
Exam Report 3
G. Celaya Charting the regional FX markets. 4
L. Kliem Equity markets: Outlook for Wall Street and FTSE TMT
and theold economy. 9
M. Wignall Mutual offsetting systems a problem for technical
analysis? 11
N. Burnton ChronographicsSM 12
Z. Harland Trading a 2 year old the real-timeperformance of a
neurogenetic T-bond futures trading system. 14
IFTA Conference 2000 16
-
WHO TO CONTACT ON YOUR COMMITTEE
CHAIRMANAdam Sorab, Credit Suisse First Boston, Five Cabot
Square, London E14 4QR. Tel: 020-7888 7240
TREASURERVic Woodhouse. Tel: 020-8810 4500
PROGRAMME ORGANISATIONMark Tennyson dEyncourt. Tel: 020-8995
5998 (eves)
LIBRARY AND LIAISONMichael Feeny. Tel: 020-7786 1322The Barbican
Library contains our collection. Michael buysnew books for it where
appropriate, any suggestions for newbooks should be made to
him.
EDUCATIONJohn Cameron. Tel: 01981-510210Clive Hale. Tel:
01628-471911George MacLean. Tel: 020-7312 7000
IFTAAnne Whitby. Tel: 020-7636 6533
MARKETINGSimon Warren. Tel: 020-7656 2212Kevan Conlon. Tel:
020-7329 6333Tom Nagle. Tel: 020-7337 3787
MEMBERSHIPSimon Warren. Tel: 020-7656 2212Gerry Celaya. Tel:
020-7730 5316Barry Tarr. Tel: 020-7522 3626
REGIONAL CHAPTERSRobert Newgrosh. Tel: 0161-428 1069Murray Gunn.
Tel: 0131-245 7885
SECRETARYMark Tennyson dEyncourt. Tel: 020-8995 5998 (eves)
STA JOURNALEditor, Deborah Owen, 108 Barnsbury Road, London N1
0ES
Please keep the articles coming in the success of the
Journaldepends on its authors, and we would like to thank all
thosewho have supported us with their high standard of work. Theaim
is to make the Journal a valuable showcase for membersresearch as
well as to inform and entertain readers.
The Society is not responsible for any material published inThe
Market Technician and publication of any material orexpression of
opinions does not necessarily imply that theSociety agrees with
them. The Society is not authorised toconduct investment business
and does not provideinvestment advice or recommendations.
Articles are published without responsibility on the part of
theSociety, the editor or authors for loss occasioned by anyperson
acting or refraining from action as a result of any viewexpressed
therein.
Results of Diploma ExaminationAPRIL 2000
DISTINCTION
Anthony Fletcher
James OConnor
Angelina Ng
PASS
Des Bailes
R. Crozier
Valerie Enguehard
David Franklin
Stephen Hatton
Michael Hugget
Julian Mahne
Martin Miller
Darren Read
Bert Reymenants
Soon Hock Gary Tan
Duncan Webb
2 MARKET TECHNICIAN Issue 38 July 2000
Networking Exam Results
ANY QUERIES
For any queries about joining the Society, attendingone of the
STA courses on technical analysis ortaking the diploma examination,
please contact:
STA Administration Services(Katie Abberton)
P.O. Box 2, Skipton BD23 6YH.
Tel: 07000 710207 Fax: 07000 710208 www.sta-uk.org
For information about advertising in the journal,please
contact:
Deborah Owen
108 Barnsbury Road, London N1 OES.
Tel/Fax: 020-7278 7220
-
Issue 38 July 2000 MARKET TECHNICIAN 3
THE PREDICTORS:
How a Band of Maverick Physicists set out toBeat Wall Street
By Thomas A. Bass (Penguin Press 18.99)
If you read James Gleicks wonderful Chaos, youll remember
theDynamical Systems Collective, a group of hippie physicists
sharing aratty student house in California in the 1970s who made a
set ofdiscoveries crucial to the infant science of chaos theory.
They foundways to map the messy, turbulent processes of the
everyday worldby considering them as flows of information. Once
they startedlooking, they found unexpected order everywhere: a
dripping tap, arising column of cigarette smoke. Later,
rechristened theEudaemonians, some of them tried to win at roulette
by putting toe-operated computers into their shoes. Weird science
against thepolyester barons of Las Vegas! Thomas Bass wrote a book
about it,The Newtonian Casino.
In the early 1990s, after several respectable years in academia,
twomembers of the Collective, Norman Packard and Doyne
Farmer,matched wits again with the world of money. Farmer, at the
LosAlamos National Laboratory, had come up with a technique
formaking non-linear predictions projecting the futures chaos
fromthe presents. At the Institute for Advanced Studies, Packard
haddevised adaptive computer algorithms that proved
spectacularlygood at recognising patterns in whatever he pointed
them at, fromsnowflakes to the spending habits of Italian municipal
bureaucrats.Together, these seemed like the keys required to make a
profitableintervention in the jetstream of capital that flows
continuouslyaround the globe from exchange to exchange, formidably
divorcedfrom the actual business of making, moving and selling
things.
As Bass explains, in a beautiful analogy, the streaming
trillions ofdollars really do behave like a fluid in a state of
turbulent flow. Theintricate structure of gas turbulence was
revealed when a photographwas taken in a wind tunnel by the light
of a high-voltage spark.Anyone who had the means to fix an
equivalent high-definitionpicture of the fractal whirlpools that
form in money couldtheoretically divert some of it their way.
Packard and Farmer and a gaggle of younger colleagues set up
inSanta Fe in an office equipped with the most expensive
Sunworkstations they could afford, and the cheapest possible
gardenfurniture to sit on, and started building a system to play
the markets.Bass writes well about the intellectual questions that
developed asthey used physics to do economics.
Effective market theory, for example, holds that price
fluctuationsare purely random when buyers and sellers have full
access to marketinformation if true, it would have made the task
impossible. Butthis is not really a popular science book; it comes
from the nexus inour culture at the moment where money and
technology make eachother exciting. Here, as in Wired magazine,
technology soaks upmoneys status as pure potential, stuff that can
satisfy every desire,while money acquires technologys quickness,
and above all, itsscalability, its power to expand suddenly by
orders of magnitude, asInternet systems can. It comes true like a
granted wish, it inflates likethe airbag in a limousine.
Basss story is a business history, told with a nice sense of
line, and anelegant eye for oddities of the interface between the
bankers in theirsuits and the physicists working in their shorts,
under MoonMountain and Sun Mountain, which grace Santa Fe like a
spiritualWonderbra.
By 1995, the Prediction Companys models were making money.
Lotsof it. Regularly. Experimental verification cant get any
solider. Youcan compare them to Marx, who also thought he had
detectedhidden structure in the market-place. His predictions
failed. The real-world experiments of the last century show that
all history is not, asit happens, the history of class struggle.
But Marxs structure did
allow for the possibility of people taking deliberate,
rational,concerted actions. The temporary laceworks of structure
that thePrediction Company spots, as they slide across the flanks
of theworlds financial waves, do not allow you to plan to make the
worlddifferent. They can only be used to make you individually
rich. Thismay well be a more innocent dream: it is also a smaller
one, and tosettle for it is to lose something.
Differences in style soon separate the former members of a
scientificcommune from the Swiss Bank Corporation, their
backers.
The price of Farmer and Packards access to venture capital is
thatthey abandon every trace of hippie egalitarianism. It takes
them justtwo days to decide to do so. They learn to fire people
(eventually,every founding member except themselves) and to spout
businessclaptrap about team playing and empowerment. Its a long way
fromFarmer saying I want to beat the system, not join it, yet
anotherdemonstration that maverick, as used in this books subtitle,
is anAmerican term meaning conformist. There is something
terriblysad about the way that they go on acting out their original
tasteswhen theyre millionaires, except on a much, much larger
budget.When you have reached a certain degree of personal wealth,
surelyits unseemly to wear an Eat the Rich T-shirt?
Francis Spufford
This article originally appeared in the book review page of
the
Evening Standard, 31.1.00.Evening Standard
Examiners Report Candidates faced a tougher examination than
previously. Highermarks were needed for both pass and distinction
and a morerigorous approach was taken to the framing of questions.
Indeed inthe opening section the choice was reduced to two
questions asopposed to three in order to avoid an easy option. Once
againassistance was given in preparing questions on a specialist
topic bythe Diploma course lecturer concerned.
