Key Points MASSK Investment People matter, results count. Massk Investment John Bishop 70 Victoria Street SW1E London www.massk.co.uk
---------------Robot-traders allow to profit on the most attractive timespans that are unavailable for human traders. Automatic trading platforms are more cost efficient, than human traders, but it requires big upfront investment. Long term studies show both higher average returns and smaller standard deviation of algorithmic funds, especially on volatile and uncertain markets. ------------------------The report aims to introduce the topic of algorithmic based trading, which is mostly represented by High Freque ncy Trading (HFT).
Moreover, we
provide you with a comparison of the quantitative approach with qualitative approach.
Blue ocean of investing An analysis of data collected between January and March 2009 shows clearly that the most profitable and at the same
time
the least
risky
trading
opportunities occur in <10 seconds timespans. Certainly, it is much easier for a computer than for human to compete in this field.
Algorithmic Trading Economics Inhouse development of algorithmic trading technology takes about 18 months
and
costs
approximately
$728 000 for a fund that manages <$500
Figure1: Maximum gain(%) and standard deviation(p.p.) of HFT,
MM capital.i It is also possible to purchase ready-to-buy solutions from providers like Tower Research Capital or DE Shaw. It is strongly advised to create economic model of the system within a company, wherease operationals modules(e.g interface)
may be outsourced. High upfront
expenses are quickly offset by savings on wages. While it is sufficient to operate HFT fund with 5 people, an average qualitative fund employed 14 people in 2010ii. What is more, the technology has very good economics of scale. A $20 million unlevered fund with five employees needs to generate at least a 12% return per year in order to break even, compared to just
1
above 1% return of a $90 million fund. Taking
SOLD AT
FAT
1
610,000YEN (US$5,041) APIECE.
YEN INSTEAD OF
FINGER COST JAPANESE
MIZUHO
BROKER
US$225M.
better
performance Quant
iii
of
funds
(on
ON OCTOBER 2002 A BEAR STEARNS TRADER MADE A 183 FACT THAT INSTEAD OF SELLING
4
MILLION SHARES HE SOLD
4
POINTS
DOWN JONES
FALL DUE TO THE
BILLION
average +0,9p.p. a year) into account, we
can
that
conclude
investing
in
M arkets suitable for HFT
algorithmic
HFT has captured a large share
technology
of the overall U.S. and European
enablesboth
equity trading volume, and is
increasing
fast gaining popularity in other
efficiency
and
cutting costs.
regions
such
as
Asia-Pacific
(expected 10% by 2015). Market to be suitable for HFT must be both very liquid and electronic to facilitate the quick
turnover of
capital. IN 2010, HFT
Figure2: HFT as a % of Equity Turnover by Volume, U.S. and by Value, Europe 2005-1010
WAS
ESTIMATED TO HAVE ACCOUNTED FOR
56% BY
VOLUME OF THE ENTIRE EQUITY TURNOVER IN THE
U.S.,
UP FROM
21% IN
The higher the data frequency, the more arbitrage opportunities appear. Of the most liquid securities, only spot foreign exchange, equities, options, and futures markets are fully automated. With improving technology, HFT is beginning to play an important role in the trading of other asset classes as well.
2005.
Human and algorithmic based trading – comparison 83% HFT
OF PROPRIETARY FIRMS IN THE
U.S.
DISADVANTAGES OF HUMAN
Figure3 :Asset Classes Traded by Proprietary HFT Firms (% of Firms), U.S., 2009
TRADING
TRADED EQUITIES IN
2009. 1.The typographical errors It is known as fat finger syndrome of stressed traders working in fast-
Euphoria Anxiety Thrill Exciteme nt
moving electronic- financial markets to press the wrong JAPAN STOCK EXCHANGE. 610,000 SHARES WERE
Optimis m
Denial Fear Desperatio
nn
Optimis mm Relief Hope
button on their keyboard. Panic
2
2.Humans are
Figure4: Behavioral Biases
emotional. Traders tend to behavior in illogical way, because they are influenced by many other than market related factors.
ADVANTAGES OF HUMAN TRADING
1. Multistage decision-making In investment companies, departments cooperate to find optimal financial solutions for their costumer. Synergies emerge from shared responsibilities and risks. However, the investment process can be time-consuming and expensive.
