Nitesh Khandelwal & Rajib Ranjan Borah
QuantInsti
Options Workshop: Algorithmic & Automated Trading
Introduction to Algorithmic Trading
Bangkok, 7-Oct-2014
Trading in the markets
If you have a profitable trading strategy, then …
If you have a profitable trading strategy, then …
• do it as frequently (don’t miss any opportunity)
• scale it up (trade as many financial instruments)
• don’t let emotions affect (greed & fear: traders’ biggest enemies)
Trading in the markets
If you have a profitable trading strategy, then …
• do it as frequently (don’t miss any opportunity)
• scale it up (trade as many financial instruments)
• don’t let emotions affect (greed & fear: traders’ biggest enemies)
Computers:
• always at their seats
• respond to opportunities in
microseconds
Trading in the markets
If you have a profitable trading strategy, then …
• do it as frequently (don’t miss any opportunity)
• scale it up (trade as many financial instruments)
• don’t let emotions affect (greed & fear: traders’ biggest enemies)
Human eye can monitor 10-
15 stocks.
Computers can track
thousands simultaneously
Trading in the markets
If you have a profitable trading strategy, then …
• do it as frequently (don’t miss any opportunity)
• scale it up (trade as many financial instruments)
• don’t let emotions affect (greed & fear: traders’ biggest enemies)
Computers have no
emotions
Trading in the markets
If you have a profitable trading strategy, then … • do it as frequently (don’t miss any opportunity)
• scale it up (trade as many financial instruments)
• don’t let emotions affect (greed & fear: traders’ biggest enemies)
Trading is all about Probability & Computations.
And computers do calculations faster!
Trading in the markets
Inevitably, machines have taken over human beings
Trading today
Inevitably, machines have taken over human beings
Trading today
Trading shifted from pits …
…to computers
…and even more computers
Trading Landscape changes
This revolution has been fast
Effect of Algo - Trading
Growth in trading activity
Options
FX
Equity
Options
FX
Equity
Growth in trading activity
Effect of Algo - Trading
Options
FX
Equity
Growth in trading activity
Effect of Algo - Trading
Unfortunately… Computers don’t think The price dropped from $796 to $775 in about 3/4 of a second, then rebounded to $793 a second later. The drop involved 307 trades and 57,255 shares from 10 exchanges + dark pools.
Effect of Algo - Trading
Pros and Cons
Trading algorithmically is generally more profitable
Trading algorithmically is generally more profitable
• Less downtime • No emotions (Greed & Fear) • React faster • Higher scalability • Accurate and faster calculations
Pros and Cons
Trading algorithmically is generally more profitable
But…
Systems are getting
more complicated
Traditional trading system
Pros and Cons
Trading algorithmically is generally more profitable
But…
Systems are getting
more complicated
Thereby Increasing
likelihood of errors
Automated trading system
Pros and Cons
Trading algorithmically is generally more profitable …
… It can be riskier
Pros and Cons
Benefits of High-Frequency Trading
• Improves scalability – Non-automated trader can monitor not a lot of
information
• Increased system uptime – Trader need not be on his toes all the time
– No system fatigue
• Faster response to events – Respond in microseconds
• Trade more opportunities – Non-automated trading sometimes cannot
respond in time while opportunity exists
• Better risk and portfolio management
Benefits to traders
Benefits to market
Spillover benefits
• Improved market liquidity • Reduced costs of trading
– Aite Grp estimates avg retail trader saves ~$500 a year on tightened bid-ask spreads
• Reduced volatility of markets (higher stability)
• Increased market efficiency • Reduced dependency on brokers (DMA) • Improved infrastructure at the exchange
– Level playing field (transparency & efficiency) – Faster exchange response time => lower exposure time for market makers => better risk management & quotes
Benefits to traders
Benefits to market
Spillover benefits
Benefits of High-Frequency Trading
Benefits
to traders
Benefits to market
Spillover benefits
• Several spillover benefits to technological sector in general: • Automatic complex event processing in
microseconds • Artificial intelligence systems greatly evolved
(algorithmic trading systems use artificial intelligence to think like humans)
• Ability to process huge amount of information flow (market data)
• Real time mission critical stable systems • Statistical systems to process & analyse
monstrous amounts of data (Big Data) • Accelerated hardware • Hardware programming (FPGA, etc)
Benefits of High-Frequency Trading
Copyright © 2014 by QuantInsti Quantitative Learning Private Limited.