There were 23 candidates of whom 3 gained distinctions, a
further12 passed and 8 failed. The average mark was 61.3% (compared
to61.3 last year, 60.8% in 1998 and 60.9% in 1997).
The percentage of candidates failing rose to 34.8%, which
ismarginally worse than the 33% students are warned to expect and
isin line with other courses at academic establishments.
There was, however, an improvement in the general quality of
workpresented and this may well stem from another change in this
yearsexam: time allowed has been extended from two and a half hours
tothree. An omission from a full complement of answers is much
rarerand fuller answers are common.
Point and Figure is an option that most candidates choose and it
isan obvious strength of the South Bank Diploma course since
mostare well conversant, if not yet expert, in the technique.
Candidates seemed well prepared and the quality of
communicationhas improved. Revision day was held nearly three weeks
ahead ofthe exam permitting more time for its influence to help
guidecandidates.
Despite the increased failure percentage, the average mark
remainsthe same. It does mean that the general standard has
improved.Students have manifestly worked hard; the lecturers on the
Diplomacourse are to be congratulated; the examination author
andadministrative team also deserve much credit.
Book Review
-
4 MARKET TECHNICIAN Issue 38 July 2000
Trading in the FX market has changed dramatically over the last
fewyears. While this is still a $1.3 trillion per day market, as
outlined inthe article All change in the FX market published in the
IFTAmagazine last year, the market has gone through a period of
change.The consolidation of the FX industry is partially due to
theintroduction of the euro and the headline-making bank
mergers,both of which have led to margins being squeezed in
traditional FXmarkets. This makes banks more eager to deal in
exotic marketsthat they would have avoided in the past, as the
margins in lessliquid currencies are still wider, hence making them
more profitablefor dealers to work in. On the other side of the
trade, the Asiancrisis of 1997/1998 highlighted the need for
corporate andinstitutional customers to be pro-active in their FX
risk managementas the sharp swings can affect returns in a quite a
large way. Chartsare very useful for spotting opportunities and
helping to determinethe risk / reward ratio that is critical to any
trade or investment,especially in regional markets where strict
adherence to theunderlying fundamentals or listening to the locals
can be very,very wrong! For example, it was inconceivable to many
in Asia inearly 1998 that by cutting interest rates the local
currency wouldstrengthen. The charts however identified potential
reversal patternswhich showed that the market was re-evaluating the
trend and thatcutting interest rates, despite what the local
experts were saying, wasbeing well received by the market.
Some key things to keep in mind are: Charts and common sensecan
help...
Basic charting techniques are best (patterns, moving
averages)
Know the local market conventions and legalities
Yield enhancement versus real risk must be considered at all
times
Dance like a butterfly, sting like a bee (the exit doors can
getrather small!)
CARRY TRADES AND YIELD ENHANCEMENT
Many investors and funds trade in regional currencies in order
toenhance the yield that is available to them. This is usually done
ona leveraged basis by borrowing funds in a low yielding currency
andinvesting the funds in a high yielding currency. The
potentialcarry gains are easy to calculate and the potential
positivemovement in the FX rates can be very attractive. Keep in
mindthough that the high yield on the currency is probably there
for areason, and that can make the potential losses on
adversemovement in the FX rates extremely high. The question that
mustbe asked on every trade is Does the potential reward outweigh
thepotential risk?.
Indicative yields for market amounts of $1 million USD or more
ondeposit for 92 days are: EUR 3.00% US 5.69 % GBP 5.68% Yen0.2%
CHF 1.75% South African Rand 10.50% Turkish Lira 25%Greek Drachma
8.25% Czech Koruna 5.13% Slovak 7.50% PolishZloty 17 % Norwegian
Krone 5.25% Icelandic Krona 10%Mexican Peso 16%
Carry trade = Lending yield borrowing yield or
Carry trade = high yield low yield which hopefully will
outweighdepreciation /devaluation risk!
Example1 million USD worth of EUR/ISK for 30 days = $6,250
(yield spread x amount x number of days)/360
on a $20 million portfolio = funding at 2.5% lending at 10 %
FXstable = $125,000 profit per month with no gearing. If this
was
done every month for 1 year there would be a $1,500,000
profit!Given the market norm of 10 times gearing, the $20 million
USDportfolio would gear up to $200 million and the profit would
be$15,000,000 per year. To repeat the main point, the question
iswhether or not the FX relationship is stable. If the spot rate
moves inan adverse manner the gearing can create a situation where
the $20million portfolio becomes a $5 million portfolio at the end
of theyear.
The Yen is the funding currency of choice for many carry
trades.With interest rates near zero the yen is attractive but as
the chartbelow shows, it is a very whippy currency. The move from
theY147.60 yen lows (USD high) of 1998 down to below 110 wiped
outmany of the carry trades that were funded with the yen as
theadverse spot rate move really hit the market. Looking at the
chartshows that a head and shoulders pattern (the first of many)
could bein place with targets near Y80. This serves to keep Yen
carry traderson their toes.
The chart below shows the THB/USD a weekly chart with 13 and50
week moving averages. The THB was until summer 1997 awonderful
currency to trade. The central bank (Bank of Thailand)ran the THB
as a basket currency, with the basket made up mostly ofYen and
Deutschemarks. Given that THB deposits paid a lot morethan Yen and
DEM deposits, borrowing the last two and lending inTHB was an easy
way to put a carry trade on, with the stabilityof the FX
relationship more or less ensured as long as the basket wasintact.
The summer of 1997 basically shattered that relationship asthe THB
was attacked by speculators, the BoT smashed thespeculators by
using the forward market and spot, driving spotsharply higher
against the USD before giving up, letting the THBfloat (no more
basket) and the Asian crisis was underway. Thelessons from the
early part of the crisis are clear, always be awarethat the exit
door in illiquid FX markets is smaller than the entry door.When
there is a panic big/offer spreads will widen out, options willbe
expensive (if available) and the disappearance of market
liquiditymeans that you really need to have a good relationship
with yourbank to trade these currencies for any hedging purposes.
From acharting perspective the blowout in 1997 was basically a
series ofbull flags on the daily charts, with market sentiment
following thetraditional fear / greed / fear / greed cycle. The USD
peak wasmarked by the market being in full panic mode, while the
firming ofthe THB against the USD in 1998 was marked by continued
disbeliefin many that the worst was over. Charts showed a pretty V
topwith a potentially massive head and shoulders pattern
dominatingactivity. Current activity is showing a potential
crossover in the 13and 50 week moving averages but for choice any
USD upmove will
Charting the regional FX markets
By Gerry CelayaThis article is a summary of a talk given to the
Society on 12th April, 2000
JPY=. Bid [Hi/Lo/Cl Bar] [(MA 50] Weekly
-
Issue 38 July 2000 MARKET TECHNICIAN 5
be limited to the 40.000 zone with the next major move
stillexpected to be in line with the major top pattern, towards
36/35.60again and lower for the USD (25.000 in a few years?). As a
sidenote, even though the USD rally in 1997 suggests that
thespeculators won the battle, George Soros is on the record as
sayingthat the BoT squeezed them and many others out of their
positionsbefore the THB really devalued, suggesting that the major
USDbuying through the 3rd and 4th quarters of 1997 was done
bycorporates and hedgers.
The chart below shows the THB/YEN relationship. As seen, the
pre-summer 1997 relationship was relatively stable. Since then the
crossplummeted (weak THB, strong YEN, hurting any carry trade)
andhas yet to show any convincing long term basing signals. For
choice,a flat trading range should continue to dominate.
The USD/SGD chart shows the Asia crisis effect pretty well
withSingapore in the middle of it all . The breakout of the flat
channel in
1997 was followed by a solid panic of buying USD. The SGDchart
is different from the other Southeast Asian FX charts becausethe
fact that the Malaysian Ringitt, its cross straits neighbour,
ispegged,has served as a dampener on activity over the last
18months. Technically, current activity shows pressure on
USDresistance and the risk of a break through 1.7360/7400 is high.
Forchoice, even though the turn above the declining trendline in
early2000 suggests that the USD will fly higher, the horizontal
resistanceline at the 1.7360/1.7400 area should hold up. A break
would bebad for the SGD and leave 1.76/1.78 open.
The SGD/JPY chart shows the effect of the strong Yen
sinceSeptember 1998 and while it could be basing at the 60 zone,
this isby no means confirmed. In fact, moving averages are still
bearishfor the cross rate; watch 66 as a turning point for a
reversal, 60 asthe zone to break in order to continue the trend
lower.