2.Consideration of soft criteria Investments base on economic reasoning, i.e. the mindset goes beyond quantitative factors and considers qualitative soft criteria as well, which cannot be covered by any algorithm. DISADVANTAGES OF ALGORITHMIC TRADING
1. Prone to crashes GOLDMAN SACHS GLOBAL EQUITIES
HAD PILED UP A LOST OF
$9.3BN
•
Knights Capital (440m loss)
•
Flash Crash ( DJIA 9%
CLOSED AS THEY
AND
$7.3BN,
down)
RESPECTIVELY. •
BATS IPO plunge (9,4% in 5
minutes) •
Microcrashes iv
2. Evidences of poor performance Goldman Sachs Global Equities & Global
Figure3 :Number of Microcrashes in recent years
3
Apha Fund closed because of high losses.
3. A static mechanism based mainly on historical data and past empirical relations, although more and more algorithms do analyze soft information and rumors.
4. There is a possibility to learn the competitor’s algorithm, so quote stuffing is very popular misleading technique, what may lead to liquidity illusion and increased volatility.
5. Procyclic behavior HFT increases market volatility. Figure3 :Comparison of market volatility between 2007 and 2011
4
DURING CRISIS
THE FINANCIAL
(WHICH
WE
MEASURE FROM JANUARY
2007 - MARCH 2009),
6. Market system disadvantages
QUANT FUNDS DID BETTER THAN QUALITATIVE FUNDS
(3.29% VERSUS -4.77%).
•
Timing Delay Costs - Any delay cost incurred
between the Initial Decision (Open on Day 1) and the Broker Placement Price. •
Market Impact Costs - Price change between the time the
Order is placed with the Broker and the eventual trade price.
7. Higher broker commissions—fixed and variable components (e.g. Equity brokers charge USD 0.0001 to USD 0.003 commissions per share of stock traded through them in addition to the exchange fees)
8. Regulation SEC has launched trading rules that pause trading completely. For every share a price band is calculated, which is 5 per cent to either side of the average price over the
5
previous five
astonishing, because of economic of scale, algorithmic trading is cheaper in
minutes. The so
comparison with traditional methods.
called “circuit breaker” will then ban trades outside
3. Perfect memory of past market events.
that range.
ADVANTAGES
4. Discipline and behavioral bias resistance.
1. Algorithmic
5. Beneficial for market liquidity improvement, and thus lower bid-ask
trading hedge
spreads.
funds in
Figure8 :S&P500 Median Bid-Ask
comparison to qualitative funds generate higher
NEWS-INDUCED THE
PRICE ADJUSTMENT OF INTEREST RATE AND FOREIGN EXCHANGE FUTURES HAPPENS WITHIN
60 SECONDS AFTER THE
NEWS IS RELEASED.
mean returns 9,8% (8,9%) and lower standard deviation 15,71 (16,71%)v The qualitative funds perform significantly better than quant funds in up markets (25% and 15% respectively). However, the quant funds do significantly better in down markets (2% versus -16%).
TEST
THE
HFT
ALGORITHM ON THE WEBSITE: HTTP://WWW.
HFTRADINGBOOK.COM.
WHILE
REGISTERING AS A NEW USER, YOU WILL NEED THE PASSWORD:
HIGH-FREQUENCY
2. Lower administration costs- although the$26,4 billion
Figure7: Number of quotes and trades between 2006 and 2012
spent on IT in financial industry in 2005 may seem
6
remain the system operational 99.99 percent of the trading time). Appendix
6.Most of the highfrequency production-bound systems written
are in
C++,
which
allows
the
client
to
be
independent
of
system’s provider. 7.The system incorporates macroeconomics information
Resources
quickly. Highfrequency studies
1.HFT: Evolution and Future. Capegemini.
of the bond market responses to macroeconomic
2.I. Aldrige: High frequency trading, A Practical Guide To Alghoritmic Strategies And Trading System. 3. Hedgefund Review
announcements show that new information is fully incorporated in bond prices just 2 minutes following its announcement.
8. The failure rate for any welldesigned HFT system should not exceed 0.01 percent (i.e., the provider guarantees to
7
i
CitiPrime Finance Per Trac Survey 2011 I.Aldridge, High Frequency Trading, A practical guide to algorithmic strategies and trading systems, Page 33
ii
iii
iv
1.
2.
v
To qualify as a down-draft candidate, the stock had to tick down at least 10 times before ticking up -- all within 1.5 seconds and the price change had to exceed 0.8%.
To qualify as a up-draft candidate, the stock had to tick up at least 10 times before ticking down -- all within 1.5 seconds and the price change had to exceed 0.8%. Based on research of 6,354 hedge funds from January 1970 through June 2009