Although great care has been taken to ensure accuracy of the information
in this presentation – however the author (and QuantInsti) accepts no
liability or warranty for the precision, correctness or completeness of any
statement, estimate or opinion. QuantInsti also accepts no liability for the
consequences of any actions taken on the basis of the information
provided.
The slides of this presentation cannot be taken separately from the whole
set of slides.
Prior approval from QuantInsti is necessary before usage of this
presentation for educational and (or) commercial purposes.
This document provides an outline of a presentation and is incomplete
without the accompanying oral commentary and discussion.
Disclaimer
System Architecture - impact on trading performance
Options Workshop: Algorithmic & Automated Trading
Nitesh Khandelwal & Rajib Ranjan Borah
QuantInsti Bangkok 7-Oct-2014
2
Table of Contents • System architecture of a traditional trading
system
• System architecture of an algorithmic trading system
• System architecture of external components
3
Table of Contents • System architecture of a traditional trading
system
• System architecture of an algorithmic trading system
• System architecture of external components
Traditional systems → Automated systems → Internal components → External components → QA
4
System Architecture of a Traditional Trading System
• Traditionally a trading system would consist of
– A system to read data from the market
– A storehouse of historical data
– A tool to analyse historical data
– A system where the trader can input his trading decisions
– A system to route orders to the exchange
Traditional systems → Automated systems → Internal components → External components → QA
5
System Architecture of a Traditional Trading System
• A system to read data from the market
Market Data Exchange
Traditional systems → Automated systems → Internal components → External components → QA
6
System Architecture of a Traditional Trading System
• A storehouse of historical data – Which could also be directly purchased from third party data vendor
directly
Market Data Exchange
Data
Warehouse
/
Storehouse
of
historical
data
Data
Vendor
Traditional systems → Automated systems → Internal components → External components → QA
7
System Architecture of a Traditional Trading System
• A subset of the database is queried and stored locally for operational uses
Market Data
Operational Data Store
Exchange
Data
Warehouse
/
Storehouse
of
historical
data
Data
Vendor
Traditional systems → Automated systems → Internal components → External components → QA
8
System Architecture of a Traditional Trading System
• The trader’s tool would then analyze current data against patterns discovered in the operational data store
Trader’s tool
Main Centre of operations
– analyzing market data
wrt to historical data in
operational data store
and generating orders
Market Data
Operational Data Store
Exchange
Data
Warehouse
/
Storehouse
of
historical
data
Data
Vendor
Traditional systems → Automated systems → Internal components → External components → QA
9
System Architecture of a Traditional Trading System
• The trader’s tool would then generate orders which will be forwarded to the order management tool
Order Manager
Market Data
Operational Data Store
Exchange
Data
Warehouse
/
Storehouse
of
historical
data
Data
Vendor
Trader’s tool
Main Centre of operations –
analyzing market data wrt
to historical data in
operational data store and
generating orders
Traditional systems → Automated systems → Internal components → External components → QA
10
System Architecture of a Traditional Trading System
• The order manager would then route the orders to the exchange
Order Manager
Market Data
Operational Data Store
Exchange
Data
Warehouse
/
Storehouse
of
historical
data
Data
Vendor
Trader’s tool
Main Centre of operations –
analyzing market data wrt
to historical data in
operational data store and
generating orders
Traditional systems → Automated systems → Internal components → External components → QA
11
System Architecture of a Traditional Trading System
• The data warehouse would also probably store records of orders sent out
Order Manager
Market Data
Operational Data Store
Exchange
Data
Warehouse
/
Storehouse
of
historical
data
Data
Vendor
Trader’s tool
Main Centre of operations –
analyzing market data wrt
to historical data in
operational data store and
generating orders
Traditional systems → Automated systems → Internal components → External components → QA
12
System Architecture of a Traditional Trading System
• The whole system could be broken down to three components
Order Manager
Market Data
Operational Data Store
Exchange
Data
Warehouse
/
Storehouse
of
historical
data
Data
Vendor
Trader’s tool
Main Centre of operations –