The USD/HKD chart below shows the extremes that the HongKong fx
rate went to during the crisis. The spike in the HKD fromthe 7.74
area to the 7.62 zone really happened, as the Hong KongMonetary
Authority took on the speculative players. The realaction from then
on was in the forward market, when rates(normally close to USD
rates) skyrocketed to 80% and even higherat one stage. The spot
rate, given the peg that the HKMA held to,did not move that much,
and has only moved to a weaker HKDrate. This chart highlights our
forecast for 7.800 in the coming 12months.
The Indonesian Rupiah chart (overleaf) shows the large
breakdownin 1997 in the IDR, with a run from 2500 to 17000 seen in
a littleover a year. Not only did the Asian crisis hit the IDR but
localevents deteriorated further which gave the IDR the second
spikeagainst the USD in the summer of 1998. Current activity
supportsthe USD with the 13 and 50 week moving averages
pushinghigher for the USD with 9200 open and spikes to 10,000
possibleon the charts.
THB=. Bid [Hi/Lo/Cl Bar] [MA 13][MA 50] Weekly
SGDJPY=R. Bid [Hi/Lo/Cl Bar] [MA 13][MA 50] Weekly
AKD=. Bid [Hi/Lo/Cl Bar] [MA 13][MA 50] Weekly
THBJPY=R. Bid [Hi/Lo/Cl Bar] [MA 13][MA 50] Weekly
SGD=. Bid [Hi/Lo/Cl Bar] [MA 13][MA 50] Weekly
-
6 MARKET TECHNICIAN Issue 38 July 2000
The JPY/IDR cross (number of IDR per Yen) weekly chart
belowshows a relatively flat trend over the last 16 months, with
the cross inthe 13/50 week moving averages suggesting that the Yen
could beset to gain against the IDR over the coming weeks towards
8000.
The chart of the USD/KRW below shows how the Korean Won wassold
off from 850 towards 1920 /1980 over a few months. Whilethe 1997
crisis was seen as a Southeast Asia problem to beginwith, the
exposure to Thai debt and short term USD exposure reallyhit the KRW
at the tail end of 1997. Technically the KRW formed alarge reversal
in the 1st quarter of 1998 and has been slowly firmingsince then.
The 13/50 week moving averages show the KRW tryingto put together a
move through 1100 for 1000; this should takesome time. When seen,
though, 1060/20 will be open.
The YEN/KRW (Korean Won per Yen) chart shows a very flat
trendwhich is expected to continue to dominate trading over the
comingmonths. The 13/50 week moving averages are flat while the
chartsshow a sidways channel dominating. This does present a
decentcarry trade given the KRW yields near USD levels (or above)
andthe very low Yen rates.
The USD/MYR (Malaysian Ringitt) chart is a favourite for
chartiststhese days. The MYR had the same breakout pattern as the
rest ofthe SE Asian currencies in 1997 but the fixing at 3.8000 and
theintroduction of controls prevented the MYR from firming with
therest over the last 18 months. Market rumours continue to roil
aboutthe fixed FX regime ending but as usual the market will need
someclear signals from the central bank to really get excited.
Looking at regional currencies closer to home, some key points
needto be made. The first and probably most important one is that
manyof the eastern European countries will be joining the EU in
thecoming 10 years and hence their currencies will be euro
candidates.While this is a significant amount of time, the FX
market is notoriousfor trading ahead of itself (or is it very
efficient at discountinginformation?). The second is that liquidity
in the FX market is nothigh in these currencies, especially outside
of London hours. Andthirdly, while derivatives are available in
some of these markets, theliquidity issue is a key one and
bid/offer spreads can take much ofthe incentive out of dealing
here.
The EUR/CZK chart below shows a head and shoulders top
pattern
IDR=. Bid [Hi/Lo/Cl Bar] [MA 13][MA 50] Weekly JPYKRW=R. Bid
[Hi/Lo/Cl Bar] [MA 13][MA 50] Weekly
MYR=. Bid [Hi/Lo/Cl Bar] [MA 13][MA 50] Weekly
EURCZK=. Bid [Hi/Lo/Cl Bar] [MA 20][MA60] Daily
JPYIDR=R. Bid [Hi/Lo/Cl Bar] [MA 13][MA 50] Weekly
KRW=. Bid [Hi/Lo/Cl Bar] [MA 13][MA 50] Weekly
-
Issue 38 July 2000 MARKET TECHNICIAN
that dominated trading in early 1999. While the central bank
wasnot happy with the CZK firming and many economists and
localmarket players were convinced that the CZK had to weaken,
thechart told a different story. The current push through
trendlineresistance is central bank inspired (heavy intervention)
but with therisk zone at 37.20 and expected to be difficult to
break, thisshould be another CZK buying opportunity. Technically
flattrading in a large band is expected to dominate activity here
forthe coming months with the risk being that the downtrend hasbeen
broken.
The Slovakian koruna has rallied strongly over the last year
againstthe euro but is now losing some ground. For choice the 42.00
areawill be able to hold off the euro gains, but, if broken, then
the turnabove the 60 day moving average would suggest that the SKK
couldback up further. Looking out though and given the decent
carryover the euro, the SKK would be seen to be a buy for further
flattrading as the top of the range should be the old head and
shouldersneckline near 44.30/50 while 41.60 supports. A bout of
SKKstrength through the latter would put 40.80/40.60 into the
sights,continuing the firming trend seen over the last year.
The chart below is of the Polish Zloty, now one of the most
liquidEastern European FX rates. The last two years have seen this
marketmature from where one would have to trade the Tbills to gain
PLNexposure, to a free floater as of yesterday! On the chart one
can seethat the PLN took the flotation news pretty well and firmed
back tothe 4.06 area against the USD. There is a floor at 4.000
though andfurther choppy trading seems likely. The local market
tends to focuson the Nasdaq as a barometer of sentiment (foreign
directinvestment into Poland is seen as one of the big reasons for
any PLNstrength) which makes it worthwhile keeping an eye on that
market.The risk would be that this is a bull flag and sustained
gains through4.20/25 would worry for 4.40/4.50 and worse.
The South African Rand has been in a breakout since the early
partof the year, and even though I kept looking for a turning point
or a
top, there has been little joy here yet. Hit by very negative
sentimentthis one could be on course for 6.85 to break with
technical targetspointing to 7.00 and 7.50. The key has been the
fact that the ZAR isnot paying for risk, interest rates have only
edged higher since the6.20 level broke and the market seems very
reluctant to take the riskon board without getting paid for it!
The Icelandic Krona is a basket currency which is tied mostly to
theeuro and euro-linked currencies as well as to the GBP and USDand
other currencies*. This FX rate offers one of the moreinteresting
regional or exotic FX plays as Iceland, population250,000 on a good
day it seems, pays over 10% on the Kronacompared with rates below
4.0% on the euro. The chart below isof the ISK/USD and for those
who wish to bottom fish for a longterm EUR/USD recovery, the ISK is
a different way to play this asthe positive yield differential over
the USD gives one some cover.Technically this is a broad range in
play but 73.80/74 needs tohold as resistance, since a turn in the
trend will be signalled by abreakthrough 73.00. The warning on
liquidity definitely applies tothe ISK as the market is usually for
5 million USD or less, andoptions are very difficult as the
bid/offer spreads can reach truckdriving levels.
The chart overleaf shows EUR/ISK which is pretty much locked in
abear trend. The key here is the 70 zone which is likely to be
astruggle, and I could see flat trading dominate in a 73/70
range,yield differential makes playing the range from the buy ISK
above72.30, sell below 70.80 angle attractive.
7
EURCZK=. Close(Bid) [Line] [MA 60][MA20] Daily
PLN=. Bid [Hi/Lo/Cl Bar] [MA 20][MA60] Daily
ZAR=. Bid [Hi/Lo/Cl] [MA 13][MA50] Weekly
ISK=. Close(Bid) [Line] [MA 60][MA20] Daily
*The ISK (Icelandic Krona) trades as a currency basket with the
majority of theweighting placed on the European rates. The major
short term risk on the currency isstill an overheating economy with
rising inflation and inflation expectations, whichsuggests that
interest rates should remain on the high side. One month money
isnear 10.00% and three month money is near 10.00% as well. The
breakdown of thebasket is:
EUR 32.14% GBP 13.39% DKK 9.16% NOK 8.26% JPY 5.39%USD 24.09%
SEK 3.85% CAD 1.37% CHF 2.35%
-
MARKET TECHNICIAN Issue 38 July 2000
The Greek Drachma is now trading with eyes on the 340.70
entryrate (March 2001) and is a trading buy for the yield.