analyzing market data wrt
to historical data in
operational data store and
generating orders
Traditional systems → Automated systems → Internal components → External components → QA
13
System Architecture of a Traditional Trading System
• The exchange (and other data sources) – i.e. the external world
Order Manager
Market Data
Operational Data Store
Exchange
Data
Warehouse
/
Storehouse
of
historical
data
Data
Vendor
Trader’s tool
Main Centre of operations –
analyzing market data wrt
to historical data in
operational data store and
generating orders
Exchange
Traditional systems → Automated systems → Internal components → External components → QA
14
System Architecture of a Traditional Trading System
• The server – which is mostly a data store
Order Manager
Market Data
Operational Data Store
Exchange
Data
Warehouse
/
Storehouse
of
historical
data
Data
Vendor
Trader’s tool
Main Centre of operations –
analyzing market data wrt
to historical data in
operational data store and
generating orders
Server Exchange
Traditional systems → Automated systems → Internal components → External components → QA
15
System Architecture of a Traditional Trading System
• The applications in the trader’s pc which do all the processing
Order Manager
Market Data
Operational Data Store
Exchange
Data
Warehouse
/
Storehouse
of
historical
data
Data
Vendor
Trader’s tool
Main Centre of operations –
analyzing market data wrt
to historical data in
operational data store and
generating orders
Application Server Exchange
Traditional systems → Automated systems → Internal components → External components → QA
16
System Architecture of a Traditional Trading System
• If data operations are simple… operational data store can be in application layer (trader’s pc)
Order Manager
Market Data
Operatio
nal Data
Store
Exchange
Data
Warehouse
/
Storehouse
of
historical
data
Data
Vendor
Trader’s tool
Main Centre of operations –
analyzing market data wrt
to historical data in
operational data store and
generating orders
Application Server Exchange
Traditional systems → Automated systems → Internal components → External components → QA
17
Table of Contents
• System architecture of a traditional trading system
• System architecture of an algorithmic trading system
• System architecture of external components
Traditional systems → Automated systems → Internal components → External components → QA
18
System Architecture of an Automated Trading System
• With the advent of DMA & automated trading, the following changes in architecture took place:
– Latency between Event Occurrence & Order Generation had to be reduced to an order of milliseconds and lower.
– Order Management had to be made more robust to handle generation of thousands of orders in a second
– Risk Management had to be done in real time and without human intervention
Traditional systems → Automated systems → Internal components → External components → QA
19
System Architecture of an Automated Trading System
• To generate orders quickly, market events are now handled in the server instead of the application - in the CEP module
Application
Order Manager
Market Data
Complex Event Processing
engine
Exchange
1
Storage
Application Server Exchange
Traditional systems → Automated systems → Internal components → External components → QA
20
System Architecture of an Automated Trading System
• Complex mathematical operations are handled in a dedicated calculation block in the server block (e.g. Options Greeks calculations)
Application
Order Manager
Market Data
Complex Event Processing
engine
Exchange
1
Storage
Application Server Exchange
Maths
Calc
Traditional systems → Automated systems → Internal components → External components → QA
21
System Architecture of an Automated Trading System
• The role of the application layer has reduced drastically – (i) an input screen for strategy settings
Application
Order Manager
Market Data
Complex Event Processing
engine
Exchange
1
Storage
Application Server Exchange
Strategy
Settings
UI
Maths
Calc
Traditional systems → Automated systems → Internal components → External components → QA
22
System Architecture of an Automated Trading System
• The role of the application layer has reduced drastically – (ii) a monitor of position & orders
Application
Order Manager
Market Data
Complex Event Processing
engine
Exchange
1
Storage
Application Server Exchange
Strategy
Settings
UI
State
Mgmt
(PnL +
Position)
Order /
Execution
Monitor
Maths
Calc
Traditional systems → Automated systems → Internal components → External components → QA
23
System Architecture of an Automated Trading System
• The role of the application layer has reduced drastically – (iii) preliminary fat finger RMS checker
Application
Order Manager
Market Data
Complex Event Processing
engine
Exchange
1
Storage
Application Server Exchange