Basically,most of the rate cuts are expected to take place closer
to the EMUentry date (like the Irish did in 1998) which should
leave theEUR/GRD to give carry opportunities up until then.
The Turkish Lira chart is a pretty easy one for trend followers
topredict! The controlled devaluation with 80% and much higher
yieldsdid provide decent returns for some time, but those days are
over.With the new IMF backed inflation fighting regime, yields
havedropped to the 25%/40% zone in the 3 month money area and theFX
is expected to devalue nearly 20% on the year, breakeven if youare
lucky! Given our criteria of risk vs. reward (need to be paid
forthe risk) the TRL is now a much more selective play (wait for
yieldsto back up on any political concerns and then jump in).
And finally, Latin America. The Mexican Peso is the most liquid
of theLATAM currencies and even trades as a reasonably deep
contract onthe IMM. The close ties with the US makes this one
attractive as ahedging or speculative currency and the high rates
often entice hotmoney into it as well. The 1994/1995 Tequila crisis
is seen prettyclearly below as traders went home for the holidays
in 1994 only tofind out that the MXN had taken a swan dive in their
absence! Thegeneral trend was for a weak MXN over the last few
years with carrytrades mostly working out depending on the entry
point (yields wellover 20% compensating for the risk). While the
Asia crisis of
1997/1998 ran its course with little in the way of sustained
pressureon the MXN, the Russian/Brazil effect of August 1998 really
hit theMXN. Given the liquidity of the MXN compared to the
Brazilean Real,the USD shifted higher in a very aggressive manner
and even touchedthe 11.10/20 zone after Brazil devalued in early
1999. Since then themarket has seen the MXN firm sharply and a case
for 9.20/9.85 as arange trade can be made. While the recent push
towards 9.20/9.15could have threatened to see 8.70 and lower for
the USD, the factthat yields in the short term instruments dipped
to 13% and lowertook away the risk premium. Holding to the view for
the rangetrading , then buying MXN below 9.20 is not that favoured
and infact a turn back to 9.60/9.65 is needed before the MXN
becomes abuy again. A sustained USD rally through 9.85 would shift
the riskhere and put 10.20/10.50 into the light, not really
favoured.
The chart of the Brazilean Real below shows the devaluation, as
onceagain a market went from We will never devalue to Whoops!.The
toppy pattern after the devaluation could be a head andshoulders in
the USD/BRL with subsequent activity being very flat.While moving
averages are convergeing and there is a risk of a shiftback to
1.90, this one seems more likely to trade in a flat mannerbetween
1.86/1.76 for the coming months. The 17%+ yields help onthe carry
trade for the BRL but keep in mind that activity is mostlydone
through the non-deliverable forward market (NDF, with noexchange of
the local currency taking place, only USDs going backand forth for
the difference in FX between entry and exit levels).Thus, in a
crisis, the exit door in the BRL gets very, very small.
SUMMARY
In summary while the potential carry gains are easy to
calculateand the potential positive movement in the regional FX
market canbe very attractive the critical factor to keep in mind is
that the highyield on the currency is probably there for a reason.
This willusually make the potential losses on adverse movement in
the FXrates extremely high as the exit door in these markets is
invariablymuch narrower than the entry door. The question that must
beasked on every trade is Does the potential reward outweigh
thepotential risk?. Charts give a very useful indication as to what
themarket is doing and can give valuable insights for
forecastingpurposes. While data is at times not that clean, this is
improving andthe charting of regional FX markets will continue to
grow inimportance for hedging and speculative purposes.
8
EURGRO=. Bid [Hi/Lo/Cl Bar] [MA 60][MA20] Daily
TRL=. Bid [Hi/Lo/Cl Bar] [MA 60][MA20] Daily
MXN=. Bid [Hi/Lo/Cl Bar] Weekly
BRL=. Bid [Hi/Lo/Cl Bar][MA 60][MA20] Daily
EURISK=. Bid [Hi/Lo/Cl Bar] [MA 60][MA20] Daily
-
Issue 38 July 2000 MARKET TECHNICIAN 9
CONCLUSIONS:
1. The new economy TMT group of stocks has recently ralliedfrom
deeply oversold positions. These initial reflex rebounds arelikely
to be followed by some back testing. That should lead tomore
extensive base-building, which, in turn, should lead into anew and
more sustainable advance later this year.
2. Undervalued old economy groups rallied during the March /
MayTMT sell-off. But progress is likely to slow for a while as
thesesectors do battle with significant overhead resistance
levels.
3. Ultimately, its possible that the two sides manage a degree
ofreconciliation. The higher quality TMT build new bases
(althoughthey may well not surpass Q1 peaks for some considerable
time)whilst a number of cyclical / value plays regain their share
of thelimelight. The issue is probably not so much new economy
versusold economy as, quite simply, a changing, evolving,
economy.
4. Uncertainty and market volatility could remain a feature a
littlelonger. But, looking ahead to later this year, markets could
bepreparing for their next advances. Medium and long term
rate-of-change momentum measures look in place for this, at least
forthe U.S. and the UK. These cycle indicators seem to be buyingthe
soft landing scenario for the Anglo Saxon economies.
CHART 1:The NASDAQs March May decline broadly speaking
foundsupport in the 50% retracement region of the October 1998
/March 2000 advance. Medium term momentum (e.g. 13 weekRoC) reached
oversold extremes not even seen in October 1998, andis starting to
turn up.
Chart 1: NASDAQ, Fibonacci Retracements, 13 Week RoC
CHART 2:But a number of technical positives are still missing,
e.g. sentiment isnot yet really bombed out. And long term momentum,
shown inchart 2, suggests that at this stage rallies largely still
provide tradingbuys. A more secure base can be seen in the 2800
area, where a61.8% retracement of an 18-month trend (see chart 1)
more or lesscoincides with the 50% level in relation to the entire
1995 2000bull market. Chart support in this region is a bonus.
Chart 2: NASDAQ, Long Term Chart; Long Term Momentum
CHART 3: Dow Industrials. The medium term momentum buy signal
shouldreinforce support levels. At the very least, it should limit
downside,particularly since the long term version has now fallen
back tobelow the zero line its first visit to that area since Q3
1998. Thatkind of combination is not saying that the index starts
to trend upfrom this point forward, but it is likely to be setting
the stage for amajor upturn a little later in the year. Price
activity over the last 12months looks to be shaping up into a
diamond: this may be acontinuation pattern. Some people see a head
and shoulders:volume would argue against that.
Chart 3: Dow Industrials, Weekly Data Momentum; Volume
CHART 4: FTSE looks broadly similar. Medium term momentum is
bottoming;the long term version is already giving a buy i.e. it is
a littleahead of the Dow. The 6400 / 6000 area has recently
re-establisheditself as support.
Chart 4: FTSE 100
CHART 5:A founder member of the TMT group, the Software &
ComputerServices sector. Sharp falls between March and May have
foundsupport in the 50% retracement region (in relation to the Q3
1999 Q1 2000 relative uptrend). But base-building, leading to a
sustainablenew uptrend, is still going to take some time.
Chart 5: Software & Computer Services Relative to the All
Share
Equity Markets: Outlook for Wall Street andFTSE, TMT and Old
Economy
By Luise Kliem, MSTAThis article is a brief summary of the talk
given to the Society on 10th May 2000
Source: Bridge Information Systems
Source: Bridge Information Systems
Source: Bridge Information Systems
Source: Bridge Information Systems
Source: Datastream
continues on page 11
-
10 MARKET TECHNICIAN Issue 38 July 2000
INTRODUCTION
Under the normal course of events an analyst examining
thehistorical record of the behaviour of an exchange traded
futurescontract for any particular delivery month would reasonably
assumethat the numerical data on prices, volume and open
interestrepresented the outcome of the trading activity for that
contract.Similarly, it would be assumed that the transformation of
these datainto a graphical form would provide a meaningful
visualrepresentation of that behaviour from which various
relationshipsmight be deduced using different analytical
techniques. However, it islikely that this no longer is the case
with certain futures contracts.