Strategy
Settings
UI
State
Mgmt (PnL
+ Position)
Order /
Execution
Monitor
Within
applicatio
n RMS
Maths
Calc
Traditional systems → Automated systems → Internal components → External components → QA
24
System Architecture of an Automated Trading System
• RMS is now automated and checked by the OMS before an order is generated
Application
Order Manager
Market Data
Complex Event Processing
engine
Exchange
1
Storage
Application Server Exchange
Strategy
Settings
UI
State
Mgmt (PnL
+ Position)
Order /
Execution
Monitor
Within
application
RMS
Maths
Calc
RMS
Traditional systems → Automated systems → Internal components → External components → QA
25
System Architecture of an Automated Trading System
• Because RMS is automated, a second level of monitoring is necessary – an overall global position monitor
Application
Order Manager
Market Data
Complex Event Processing
engine
Exchange
1
Storage
Application Server Exchange
Strategy
Settings
UI
State
Mgmt (PnL
+ Position)
Order /
Execution
Monitor
Within
application
RMS
Maths
Calc
RMS
Admin
Monitor
Traditional systems → Automated systems → Internal components → External components → QA
26
System Architecture of an Automated Trading System
• With Multiple destinations being connected to the automated systems, standardized protocols (FIX) became the norm
Application
Order Manager
Market Data
Complex Event Processing
engine
Exchange
1
Storage
Application Server Exchange
Strategy
Settings
UI
State
Mgmt (PnL
+ Position)
Order /
Execution
Monitor
Within
application
RMS
Maths
Calc
RMS
Exchange
2
F
I
X
F
I
X
Admin
Monitor
Traditional systems → Automated systems → Internal components → External components → QA
27
System Architecture of an Automated Trading System
• This also necessitated adding data normalizer block in the market data adaptors to convert data from multiple exchanges into a standard format
Application
Order Manager
Market Data
Complex Event Processing
engine
Exchange
1
Storage
Application Server Exchange
Strategy
Settings
UI
State
Mgmt (PnL
+ Position)
Order /
Execution
Monitor
Within
application
RMS
Maths
Calc
RMS
Exchange
2
F
I
X
F
I
X
Data
Normalizer
Admin
Monitor
Traditional systems → Automated systems → Internal components → External components → QA
28
System Architecture of an Automated Trading System
• Moreover, an Order Router had to be added to the OMS to route orders from the same OMS to multiple exchanges
Application
Order Manager
Market Data
Complex Event Processing
engine
Exchange
1
Storage
Application Server Exchange
Strategy
Settings
UI
State
Mgmt (PnL
+ Position)
Order /
Execution
Monitor
Within
application
RMS
Maths
Calc
RMS
Exchange
2
F
I
X
F
I
X
Data
Normalizer
Order
Router Admin
Monitor
Traditional systems → Automated systems → Internal components → External components → QA
29
System Architecture of an Automated Trading System
• Regulatory requirements have complicated storage requirements – requiring storage of trade information in addition to market data
Application
Order Manager
Market Data
Complex Event Processing
engine
Exchange
1
Storage
Application Server Exchange
Strategy
Settings
UI
State
Mgmt (PnL
+ Position)
Order /
Execution
Monitor
Within
application
RMS
Maths
Calc
RMS
Exchange
2
F
I
X
F
I
X
Data
Normalizer
Order
Router
Back
office
record
MktData
Store
Admin
Monitor
Traditional systems → Automated systems → Internal components → External components → QA
30
System Architecture of an Automated Trading System
• And obviously it should be able to read information from third party vendors as well
Application
Order Manager
Market Data
Complex Event Processing
engine
Exchange
1
Storage
Application Server Exchange
Strategy
Settings
UI
State
Mgmt (PnL
+ Position)
Order /
Execution
Monitor
Within
application
RMS
Maths
Calc
RMS
Exchange
2
F
I
X
F
I
X
Data
Normalizer
Order
Router
Back
office
record
MktData
Store
Data
Vendor
Admin
Monitor
Traditional systems → Automated systems → Internal components → External components → QA
31
System Architecture of an Automated Trading System
• Moreover, the CEP engine has its own storage requirements of event history to identify future opportunities
Application
Order Manager
Market Data
Complex Event Processing
engine
Exchange
1
Storage
Application Server Exchange
Strategy
Settings
UI
State
Mgmt (PnL
+ Position)
Order /
Execution
Monitor
Within
application
RMS
Maths
Calc
RMS
Exchange
2
F
I
X
F
I
X
Data
Normalizer
Order
Router
Back
office
record
MktData
Store
Event
History
Data
Vendor
Admin
Monitor
Traditional systems → Automated systems → Internal components → External components → QA
32
System Architecture of an Automated Trading System
• Complex functionality third party applications have to be tightly integrated with all the blocks
Application
Order Manager
Market Data
Complex Event Processing
engine
Exchange
1
Storage
Application Server Exchange
Strategy
Settings
UI
State
Mgmt (PnL
+ Position)
Order /
Execution
Monitor
Within
application
RMS
Maths
Calc
RMS
Exchange
2
F
I
X
F
I
X
Data
Normalizer
Order
Router
Back
office
record
MktData
Store
Event
History
Adaptor for
third party
apps – R,
Matlab, etc
Data
Vendor
Admin
Monitor
Traditional systems → Automated systems → Internal components → External components → QA
33
System Architecture of an Automated Trading System
• Independent data management tools to verify the sanity of the information have to configured
Application
Order Manager
Market Data
Complex Event Processing
engine
Exchange
1
Storage
Application Server Exchange
Strategy
Settings
UI
State
Mgmt (PnL
+ Position)
Order /
Execution
Monitor
Within
application
RMS
Maths
Calc
RMS
Exchange
2
F
I
X
F
I
X
Data
Normalizer
Order
Router
Back
office
record
MktData
Store
Event
History
Adaptor for
third party
apps – R,
Matlab, etc
Data
Retrieval
Data
Vendor
Admin
Monitor
Traditional systems → Automated systems → Internal components → External components → QA
34
System Architecture of an Automated Trading System
• To be able to back-test strategies, two components are required: (i) replay of stored data
Application
Order Manager
Market Data
Complex Event Processing
engine
Exchange
1
Storage
Application Server Exchange
Strategy
Settings
UI
State
Mgmt (PnL
+ Position)
Order /
Execution
Monitor
Within
application
RMS
Maths
Calc
RMS
Exchange
2
F
I
X
F
I
X
Data
Normalizer
Order
Router
Back
office
record
MktData
Store
Event
History
Adaptor for
third party
apps – R,
Matlab, etc
Data
Retrieval
Data
Vendor
Replay
of
stored
data
Admin
Monitor
Traditional systems → Automated systems → Internal components → External components → QA
35
System Architecture of an Automated Trading System
• To be able to back-test strategies, two components are required: (ii) a simulator destination instead of an actual exchange
Application
Order Manager
Market Data
Complex Event Processing
engine
Exchange
1
Storage
Application Server Exchange
Strategy
Settings
UI
State
Mgmt (PnL
+ Position)
Order /
Execution
Monitor
Within
application
RMS
Maths
Calc
RMS
Admin
Monitor
Exchange
2
F
I
X
F
I
X
Data
Normalizer
Order
Router
Back
office
record
MktData
Store
Event
History
Adaptor for
third party
apps – R,
Matlab, etc
Data
Retrieval
Data
Vendor
Replay
of stored
data
Simulator
exchange
Traditional systems → Automated systems → Internal components → External components → QA
36
Table of Contents • System architecture of a traditional trading
system
• System architecture of an algorithmic trading system
• System architecture of external components
Traditional systems → Automated systems → Internal components → External components → QA
37
System Architecture – external components
• Standard vs Native – FIX vs Proprietary
Standard APIs Native APIs
Traditional systems → Automated systems → Internal components → External components → QA
38
Relevant factors for analyzing different markets → Current landscape in different geographies → The future → QA
Technology Protocols globally One of the most popular protocol is FIX
Financial Information eXchange protocol
Data transferred as sequence of tag=value pairs
e.g.: 8=FIX.4.2 | 9=77 | 35=0 | 49=MCXSXTRADE | 56=xxxxx | 34=24
| 43=N | 52=20140607-05:14:05 | 112=DNLDCOMPLETE | 10=127
1 Account
2 AdvId
3 AdvRefID
4 AdvSide
5 AdvTransType
6 AvgPx
7 BeginSeqNo
8 BeginString
9 BodyLength
10 CheckSum
11 ClOrdID
12 Commission
13 CommType
14 CumQty
15 Currency
16 EndSeqNo
17 ExecID
18 ExecInst
19 ExecRefID
20 ExecTransType
21 HandlInst
22 IDSource
23 IOIid
24 IOIOthSvc (no longer used)
25 IOIQltyInd
26 IOIRefID
27 IOIShares
28 IOITransType
29 LastCapacity
30 LastMkt
31 LastPx
32 LastShares
33 LinesOfText
34 MsgSeqNum
35 MsgType
36 NewSeqNo
37 OrderID
38 OrderQty
39 OrdStatus
40 OrdType
41 OrigClOrdID
42 OrigTime
43 PossDupFlag
44 Price
45 RefSeqNum
46 RelatdSym
47 Rule80A(aka OrderCapacity)
48 SecurityID
49 SenderCompID
50 SenderSubID
51 SendingDate (no longer used)
52 SendingTime
53 Shares
54 Side
55 Symbol
56 TargetCompID
57 TargetSubID
39
System Architecture – external components
Message Rates/ order throttles
1. Enforced using message rate per
second
2. Invitation informing available
message slots
3. Intermediate layer – TAP Server
4. Bandwidth considerations
Traditional systems → Automated systems → Internal components → External components → QA
Technology and Hardware innovations
Options Workshop: Algorithmic & Automated Trading
QuantInsti 7-Oct-2014
41
Agenda • Understanding Performance factors
• Improving Performance – software
• Improving Performance – hardware
42
Agenda • Understanding Performance factors
• Improving Performance – software
• Improving Performance – hardware
43
How critical is performance ?