ENTER MOS
The reason is the existence of the Mutual Offset System or
MOSwhich is provided by the Chicago Mercantile Exchange (CME)
inconjunction with the Singapore International Monetary
Exchange(SIMEX) for clearing members on both exchanges. What is
MOS? Itis a facility to allow new or liquidating trades to be
executed onSIMEX by CME members and then be transferred back to the
CME.The opposite is also catered for e.g. SIMEX firms can have
tradesexecuted on the CME and then be transferred back to SIMEX.
Thefacility is currently available for three interest rate futures
contractsi.e. 3 month Eurodollar, Euroyen, Euroyen Libor, together
with theJapanese Government Bond. There is also a MOS arrangement
,termed a Link , between SIMEX and the International
PetroleumExchange (IPE), London for Brent Crude Oil futures.
Theimplications of such MOS for technical analysis is discussed
here inthe context of the CME-SIMEX system.
MOS allows a participant to effectively borrow the liquidity
inSIMEX when the CME is closed and vice versa, and have
theconvenience of only having to post margin for one exchange i.e.
their own local one. In addition, a participant whose localexchange
was, say, the CME, could establish a position in SIMEXthen have two
choices open to them:
(1) have the position brought back to the CME at the price
theirorder got filled on SIMEX, to be processed through the
CMEclearing house [as if the order had been filled in the CME
pit],thus being marked to the market against the CME
marketsettlement price.
(2) leaving the trade with SIMEX and having it processed
throughthe clearing house system there, thus being marked to
themarket against the SIMEX settlement price.
Given such convenience, MOS must undoubtedly be a
welcomedevelopment from a hedgers viewpoint. However, a personal
view isthat what is generally good news for the hedging community
usuallyspells bad news for the speculating community. MOS appears
to beno exception.
SOME INFERRED DYNAMICS
However, before spelling out the nature of the bad news it
isnecessary to explore the consequences of participants
takingadvantage of the MOS facility, in terms of the impact on the
internaldynamics of the auctions at the Chicago and Singapore ends
of thesystem. If, say, a Chicago based participant has an order
executed onSIMEX, which represents the initiation of a new
position, then thearrival of that order in the Singapore pit will
impact on the ongoingauction, by influencing a shift in the price,
in some way dependingon the liquidity conditions prevailing at the
time of its arrival inrelation to the size of the order. For the
order to be filled, some otherparticipant, be it a floor broker or
a local will have to take the other
side of the trade. Consequently the creation of the transaction
will berecorded in the exchanges time and sales register for that
market.Thus the price level and number of contracts involved will
have beencaptured and be available to be fed into the clearing
house system should the Chicago based participant decide to leave
the positionin Singapore.
What happens if (s)he so elects? The answer is that a footprint
ofthat trade, in the form of a data point, will subsequently appear
inthe tick price data record for that day. If the exchange is in
the habitof issuing volume at tick price then the volume associated
with thetrade will also appear there. Secondly, if the position is
held after theClose it will be recognised as new commitment and
factored into theclearing house calculation of the open interest
figure for that day. If itwas a relatively large order in relation
to the typical orders of the day,it could well end up helping to
generate an increase in the openinterest figure, as it reflects the
initiation of a new position(s). So farwhat has been described is
the normal i.e. non-MOS process,which would occur irrespective of
the geographical location of theparticipant.
However, what if our Chicago participant elects to have the
positiontransferred back to Chicago? It will be marked to the
market usingthe CME settlement price. As such it will end up being
factored intothe clearing house calculations that produce the open
interest figurefor that day. The interesting question then arises.
What happens tothe transactional volume figure associated with the
trade? This hasalready been captured by the SIMEX time and sales
computers aswas the price tick(s) involved. As such they are a
unique constituentelement(s) of the historical record of the
auction microstructure atSIMEX during that trading session in which
our Chicago participanttook part. As such these data can hardly be
transferred to CMEalongside the fill price(s) of the local trade(s)
it would make amockery of the time and sales data for that days
auction on theCME, because the trade(s) never took place there! The
answer is thatthe transactional volume and price ticks remain with
SIMEX, but thetrade makes no contribution to its clearing house
calculations of thenew open interest figure for the day in
question, for that SIMEXcontract.
THE BAD NEWS
How these data, relating to the trades transferred from
theexchange where the position was initiated to the exchange
wherethe participant wishes to have the trade marked to the market
andthus carried until it is either offset or delivery is
made/taken, arehandled, is a key issue. This is because under the
normal course ofevents the price tick record, which goes to make up
the summaryor short-hand statement of the familiar Open, High, Low,
Closeprice bar, together with the days transactional volume and
thechange in open interest represents the outcome of the how of
theway that the various orders were processed in the pit, in terms
oftheir size and sequence of execution.
As such the record encapsulates the cause and effect
relationshipbetween the three variables of price, volume and open
interest.Knowing this, it is normally quite realistic to undertake
an analysis ofthese data and have a legitimate expectation that
whatever cause-effect relationship might be discovered for that day
[possibly inconjunction with a number of previous days data] it is
valid, and canthus legitimately be allowed to influence an analyst
or traderssubsequent view of the likely future behaviour of the
market ofinterest.
If, however, as appears to be the case with a market having a
MOSfacility, where it seems impossible to maintain the integrity of
thelegitimate tick price/volume/open interest relationship that
reflects
MUTUAL OFFSETTING SYSTEMS A PROBLEM FOR TECHNICAL ANALYSIS?
By Dr Michael Wignall MSTA
-
Issue 38 July 2000 MARKET TECHNICIAN 11
the auction process, the results of any analysis will likely be
faulty.This is because the cause-effect relationship between the
threevariables will have been corrupted by the MOS trades which
theirowners decided to transfer back to their local exchange.
IS WHAT YOU SEE THE WAY IT WAS?
Before the advent of MOS the answer to this question for the
pricechart of any market of interest would simply have been
Yes.However, with the existence of MOS, the answer must now
beMaybe. It will depend on whether the contract under
scrutinytrades on more than one exchange and whether it has been
blessed[cursed?] with a MOS facility or some equivalent. If it has,
then theanswer is Unlikely and the analyst and trading system
developerwill be inviting trouble if they ignore the implications.
Ultimately itwill come down to the volume of MOS transactions
involvinginter-exchange transfers in relation to the volume of
MOStransactions not involving inter-exchange transfers and the
volumeof non-MOS transactions for the contract of interest. The
greaterthe proportion of MOS transferred transactions in any one
contract,over time, the greater the likely distortion of the
historical record ofthe dynamic relationship between intra-day
price changes,transactional volume and inter-day open interest
changes for thecontract in question.
SOME CONSEQUENCES
As pressures for 24 hour hedging facilities grow to recognise
existing24 hour global trade flows, it is likely that mutual
offsetting willbecome an increasingly common feature of contracts
as commodityexchanges around the world set up partnerships based on
jointinterest. Why accept the costs of fighting for market share
for a newfinancial product, or retain a share for an existing
product, when therevenue from its usage in the form of exchange
fees can literally beshared between exchanges co-operating, rather
than competing,across time zones?
Assuming the interpretation of the implications of the mutual
offsetprocess is correct, then the likely consequence is that
conventionaltechnical indicators which factor volume and open
interest intoprice changes can be expected to demonstrate erratic
behaviour insome markets, but not so in others. Whereas in the
MOS-free worldof the past such indicators would have been expected
to giverelatively uniform results across any selection of markets,
thiscomfortable simplicity has likely passed away. In the case of
futures,for example, erratic behaviour can be expected from the
operationof the Herrick Payoff Index, and were MOS to be
eventuallyextended to cater for options on futures, then the Hines
Indexwould be so affected. Thus charts, whether they be hard copy
orscreen-based might one day, when MOS markets become
moreprevalent, display a disclaimer by the more perceptive and
riskconscious data vendors:
This market has a mutual offset facility with another
exchangewhich may consequently render unreliable any analytical
methodswhich factor volume and open interest into the price
history.Analysts making recommendations or traders basing decisions
onsuch methods do so entirely at their own risk.
CONCLUSIONS
In other words no assurance can any longer be implied that what
isbeing offered via a chart is how it actually was. Unfortunately a
MOSsuch as the CME-SIMEX, has the ability to destroy the
hithertoinseparable linkage between tick price, volume at tick
price, dailyvolume and changes in open interest. Perhaps the most
prudentapproach would be for analysts and traders with a
speculatorsmindset to avoid futures markets which have MOS assuming
theytake the trouble to discover which do.