Hundreds of millions of dollars spent by various people to reduce the
round trip time between Chicago and New York from 13 milliseconds to
around 9 milliseconds
44
How critical is performance ?
Hundreds of millions of dollars spent by various people to write code in
hardware itself instead of implementing it in software
45
What is Performance? • Internal to the system
– Lots of data – fast processing
– Fast decision making
– Fast risk management
• External to the system
– Fast transfer of data, both outgoing and incoming
– Fast processing of requests at the destination
46
Microbursts
47
Latencies: Exchange to Network
• Typical latencies from exchange to network
48
Latencies: Within network • Typical latencies within internal network
49
Agenda • Understanding Performance factors
• Improving Performance – software
• Improving Performance – hardware
50
Race to 0 latency – Software
• Re-tune TCP Stack. TCP hacks (TCP_NODELAY, sendfile(2))
• Moving the software as close to hardware as possible.
• Move from reusable object oriented code towards extremely efficient code using lower level of programming
• Avoiding disk writes, in memory storage
• Program Runtime – isolcpus, stack bypass
• Lock Free Codes
• Operating System – RT kernels
51
Agenda • Understanding Performance factors
• Improving Performance – software
• Improving Performance – hardware
52
Race to 0 latency – Hardware
• Servers – Fastest Cores, Cache, • Fastest Network Infra (Switches, Routers ) • Solid State Drives • Microwave Transmission • FPGA – ASIC implementation • Custom Chips • TCP Offload / Kernel Bypass by hardware DMA • Hardware message filters • Line arbitration
53
Latency race leading to microwaves
• Chicago to New York – Ordinary Cable round trip time = 14.5 milliseconds – Trading leased lines = 13.1 milliseconds – Oct 2012: Spread Network lays straighter cables resulting
in 12.98 milliseconds • Huge benefits to Spread as lot of players shifted there
– 2013-14: McKay Brothers and Tradeworx offer microwave connection between these cities
(9 & 8.5 milliseconds respectively) • Theoretical limit of physics = 7.5 to 8 milliseconds
54
Any Questions?
55
Copyright © 2014 by QuantInsti Quantitative Learning Private Limited.
Although great care has been taken to ensure accuracy of the information
in this presentation – however the author (and QuantInsti) accepts no
liability or warranty for the precision, correctness or completeness of any
statement, estimate or opinion. QuantInsti also accepts no liability for the
consequences of any actions taken on the basis of the information
provided.
The slides of this presentation cannot be taken separately from the whole
set of slides.
Prior approval from QuantInsti is necessary before usage of this
presentation for educational and (or) commercial purposes.
This document provides an outline of a presentation and is incomplete
without the accompanying oral commentary and discussion.
Disclaimer
Risk Management – for HFT systems
Options Workshop: Algorithmic & Automated Trading
Nitesh Khandelwal & Rajib Ranjan Borah
QuantInsti
Bangkok 7-Oct-2014
Table of Contents
• Case studies of major risk incidents globally
• Algorithmic Trading Related Risks
Major algorithmic trading incidents globally - I
• Credit Suisse, Nov 2007 – Incident:
• Hundreds of thousands of cancel orders sent to the exchange
• Orders clogged NYSE and affected trading of over 900 stocks
– Reasons: • Trader implemented code which could change parameters
on clicking on spin button
(without any need for confirmation)
• With each click, orders were cancelled and resent
– Fine/ Losses: • $150,000 fine
Major algorithmic trading incidents globally - II
• Infinium Capital, Feb 2010 – Incident:
• 4612 trades on crude oil futures in 24 seconds
– Reasons: • Strategy was designed to trade energy ETFs on the basis of
crude prices
• Trader configured crude oil futures on the basis of energy ETFs
• Moreover, RMS was designed on the basis of ETF prices, not crude prices
– Fine/ Losses: • $850,000 fine by CME
• Deutsche Bank, June 2010 – Incident:
• Sent orders for 1.24 million Nikkei 225 Futures & 4.82 million Nikkei 225 mini-futures in first few minutes
• More than 10 times normal volume
• Market dropped 1% on orders
– Reasons: • Pair trade strategy used value of mini-Nikkei to quote
Nikkei. At start of day, there was no liquidity in mini-Nikkei
• Error recognized immediately, 99.7% orders cancelled
– Fine/ Losses: • Forced to close Algorithmic trading desk in Tokyo
Major algorithmic trading incidents globally - III
Major algorithmic trading incidents globally - IV
• Knight Capital, Aug 2012 – Incident:
• Traded 154 stocks at bizarre prices (4 million trades for 397 million shares in 45 minutes): alternately bought at higher prices and sold at lower prices
– Reasons: • Accidentally installed test software which incorporated an old
piece of code designed 9 years ago
• In one out of 8 production servers, new code was not installed by a technician
• No process for second technician to review
– Fine/ Losses: • Trading loss of $460 million in 45 minutes. Fine of $12 million
• Knight Capital had to be rescued by Getco
Major algorithmic trading incidents globally - V
• Goldman Sachs, Aug 2013
– Incident:
• Traded stock options at very erroneous prices at the exchange
– Reasons:
• Indication of interests were sent as actual orders to the exchange
– Fine/ Losses:
• Trading loss of $100 million
Major algorithmic trading incidents globally - VI
• Everbright Securities, Aug 2013 – Incident:
• Rogue algorithm kept buying – index moved up 6% intraday
• Did not inform regulators, shorted the artificial bubble – banned from prop trading forever for insider trading
– Reasons: • Algorithm sent buy orders for ETFs worth 23.4 billion
Yuan (7.27 billion got traded) – no risk check in OMS
– Fine/ Losses: • Banned from prop trading forever for insider trading
Major algorithmic trading incidents globally - VII
• HanMag Securities, Dec 2013 – Incident:
• HanMag exercised wrong call and put options
• 36,100 trades in a few minutes
– Reasons: • Error in automated profit taking trade program
(interchanged puts with calls)
– Fine/ Losses: • Some firms returned money back to HanMag (Optiver
returned $600k trading profits)
• Eventual loss of 57 billion Korean Won
• United Airlines flash crash
– Incident:
• On Sep. 7, 2008 United Airlines had a downward price spike
– Reasons:
• Google’s news bots picked up an old 2002 story about United Airlines possibly filing for bankruptcy
• News Analytics based automated traders reacted to it
Major algorithmic trading incidents - VIII
• Dow Jones Flash Crash
– Incident: • On Apr 23, 2013 Markets dropped 0.8% momentarily
– Reasons: • Twitter account of news publisher hacked – false news of White
house explosion
• News Analytics based automated traders reacted to it
Major algorithmic trading incidents - IX
Table of Contents
• Case studies of major risk incidents globally
• Algorithmic Trading Related Risks
Risks related to Algorithmic Trading
• Orders flow without human control – Higher reliance on technology implies increased sys-ops risk – In case of a wrong input, algorithm will execute at a wrong
level – If price feed goes down, algorithm will send orders based on
stale data • price feed could go down because of physical connectivity,
exchange disconnection, software crash, etc
• Before a human can realize (and then respond), tremendous damage would have been done already
• Trades happen at such a fast pace, that positions could reach a dangerous level in no time – Real-time monitor of positions, exposures, regulation checks
• Risks specific to automated trading can be classified into the following categories: – Access – Consistency – Quality – Algorithm – Technology – Scalability