Computer based trading system developers who typically
viewmarkets simply as collections of time series data needing to
bemanipulated, and who lack any real understanding, or even
interest inthe how or the why the numbers came into existence in
the form theydid, are particularly vulnerable to the trap offered
by such markets.
Of course speculator participants who remain ignorant of
theproblem will continue to unwittingly provide the liquidity that
thehedging community needs. As ever, hedgers bliss can be defined
interms of speculators ignorance, and MOS represents just one
morepotential facet of the latter attribute. Speculators wondering
why theactual future, being experienced through the constantly
unfoldingpresent, does not materialize in the form that their
definition of thefuture would have them believe it should, may
unknowingly betrading a market with a Mutual Offset System. One in
whichsignificant inter-exchange transfers are taking place.
REFERENCES
Chicago Mercantile Exchange, CME/SIMEX Mutual Offset System;The
Worlds Most Successful Trading Link, CME, n.d., Chicago,
IL.,http://www.cme.com/market/interest/ mos.html
Chicago Mercantile Exchange, Euroyen Futures At The CME,
CME,n.d., Chicago, IL., http://www.cme.com/ market/euroyen.html
Chicago Mercantile Exchange, CME Contract Specifications
ForInterest Rate Futures And Options, CME, Jan. 06, 2000,
Chicago,IL., http://www.cme.com/clearing/spex/
csinterestrate.html
Chicago Mercantile Exchange, Interest Rate Marketing Dept.
toWignall, Jan. 24, 2000, (e-mail)
Hines, Ray, Hines Index, Technical Analysis Of Stocks
&Commodities, Vol. 7, No. 4, (April) 1989.
Neil, Trevor, The Herrick Payoff Index: making use of
futurescontracts open interest, Market Technician, Issue No. 25,
(March)1996.
Wignall to CME Currency & Interest Rate Marketing Dept.,
Jan. 24,2000, (e-mail)
Authors postscript: SIMEX has since merged with the
SingaporeStock Exchange to form what is known simply as the
SingaporeExchange, or SAX. The MOS facility remains in being.
Copyright Michael Wignall 2000
The writer (who, contrary to conventional wisdom, is seeking
todevelop the Holy Grail of computer based trading systems)welcomes
comments on this article, via the journals Letters section,or tel.
01923 450681, or email [email protected]
continued from page 9
CHART 6: Old economy sector Restaurants, Pubs & Breweries.
Spikedowntrend reversals by these groups have been a
commonoccurrence in the market this year. They look to be
bottoming, butthey have a lot of overhead resistance areas to deal
with, both inabsolute and relative terms.
Chart 6: Restaurants, Pubs & Breweries, Relative to the All
Share
Reprinted by permission. Luise Kliem, Analyst, Merrill
LynchCopyright 2000 Merrill Lynch, Pierce Fenner & Smith
Incorporated
Source: Datastream
-
12 MARKET TECHNICIAN Issue 38 July 2000
INTRODUCTION
Financial markets like futures and commodity markets
produceenormous amounts of data for analysis. It is much quicker
and easierto analyse a graph than it is to analyse copious amounts
of tabulardata. So it is not surprising that many people choose to
display datain a graphical format. How this data is actually
represented on achart, and what information is highlighted, depends
on the nature ofthe plot type.
Different plot types make certain information more obvious
thanothers. Bar charts, line charts, point and figure charts,
candlestickcharts and CBOT Market Profile are all different plot
types.ChronoGraphicsSM is a new plot type and, as such, should be
thoughtof in the same terms.
However ChronoGraphics offers very easy access to
moreinformation than that offered by other plot methods. It tells
you:
what happened first in a period the high or the low? how fast
the move was between the high and the low was it
quick or slow? when during the period the high or low happened-
early or late?
What this means in terms of added analysis capabilities we
shallcover later. First, let us examine the way a ChronoGraphics
bar isdrawn. The easiest way to do this is to examine the way that
atraditional bar is constructed and then to compare this with the
waythat a ChronoGraphics bar is constructed.
Construction of Traditional bars andChronoGraphics bars
A traditional bar is constructed by finding the highest high
tradedduring the trading period under consideration and then the
lowestlow. The difference between the two gives the periods trading
range;and it is represented by a single vertical bar, drawn from
the low tothe high. The bar also includes a mark, showing the close
(usually thelast trading price) of the period. When plotting a
marketsperformance over more than one time period, each bar is
separatedfrom the neighbouring bars by a constant distance. The
constructionof a weekly bar, which embraces five trading days, is
shown in Figure1.0 below.
Fig 1.0With ChronoGraphics, however, the bar is drawn from the
actual lowto the actual high. This reflects the time element. See
Fig 1.1opposite.
Fig 1.1If the price action for the week had been different, but
the high, lowand close had been the same, the traditional weekly
bar would havebeen as in Fig 1.2 opposite.
Fig 1.2However using the ChronoGraphics methodology there would
havebeen a clear indication that the action during the period had
beenvery different, as in Fig 1.3.
Fig 1.3So in Figure 1.1 we have a forward, or positive, sloping
bar and inFigure 1.3,we have a negative, or backward, sloping bar.
The slopeof the bars shows us the exact sequence of events during
the period.It represents the directional characteristic of a bar.
We can also tellfrom the angle of this slope whether the move
occurred quickly or ifit took most of the period.
We have two more bits of information to add to complete
theChronoGraphics bar. These are the Open and the Close. On
aChronoGraphics bar, the colour between the Open and the Close
isshown in a different colour from that which is otherwise
usedbetween the High and the Low extremes. See Figure 2.
ChronoGraphics SM
By Nick Burnton
Fig 1.0
Fig 1.1
Fig 1.2
Fig 1.3
5 Days
High
Low
X axis Time
Y a
xis
Pr
ice High
Close
Standard WeeklyBar Chart
Low
5 Days
5 Days
5 Days
High
High
High
Low
LowLow
X axis Time
X axis Time
X axis Time
Y a
xis
Pr
ice
Y a
xis
Pr
ice
Y a
xis
Pr
ice
High
Close
Standard WeeklyBar Chart
Low
High
Close
Standard WeeklyBar Chart
Low
High
Close
Standard WeeklyBar Chart
Low
ChronoGraphicsBAR
ChronoGraphicsBAR
-
Issue 38 July 2000 MARKET TECHNICIAN 13
Fig 2By adding in this information we are now also able to
define eachChronoGraphics Bar in terms of its behavioural
characteristics: This isbecause the Close may be either above or
below the Open.
If a Close is above the Open in a positive sloping
ChronoGraphics bar(Fig. 2.1 and Fig. 2.2), or if the Close is below
the Open in anegative sloping ChronoGraphics bar (Fig. 2.5 and Fig.
2.6), thenthat bar is deemed to be normal.
If, however, the Close falls back below the Open in a positive
slopingbar (Fig.2.3 and Fig. 2.4), or if the Close rises back above
the Openin a negative sloping bar (Fig.2.7 and Fig. 2.8), then that
bar isdeemed to be abnormal.
Note that for an abnormal bar, the colours of the ChronoGraphic
barinvert.
The importance of an abnormal ChronoGraphics bar
(whetherpositive or negative sloped) is that it is an item of
additionalinformation, which has trading implications. Evidence
suggests thatonly about 10% of ChronoGraphics bars are
abnormal.
ChronoGraphics bars can therefore be defined by their
directionaland behavioural characteristics.
Applications of ChronoGraphics
The extra information that is supplied by ChronoGraphics bars
can beused in a number of different areas of Technical Analysis.
Chapters ofa book could be devoted to some of these areas. However,
here is avery brief overview:
Multiple time framesMultiple time frames have been difficult to
incorporate on one chartusing traditional bars. Quite simply, they
become too difficult toread. ChronoGraphics allows the overlay of a
number of periods,with no confusion, and with clear and easy to
read trend lines.
Trend linesWhen using traditional bars, trend lines drawn using
different timeframes tend to give different results. ChronoGraphics
integratesdifferent time frames on one chart, and thereby gives
consistencyof trend line signals. Also ChronoGraphics enables trend
lines to bedrawn that, although relevant to trading decisions,
would not havebeen apparent using traditional methodology.
Chart patternsWith each bar now being defined in more detail,
classic chartpatterns can be seen and analysed in more detail.
Also
ChronoGraphics has its own unique patterns given by the
variousangles, colours and lengths of the bars.