• These risks have to be handled pre-order – Within the application – Before generating an order in the Order Management
System
• Moreover, it is pertinent that the trader understands the internal working of the black-box
Risks related to Algorithmic Trading
RISK Handled in
Methodology App OM
Access Connectivity to an exchange goes down
Y Y Heart-beats
Exchange disconnects you Y Y Heart-beats
Network issue Y Hardware, Operating System
Consistency Market Data is stale Y Y Time-stamp
Analytics are running in real-time (huge processing time)
Y Time-stamp
OM adaptor is responding in real time
Y Time-stamp
Quality Market - data is garbled Y Common RMS rule
Loss of liquidity during high-volatility
Y Common RMS rule
Risks related to Algorithmic Trading
RISK
Risk Handled in Methodology
App OM
Algorithmic Margin breached Y Common RMS rule
Exposure limit set by exchange
Y Common RMS rule
Risk limits exceeded Y
Check for acknowledgements before sending order
Incorrect strategy setting leading to continual mistrades
Y Y PnL fluctuation check
-do- Y Order throttle rate
-do- Y Fat finger settings check
-do- Y Max Value Traded
Incorrect order generation Y Y Price range check
Order throttle Y Exchange reject limit
Risks related to Algorithmic Trading
RISK
Risk Handled in
Methodology
App OM
Technology Hard disk gets full Independent check
Virus /Trojan Firewall, Anti-virus
System Crash Operating System
Application crash Y
Heart-beat to check application
Protocol Mismatch Third-party software compatibility check
Risks related to Algorithmic Trading
RISK
Handled in
Methodology App OM
Scalability Number of applications & portfolios that can be handled Y Y
Number of exchanges that can be connected Y
Number of symbols that can be handled Y Y
Order of complexity of computations Y
Risks related to Algorithmic Trading
Risks related to Algorithmic Trading
Knowledge is the Key
E-PAT
Statistics and Econometrics
Financial Computing & Technology
Algorithmic & Quantitative Trading
QI’s E-PAT course
E-PAT
Statistics and Econometrics
Financial Computing & Technology
Algorithmic & Quantitative Trading
E-PAT course structure - module I
Basic Statistics
Advanced Statistics
Time Series Analysis
Probability and Distribution Statistical Inference Linear Regression
Correlation vs. Co-integration ARIMA, ARCH-GARCH Models Multiple Regression
Stochastic Math Causality Forecasting
E-PAT
Statistics and Econometrics
Financial Computing & Technology
Algorithmic & Quantitative Trading
E-PAT course structure - module II
Programming
Technology for Algorithmic Trading
Statistical Tools
Intro to Programming Language(s) Programming on Algorithmic
Trading Platforms
System Architecture Understanding an Algorithmic
Trading Platform Handling HFT Data
Excel & VBA Financial Modeling using R Using R & Excel for Back-testing
E-PAT
Statistics and Econometrics
Financial Computing & Technology
Algorithmic & Quantitative Trading
E-PAT course structure - module III
Trading Strategies
Derivatives & Market Microstructure
Managing Algo Operations
Statistical Arbitrage Market Making Strategies Execution Strategies Forecasting & AI Based Strategies Pair Trading Strategies Trend following Strategies
Option Pricing Model Dispersion Trading Risk Management using Higher
Order Greeks Option Portfolio Management Order Book Dynamics Market Microstructure
Hardware & Network Regulatory Framework Exchange Infrastructure & Financial
Planning (Costing) Risk Management in Automated
systems Performance Evaluation & Portfolio
Management
E-PAT
Statistics and Econometrics
Financial Computing & Technology
Algorithmic & Quantitative Trading
Project work
E-PAT course structure - project
Contacts For 4-month Executive Program in Algorithmic Trading:
E-PAT: 4 month weekend online program (3hrs every Sat + Sun)
• Statistics & Econometrics
• Quantitative Trading
• Financial Computing
For algorithmic trading advisory: [email protected]
To reach us directly: [email protected] , [email protected]
Copyright © 2014 by QuantInsti Quantitative Learning Private Limited.
Although great care has been taken to ensure accuracy of the information in this presentation – however the author (and QuantInsti) accepts no liability or warranty for the precision, correctness or completeness of any statement, estimate or opinion. QuantInsti also accepts no liability for the consequences of any actions taken on the basis of the information provided.
The slides of this presentation cannot be taken separately from the whole set of slides.
Prior approval from QuantInsti is necessary before usage of this presentation for educational and (or) commercial purposes.
This document provides an outline of a presentation and is incomplete without the accompanying oral commentary and discussion.
Disclaimer
Top Related