System testing One of the drawbacks of traditional system
testing is that thealgorithms in software assume that the price
extreme closest to theOpen was the one that traded first. This is
not always necessarilytrue and can give false system performance
results. ChronoGraphicsshows the chronology of High, Low, Open and
Close.
FormulaeGenerally speaking, current formulae used in TA are
restricted intheir ability to adapt to market conditions due to the
one-dimensional time element. However with the added
informationfrom the slope of the bar (in terms of direction and
angle), as wellas the time element of the extremes, the potential
for derivingnew and better formulae has increased
significantly.
StrategyChronoGraphics can be used to focus attention on
opportunities inthe market such as highlighting abnormal market
conditions. Thiscan be combined with other forms of analysis to
formulate astrategy for trading.
CONCLUSION
ChronoGraphics bars are a new way of presenting price
information.They provide much more information per bar than was
traditionallythe case. The slope of the bar provides information
about thedirection and speed of the market movement, and the
colours of thebars provide information about the chronology in
between the openand the close. In principle ChronoGraphics
integrates the informationfrom traditional bars and candlestick
bars, and goes beyond both.
Probably the most exciting aspect of using ChronoGraphics is
thatthey provide new building blocks for research to be carried
out.ChronoGraphics will be available from Bridge Information
Systems onBridgeStation as part of their charting software packages
of Athenaand AthenaExpert as from the middle of July.
Chronographics willalso be available on Aspen Graphics for
BridgeFeed. Bridge has aPatent pending on ChronoGraphics.
Nick Burnton is the European Director of Technical Analysis at
BridgeInformation Systems.
Copyright Nicholas Burnton 2000
Fig 2
Fig 2
Fig 2.1 Fig 2.2Positive Normal
Fig 2.3 Fig 2.4Positive Abnormal
Fig 2.5 Fig 2.6Negative Normal
Fig 2.7 Fig 2.8Negative Abnormal
Open Open
Low
High
CloseClose
Bytes & PiecesThe Global Exchange Database from
Grade Publishing
Finding a reliable source of data is never easy, so you may
wishto check out the Global Exchange database. As the namesuggests
it covers all the worlds major exchanges and have anextensive
database of stock, commodity and derivativeinstruments. The data is
updated daily via the Web or can beFTPd to your own network, with
an extensive database, manyTechnicians may find this an invaluable
resource.
For details see Grade Publishing Ltd: Tel: 020 8672 6655
* * *
Need another Brain?Artificial Intelligence can help!
The use of artificial intelligence modelling technologies
continuesto become more accessible with the release of
TradingSolutionsa technical modelling programme by
NeuroDimensions.TradingSolutions combines traditional indicators
with neuralnetworks and genetic algorithms, with step by step
wizards toallow you to chart and model times series data. It also
has theability to read standard format ASCII or Metastock data.
This isthe most user friendly program of this type I have seen,
withguided instruction and a step by step tutorial.
Contact NeuroDimensions: Tel: 001 352 377 5144
-
14 MARKET TECHNICIAN Issue 38 July 2000
INTRODUCTION
In early 1999 I had a paper in the March issue of the
MarketTechnician entitled, Using Non-linear neurogenetic models
withprofit related objective functions to trade the US T-Bond
future. Thepaper described a neural network based bond futures
trading system,the performance of which I later presented at a talk
I gave to the STAin February 2000.
The paper was originally written for an academic conference and,
assuch, was not particularly trader friendly. Given that readers of
theMT tend to be practitioners rather than academics, in this
article Iexplain how I built the system from a traders perspective
and, moreimportantly, show how the system has performed over the
last 2years, during which it has been running in real-time.
It is an EOD system and is 100% mechanical with no
over-rides.The system is in fact fairly simple even though the
methods used tobuild it may appear complex if one has little
experience in usingneural networks. It uses five inputs in all; two
formed from Bondfutures prices and three from the S&P500
futures two of whichare merely lagged values of the one distinct
S&P input see figure1. A genetic training algorithm was used as
this allows one greaterflexibility with regards to the choice of
objective function (i.e., MSE,Total Profit etc.) than the more
commonly used gradient-basedalgorithms.
METHODOLOGY
The steps taken to build the system were as follows:
1) Using historical closing data derived from the 30 year T-Bond
andS&P500 futures contracts, spanning from the 12th April 1983
tothe 10th Oct 1994 (approx. 3000 days of in-sample data),design
inputs/indicators which show correlation to future T-bondfutures
price returns.
2) Using the derived indicators as inputs, test a number of
neuralnetwork architectures and optimisation functions. Choose
thearchitecture with the least free parameters and the most
stableand profitable performance over the full in-sample
dataset.Figure 1 shows the resulting optimal neural
networkarchitecture.
3) Train 20 separate networks with the specified architecture
toconvergence by maximising total log returns.
4) Use the outputs of the resulting networks in conjunction with
amajority trading rule to create daily buy and sell decisions.
5) Run networks on out of sample data from 10th Oct 1994 to
the30th June 1998 and examine performance (approx. 900 days outof
sample data).
No stops were used for the system after a number of tests
showedthem to be detrimental to overall performance. My research
hasshown that this is generally the case with trading systems
designedusing this methodology a result somewhat at odds with
thecommonly held notion that stops are de rigueur. It may be
thatincluding stops in the initial optimisation process would yield
differentresults as opposed to adding them ex post facto. In actual
trading theidea is to have enough capital in the trading account to
deal withdrawdowns and to use a peace of mind stop placed at a
distancesuch that it will only be hit in extreme trading
conditions. A standardmonte carlo simulation was used to analyse
drawdowncharacteristics.
Figure 1. Neural Network architecture used for final system.
Figure 2. In-sample and out of sample equity curve.
Figure 2 shows the cumulative equity curve over the whole in and
outof sample periods. It can be seen from the chart that the
performanceis relatively stable and exhibits a steadily upward
sloping equitycurve. The figure for slippage and commission was
derived fromexperience trading with real prices and amounted to
average slippageper round turn (RT) of 1 tick ($31.25) and
commission of $20 i.e., atotal of US $51.25 per RT. The slippage of
1 tick per RT is notsurprising given that when using the open(t+1)
to open(t+2) to tradethere is no significant difference to the
results. Table 1 shows aperformance summary from the same period
(the results have beenamalgamated over both in and out of sample
periods for the sake ofbrevity all figures include slippage &
commission).
Table1. In-sample and out of sample performance summary.
Trading a 2 year old the real-time performanceof a neurogenetic
T-bond futures trading system.
By Zac Harland www.kruegerresearch.com
Total net profit $202,183 Ratio avg win/avg loss 1.24
Total no. of trades 710 Avg trade $284
% profitable 57% Max consec. winners 9
No. winning trades 410 Max consec. losers 7
No. losing trades 300 Avg bars in winners 6
Largest winning trade $9,698 Avg bars in losers 5
Largest losing trade -$4,238 Max drawdown -$10,389
Average winning trade $1,199 Profit factor 1.69
Average losing trade -$965 Sharpe Ratio 1.3
-
Issue 38 July 2000 MARKET TECHNICIAN 15
Real time Results
The system has been operating in real-time since the 30th
June1998. Its cumulative equity curve can be seen in figure 3,
along withsome standard measures of performance in table 2. All
figuresinclude slippage & commission as referred to above.
Figure 3. Real-time performance.
Table 2. Real-time performance summary.
It is evident from the real-time results that the system
continues toperform within expectations. In fact it has over
performed withregards to average trade figures and so one would
expect somereversion to the mean as time passes. Figure 4 shows the
UnderwaterEquity Curve of the system during the real-time period.
Note that thechart depicts the underwater equity that would have
beenexperienced if one was trading a single contract, assuming a
$20,000account. This is somewhat artificial if ones money
managementpolicy dictates an increase in the number of contracts
traded asaccount equity rises and results in equity drawdowns, as
apercentage, decreasing unrealistically towards the right hand side
ofthe chart.
Figure 4. Underwater Equity Curve
DISCUSSION
A point worth raising is that there is nothing inherently
complexabout the way in which the neural network was utilised. It
wouldhave been perfectly possible, with the same inputs, to design
atrading system using traditional methods of technical
analysishowever, it is more than likely that this would have
resulted in lessthan optimal performance. The justification for
using a neuralnetwork is that when used correctly they are one of
the bettertools for extracting trading signals from
inputs/indicators sufferingfrom low signal to noise ratios . They
are certainly more efficaciousthan the brute force optimisers found
in many trading packagestraditionally brought to bear in trading
system design.
The networks are not designed to be re-trained, as it is
assumedthat the inputs themselves will adapt to current market
conditions.For an article elucidating the pros & cons of using
a neuralnetwork as opposed to traditional methods please visit my
websiteat www.kruegerresearch.com.
In the initial search for reliable inputs many of the more
commonlyknown indicators were tested but found wanting; they did
notexhibit sufficiently stable relationships with T-Bond returns to
beconsidered for the model. If noise inputs are used the
neuralnetwork will suffer from the garbage in, garbage out
syndromeand no amount of optimisation, connectionist or otherwise,
willcreate a workable trading model. Of course, it is not just
sufficientto find inputs that contain some measurable and usable
signal in-sample; the signal content must remain stationary in
itsrelationship to both historical and future T-Bond returns.
Designinginputs with these attributes is no trivial endeavour a
fact anyonewho builds trading models will attest to. In fact
input/indicatorconstancy is arguably the most important issue
concerningmechanical trading system design, regardless of the
methodologyused (of course money management is also very
important,though it is only applicable once ones model exhibits a
statisticaledge).
To arrive at the inputs for this system I used a number
ofcorrelation measures including mutual information and
coherenceamongst others. It is not advisable to use linear
correlationmeasures as they will not detect non-linear
regularities, if theyexist in the data. Moreover, if one is testing
for linear correlationsit is prudent to use a non-parametric
statistic as prices and pricereturns tend not to conform to most
standard statisticalassumptions i.e., normality, stationarity,
homoskedasticity etc. SeeRzempoluck (1998) for further information
software is includedfor the calculation of linear correlation,
mutual information andcoherence.
THE FUTURE
The most important consideration at this juncuture is whether
ornot the system will continue to perform in the future. Since
theoriginal system was built it has been updated with the objective
ofrendering it more robust to potential future input degradation.
Todo this, two additional systems were created, each constructed
inan identical manner but using different inputs. The outputs of
allthree systems are combined using a majority rule again to
providethe final signal. The overall performance should be improved
dueto low correlation between each individual systems trading
signals.The system is now running live on my website for those
readersinterested in viewing the signals in real-time.
REFERENCES
Harland, Z. 1999. Using nonlinear neurogenetic models with
profitrelated objective functions to trade the US Tbond future.
TheMarket Technician. March 1999.
Rzempoluck, E.J. 1998. Neural network data analysis
usingsimulnet. Springer-Verlag. New York.
Total net profit $27,024 Ratio avg win/avg loss 1.64
Total no. of trades 80 Avg trade $337
% profitable 53% Max consec. winners 4
No. winning trades 42 Max consec. losers 5
No. losing trades 38 Avg bars in winners 6
Largest winning trade $6,667 Avg bars in losers 6
Largest losing trade -$3,238 Max drawdown -$6,006
Average winning trade $1,435 Profit factor 1.81
Average losing trade -$875 Sharpe Ratio 1.7
-
16 MARKET TECHNICIAN Issue 38 July 2000
PROGRAMME
THURSDAY, OCTOBER 5, 2000
9.00-13.00 Walkabout with John Brooks
13.00-14.30 Lunch
14.30-15.15 Wieland Staud
15.15-16.00 Rudolf Wittmer
16.00-16.30 Coffee break
16.30-17.30 (with spouses) Adrienne L Toghraie: Come out on
top
18.00 Buses leave for KlusterEberbach (Wine tasting,Sightseeing,
Dinner)
FRIDAY, OCTOBER 6, 2000
09.00-09.45 Hartmut Sieper:. Three-Dimensional
TechnicalAnalysis
09.45-10.45 Erich Florek: KISS Dilemma Avoiding
KISS-philosophyNegatives and EvolvingTechnical Analysis Beyond
it.
10.45-11.30 Coffee break
11.30-12.30 Japan Hour
Takehide Matoba: A new Money ManagementTechnique Combining
andOscillator Indicator andMACD
Kouchi Suzuki:The application of theIchimoku Balance Table (no.
2)
MinoruEda:The Japanese TraditionalChart (Makino chart) and
itsModernistic Interpretation.
12.30-14.00 Lunch
14.00-14.45 Jorge Bolivar:Quantitative PatternRecognition
ForecastingMethodology.
14.45-15.30 Patrick Young
15.30-16.00 Coffee break
16.00-16.45 Joachim Goldberg:Everything Under Control?Or Bet on
the Wrong Horse?
16.45-17.30 Prof. Hank Pruden
Free eveningDinner with Ouants / Press Event
SATURDAY, OCTOBER 7, 2000
09.00-09.10 Welcome address by BrunoEstier, IFTA Chairman,
andMike Esptein, Chair of theTA/QU meeting.
09.0-10.00 David Damant
10.00-10.45 Emmanuel Acar:Inferences of Stop-loss onReturn
DistributionComparison of Various AssetClasses.
10.45-11.30 Coffee break
11.30-12.30 Ian Notley/John Brooks
12.30-14.00 Lunch
14.00-14.45 Ralph Acampora
14.45-15.15 Prof. Paul Weher, John F. Murray:Is Technical
Analysis in the FXMarket Profitable? A GeneticProgramming
Approach.
15.15-15.45 Pierre Lequeux: Profitability of TechnicalIndicators
at High Frequencies.
15.45-16.15 Coffee break
16.15-16.45 Prof. Andrew W. Lo:Foundations of TechnicalAnalysis:
ComputationalAlgorithms. StatisticalInference and
EmpiricalImplementation.
16.45-17.15 Panel discussion:Technical Analysis A New Research
Path forQuantitative and AcademicsMethods. Where Are We SoFar?
Prof. Hank Pruden, Prof Stephen Satchell, Prof.Mark Taylor, Robert
Schwob,David Darmant.
18.00 Visit to the Guttenbergmuseum (where the historyof
information began).
SCHEDULED SPEAKERS
Ralph Acampora, CMT, Managing Director,Global Equity Research,
PrudentialSecurities and founding member of IFTA.
Linda Bradford Raschke, Author of StreetSmarts expert in
short-term tradingmethodologies, etc.
David Damant, President of the EFFAS
Erich Florek, Manager and Director ofTechnical Trading, Midas
Trading House.
Joachim Goldberg, Global Head forTechnical Analysis and
BehaviouralFinance, Deutsche Bank. Author ofBehavioral Finance,
Finanz Buch Verlag.
Ian Notley and John Brooks, Yelton FiscalInc., respectively
Publisher and Directorof Sales for Institutional Informationservice
Notleys Notes, specialising onlong-term cycles in the
internationalfinancial and commodity markets.
Joerg Schreiweis, Chief of Equity Sales, DGBank. Advisory board
member andformer Chairperson of the VTAD.
Hartmut Sieper, Consultant, Author ofnumerous books and articles
onTechnical Analysis and current boardmember of the VTAD.
Wieland Staud, Managing Advisor forSinus Funds. Well-known
German ElliottWave Analyst.
Adrienne Laris Toghraie, President ofTrading on Target.
Rudolf Wittmer, President of WHS GmbH,which specialises in
system developmentand a leading expert in the field oftrading
systems.
Patrick Young, Author of Capital MarketRevolution, Prentice Hall
1999. Founderand Editor of the electronic magazine,Applied
Derivatives Trading.
Prof. Andrew W. Lo, MIT Sloan School ofManagement (via video
conference).
Prof. Paul Weller, John F. Murray,University of Iowa.
Pierre Lequeux, ABN AMRO LondonResearch.
Prof. Hank Pruden, Golden Gate University.
Prof. Stephen Satcheli, CambridgeUniversity.
Prof. Mark Taylor, Oxford University.
Robert Schwob, Style Research.
HOTEL ACCOMMODATION
Mainz Hilton (The Conference Hotel)Rheinstrasse 68Tel:
+49/6131/2450Fax: +49/6131/245589DM 280/night for double or single
room,plus DM 30 pp for breakfast.
When reserving please be sure to specifythat your reservation is
in connection withthe IFTA 2000 Conference.
IFTA Conference 2000 The Challenges ofInformation Overload in
Todays Marketplace
October 5-7, 2000, Mainz Hilton, Mainz
CONFERENCE FEES3-day conference, including all lectures,
luncheons and evening events.
Early Registration(Until July 31, 2000) Registration Fee
IFTA and VTAD DM 1.450, -- DM 1.590,--Non-IFTA members DM
1.750,-- DM 1.890,--Spouse program DM 900,--